diff --git a/server.py b/server.py index 048e92c..dade96e 100644 --- a/server.py +++ b/server.py @@ -8,6 +8,7 @@ import logging import logging.handlers # For RotatingFileHandler import shutil +import struct import time import uuid import yaml # For loading presets @@ -251,6 +252,35 @@ async def get_main_script(): # These functions support smart audio chunk concatenation with crossfading +def _create_wav_header( + sample_rate: int, channels: int = 1, bits_per_sample: int = 16 +) -> bytes: + byte_rate = sample_rate * channels * (bits_per_sample // 8) + block_align = channels * (bits_per_sample // 8) + max_32bit_int = 0xFFFFFFFF + riff_chunk_size = max_32bit_int + data_size = riff_chunk_size - 36 + header = b"RIFF" + header += struct.pack(" bytes: + audio = np.clip(audio, -1.0, 1.0) + audio_int16 = (audio * 32767).astype(np.int16) + return audio_int16.tobytes() + + def _generate_equal_power_curves(n_samples: int): """ Generate equal-power crossfade curves using cos²/sin² functions. @@ -500,7 +530,9 @@ async def save_settings_endpoint(request: Request): ) message = "Settings saved successfully." if restart_needed: - message += " A server restart may be required for some changes to take full effect." + message += ( + " A server restart may be required for some changes to take effect." + ) return UpdateStatusResponse(message=message, restart_needed=restart_needed) else: logger.error( @@ -917,7 +949,14 @@ async def custom_tts_endpoint( status_code=400, detail="Text processing resulted in no usable chunks." ) + base_seed = request.seed if request.seed is not None else get_gen_default_seed() + for i, chunk in enumerate(text_chunks): + + current_chunk_seed = base_seed + if base_seed is not None and base_seed >= 0: + current_chunk_seed = base_seed + i + logger.info(f"Synthesizing chunk {i+1}/{len(text_chunks)}...") try: chunk_audio_tensor, chunk_sr_from_engine = engine.synthesize( @@ -942,9 +981,7 @@ async def custom_tts_endpoint( if request.cfg_weight is not None else get_gen_default_cfg_weight() ), - seed=( - request.seed if request.seed is not None else get_gen_default_seed() - ), + seed=current_chunk_seed, language=( request.language if request.language is not None @@ -981,9 +1018,6 @@ async def custom_tts_endpoint( ) perf_monitor.record(f"Speed factor applied to chunk {i+1}") - # ### MODIFICATION ### - # All other processing is REMOVED from the loop. - # We will process the final concatenated audio clip. processed_audio_np = current_processed_audio_tensor.cpu().numpy().squeeze() all_audio_segments_np.append(processed_audio_np) @@ -1007,21 +1041,18 @@ async def custom_tts_endpoint( ) try: # ### SMART AUDIO STITCHING ### - # Local constants - adjust these values to tune stitching behavior - SENTENCE_PAUSE_MS = 200 # Desired audible silence between sentences - CROSSFADE_MS = 20 # Crossfade duration for smart mode (10-50ms recommended) - SAFETY_FADE_MS = 3 # Minimal edge fade for fallback mode (2-5ms) - ENABLE_DC_REMOVAL = False # Set True if you hear low-frequency thumps - DC_HIGHPASS_HZ = 15 # High-pass cutoff for DC removal - PEAK_NORMALIZE_THRESHOLD = 0.99 # Normalize if peak exceeds this - PEAK_NORMALIZE_TARGET = 0.95 # Target peak after normalization - - # Read smart stitching toggle from config (defaults to True) + SENTENCE_PAUSE_MS = 200 + CROSSFADE_MS = 20 + SAFETY_FADE_MS = 3 + ENABLE_DC_REMOVAL = False + DC_HIGHPASS_HZ = 15 + PEAK_NORMALIZE_THRESHOLD = 0.99 + PEAK_NORMALIZE_TARGET = 0.95 + enable_smart_stitching = config_manager.get_bool( "audio_processing.enable_crossfade", True ) - # --- Sample rate validation --- if not engine_output_sample_rate or engine_output_sample_rate <= 0: logger.error( f"Invalid sample rate: {engine_output_sample_rate}, " @@ -1034,22 +1065,16 @@ async def custom_tts_endpoint( ) elif len(all_audio_segments_np) == 1: - # Single chunk - no stitching needed final_audio_np = all_audio_segments_np[0] logger.info("Single audio chunk - no stitching required") elif enable_smart_stitching: - # --- Smart mode: true crossfading with silence insertion --- fade_samples = int(CROSSFADE_MS / 1000 * engine_output_sample_rate) - - # Calculate silence buffer with compensation for crossfade overlap - # Each crossfade removes fade_samples from silence (one at each end) desired_silence_samples = int( SENTENCE_PAUSE_MS / 1000 * engine_output_sample_rate ) silence_buffer_samples = desired_silence_samples + (fade_samples * 2) - # Preprocess chunks: convert to float32 and optionally remove DC offset chunks = [] for chunk in all_audio_segments_np: processed = chunk.astype(np.float32, copy=True) @@ -1059,18 +1084,11 @@ async def custom_tts_endpoint( ) chunks.append(processed) - # Start with first chunk result = chunks[0] - # Stitch remaining chunks with crossfaded silence gaps for i in range(1, len(chunks)): - # Create silence buffer (oversized to compensate for crossfade overlap) silence = np.zeros(silence_buffer_samples, dtype=np.float32) - - # Crossfade: current result → silence (speech fades into silence) result = _crossfade_with_overlap(result, silence, fade_samples) - - # Crossfade: result → next chunk (silence fades into speech) result = _crossfade_with_overlap(result, chunks[i], fade_samples) final_audio_np = result @@ -1080,7 +1098,6 @@ async def custom_tts_endpoint( ) else: - # --- Fallback mode: minimal safety edge fades, no silence --- fade_samples = int(SAFETY_FADE_MS / 1000 * engine_output_sample_rate) num_chunks = len(all_audio_segments_np) @@ -1092,8 +1109,8 @@ async def custom_tts_endpoint( processed = _apply_edge_fades( chunk, fade_samples, - fade_in=(not is_first), # No fade-in on first chunk - fade_out=(not is_last), # No fade-out on last chunk + fade_in=(not is_first), + fade_out=(not is_last), ) processed_chunks.append(processed) @@ -1103,10 +1120,8 @@ async def custom_tts_endpoint( f"{SAFETY_FADE_MS}ms linear fades" ) - # --- Ensure float32 dtype for all code paths --- final_audio_np = final_audio_np.astype(np.float32, copy=False) - # --- Normalize to prevent clipping --- peak_amplitude = np.abs(final_audio_np).max() if peak_amplitude > PEAK_NORMALIZE_THRESHOLD: final_audio_np = final_audio_np * (PEAK_NORMALIZE_TARGET / peak_amplitude) @@ -1116,7 +1131,6 @@ async def custom_tts_endpoint( perf_monitor.record("Audio chunks stitched") - # --- Global Audio Post-Processing (applied to complete stitched audio) --- if config_manager.get_bool("audio_processing.enable_silence_trimming", False): final_audio_np = utils.trim_lead_trail_silence( final_audio_np, engine_output_sample_rate @@ -1140,7 +1154,6 @@ async def custom_tts_endpoint( ) perf_monitor.record("Global unvoiced removal applied") - # --- Warn about potentially conflicting settings --- if enable_smart_stitching and config_manager.get_bool( "audio_processing.enable_silence_trimming", False ): @@ -1224,6 +1237,119 @@ async def custom_tts_endpoint( ) +async def stream_tts_generator( + request: CustomTTSRequest, + audio_prompt_path: str = None, + target_sample_rate: int = 24000, +): + chunk_size = request.chunk_size if request.chunk_size else 120 + raw_chunks = utils.chunk_text_by_sentences(request.text, chunk_size) + text_chunks = [c for c in raw_chunks if c.strip()] + + CROSSFADE_MS = 20 + + yield _create_wav_header(target_sample_rate) + + buffered_tail: Optional[np.ndarray] = None + engine_sr: Optional[int] = None + fade_samples: int = 0 + + base_seed = request.seed if request.seed is not None else get_gen_default_seed() + + for i, chunk_text in enumerate(text_chunks): + try: + current_chunk_seed = base_seed + if base_seed is not None and base_seed >= 0: + current_chunk_seed = base_seed + i + + import asyncio + + loop = asyncio.get_running_loop() + + chunk_audio_tensor, sr = await loop.run_in_executor( + None, + lambda: engine.synthesize( + text=chunk_text, + audio_prompt_path=audio_prompt_path, + temperature=request.temperature or get_gen_default_temperature(), + exaggeration=request.exaggeration or get_gen_default_exaggeration(), + cfg_weight=request.cfg_weight or get_gen_default_cfg_weight(), + seed=current_chunk_seed, + language=request.language or get_gen_default_language(), + ), + ) + + if chunk_audio_tensor is None: + continue + + speed = request.speed_factor or get_gen_default_speed_factor() + if speed != 1.0: + chunk_audio_tensor, _ = utils.apply_speed_factor( + chunk_audio_tensor, sr, speed + ) + + audio_np = chunk_audio_tensor.cpu().numpy().squeeze().astype(np.float32) + + if engine_sr is None: + engine_sr = sr + fade_samples = int(CROSSFADE_MS / 1000 * engine_sr) + + if len(audio_np) < fade_samples * 2: + if buffered_tail is not None: + yield _float32_to_pcm16(buffered_tail) + buffered_tail = None + yield _float32_to_pcm16(audio_np) + continue + + if buffered_tail is not None: + new_head = audio_np[:fade_samples] + fade_out, fade_in = _generate_equal_power_curves(fade_samples) + crossfaded_region = (buffered_tail * fade_out) + (new_head * fade_in) + yield _float32_to_pcm16(crossfaded_region) + body = audio_np[fade_samples:-fade_samples] + else: + body = audio_np[:-fade_samples] + + yield _float32_to_pcm16(body) + buffered_tail = audio_np[-fade_samples:] + + except Exception as e: + logger.error(f"Streaming error on chunk {i}: {e}") + break + + if buffered_tail is not None: + yield _float32_to_pcm16(buffered_tail) + + +@app.post("/tts/stream", tags=["TTS Generation"]) +async def tts_stream_endpoint(request: CustomTTSRequest): + logger.info(f"Stream request received for: {request.text[:30]}...") + audio_prompt_path_str = None + if request.voice_mode == "predefined" and request.predefined_voice_id: + p = ( + get_predefined_voices_path(ensure_absolute=True) + / request.predefined_voice_id + ) + if p.exists(): + audio_prompt_path_str = str(p) + elif request.voice_mode == "clone" and request.reference_audio_filename: + p = ( + get_reference_audio_path(ensure_absolute=True) + / request.reference_audio_filename + ) + if p.exists(): + audio_prompt_path_str = str(p) + + if not engine.MODEL_LOADED: + raise HTTPException(status_code=503, detail="Model not loaded") + + target_sr = get_audio_sample_rate() + return StreamingResponse( + stream_tts_generator(request, audio_prompt_path_str, target_sr), + media_type="audio/wav", + ) + + @app.post("/v1/audio/speech", tags=["OpenAI Compatible"]) async def openai_speech_endpoint(request: OpenAISpeechRequest): # Determine the audio prompt path based on the voice parameter @@ -1235,7 +1361,7 @@ async def openai_speech_endpoint(request: OpenAISpeechRequest): if voice_path_predefined.is_file(): audio_prompt_path = voice_path_predefined elif voice_path_reference.is_file(): - audio_prompt_path = voice_path_reference + audio_prompt_path = reference_audio_path else: raise HTTPException( status_code=404, detail=f"Voice file '{request.voice}' not found." diff --git a/start.py b/start.py index 98a7e93..23ce577 100644 --- a/start.py +++ b/start.py @@ -1,2637 +1,2637 @@ -#!/usr/bin/env python3 -""" -Chatterbox TTS Server - Cross-Platform Launcher Script -======================================================= - -A user-friendly launcher with automatic setup, virtual environment -management, hardware detection, dependency installation, and server startup. - -Features: -- Cross-platform support (Windows, Linux, macOS) -- Automatic GPU detection (NVIDIA, AMD) -- Interactive hardware selection menu -- Virtual environment management -- Dependency installation with progress indication -- Server startup with health checking -- Reinstall/upgrade support - -Usage: - Windows: Double-click start.bat or run: python start.py - Linux: Run: ./start.sh or: python3 start.py - -Options: - --reinstall, -r Remove existing installation and reinstall fresh - --upgrade, -u Upgrade to latest version (keeps hardware selection) - --cpu Install CPU version (skip menu) - --nvidia Install NVIDIA CUDA 12.1 version (skip menu) - --nvidia-cu128 Install NVIDIA CUDA 12.8 version (skip menu) - --rocm Install AMD ROCm version (skip menu) - --portable Use portable Python environment (Windows, skip prompt) - --no-portable Use standard virtual environment (Windows, skip prompt) - --verbose, -v Show detailed installation output - --help, -h Show this help message - -Requirements: - - Python 3.10 or later - - Internet connection for downloading dependencies -""" - -import argparse -import hashlib -import json -import os -import platform -import re -import shutil -import socket -import stat -import subprocess -import sys -import threading -import time -import urllib.request -import zipfile -from datetime import datetime -from pathlib import Path - -# ============================================================================ -# CONFIGURATION -# ============================================================================ - -# TESTING FLAG: Set to True to simulate Python 3.11+ on Windows -# (forces embedded Python fallback even if actual Python version is <3.11) -# This is useful for testing the embedded Python path without installing Python 3.11+ -TEST_EMBEDDED_PYTHON_PATH = False - -# Virtual environment settings -VENV_FOLDER = "venv" -SERVER_SCRIPT = "server.py" -CONFIG_FILE = "config.yaml" - -# Embedded Python settings (Windows fallback for Python 3.11+) -EMBEDDED_PYTHON_DIR = "python_embedded" -EMBEDDED_PYTHON_VERSION = "3.10.11" -EMBEDDED_PYTHON_URL = ( - f"https://www.python.org/ftp/python/{EMBEDDED_PYTHON_VERSION}/" - f"python-{EMBEDDED_PYTHON_VERSION}-embed-amd64.zip" -) -GET_PIP_URL = "https://bootstrap.pypa.io/get-pip.py" - -# SHA-256 hash of the embeddable zip for integrity verification. -# Set to "" to skip verification (not recommended for production). -# To compute: download the file from EMBEDDED_PYTHON_URL, then run: -# python -c "import hashlib; print(hashlib.sha256(open('python-3.10.11-embed-amd64.zip','rb').read()).hexdigest())" -EMBEDDED_PYTHON_SHA256 = "" - -# Installation type identifiers -INSTALL_CPU = "cpu" -INSTALL_NVIDIA = "nvidia" -INSTALL_NVIDIA_CU128 = "nvidia-cu128" -INSTALL_ROCM = "rocm" - -# Requirements file mapping -REQUIREMENTS_MAP = { - INSTALL_CPU: "requirements.txt", - INSTALL_NVIDIA: "requirements-nvidia.txt", - INSTALL_NVIDIA_CU128: "requirements-nvidia-cu128.txt", - INSTALL_ROCM: "requirements-rocm.txt", -} - -# Human-readable names for installation types -INSTALL_NAMES = { - INSTALL_CPU: "CPU Only", - INSTALL_NVIDIA: "NVIDIA GPU (CUDA 12.1)", - INSTALL_NVIDIA_CU128: "NVIDIA GPU (CUDA 12.8 / Blackwell)", - INSTALL_ROCM: "AMD GPU (ROCm 6.4)", -} - -# Chatterbox fork URL (used for CUDA 12.8 installation) -CHATTERBOX_REPO = "git+https://github.com/devnen/chatterbox-v2.git@master" - -# Timeout settings (seconds) -# First run downloads large model files (~2GB). Subsequent starts are much faster. -SERVER_STARTUP_TIMEOUT = 1800 -PORT_CHECK_INTERVAL = 0.5 - -# Global verbose mode flag (set from args) -VERBOSE_MODE = True - - -# ============================================================================ -# ANSI COLOR SUPPORT -# ============================================================================ - - -class Colors: - """ANSI color codes for cross-platform colored terminal output.""" - - CYAN = "\033[96m" - GREEN = "\033[92m" - YELLOW = "\033[93m" - RED = "\033[91m" - BOLD = "\033[1m" - DIM = "\033[2m" - RESET = "\033[0m" - - # Status icons - ICON_SUCCESS = "✓" - ICON_ERROR = "✗" - ICON_WARNING = "⚠" - ICON_INFO = "→" - ICON_WORKING = "●" - - @staticmethod - def is_windows(): - """Check if running on Windows.""" - return platform.system() == "Windows" - - @staticmethod - def is_linux(): - """Check if running on Linux.""" - return platform.system() == "Linux" - - @staticmethod - def is_macos(): - """Check if running on macOS.""" - return platform.system() == "Darwin" - - @classmethod - def enable_windows_colors(cls): - """Enable ANSI color support on Windows 10+.""" - if cls.is_windows(): - try: - import ctypes - - kernel32 = ctypes.windll.kernel32 - # Enable ANSI escape sequences on Windows 10+ - kernel32.SetConsoleMode(kernel32.GetStdHandle(-11), 7) - except Exception: - # If this fails, colors just won't work (non-fatal) - pass - - -# Enable Windows colors at module load time -Colors.enable_windows_colors() - - -# ============================================================================ -# PRINT HELPER FUNCTIONS -# ============================================================================ - - -def print_banner(): - """Print the startup banner.""" - print() - print("=" * 60) - print(" Chatterbox TTS Server - Launcher") - print("=" * 60) - print() - - -def print_header(text): - """Print a section header.""" - print(f"\n{Colors.CYAN}{text}{Colors.RESET}") - - -def print_step(step, total, message): - """Print a numbered step.""" - print(f"\n[{step}/{total}] {message}") - - -def print_substep(message, status="info"): - """ - Print a sub-step with status indicator. - - Args: - message: The message to print - status: One of "done", "error", "warning", "info" - """ - icons = { - "done": (Colors.GREEN, Colors.ICON_SUCCESS), - "error": (Colors.RED, Colors.ICON_ERROR), - "warning": (Colors.YELLOW, Colors.ICON_WARNING), - "info": (Colors.RESET, Colors.ICON_INFO), - } - - color, icon = icons.get(status, (Colors.RESET, Colors.ICON_INFO)) - print(f" {color}{icon}{Colors.RESET} {message}") - - -def print_success(text): - """Print a success message in green.""" - print(f"{Colors.GREEN}{text}{Colors.RESET}") - - -def print_warning(text): - """Print a warning message in yellow.""" - print(f"{Colors.YELLOW}{text}{Colors.RESET}") - - -def print_error(text): - """Print an error message in red.""" - print(f"{Colors.RED}{text}{Colors.RESET}") - - -def print_status_box(host, port): - """Print the final status box with server information.""" - display_host = "localhost" if host == "0.0.0.0" else host - url = f"http://{display_host}:{port}" - - print() - print("=" * 60) - print(f" {Colors.GREEN}🎙️ Chatterbox TTS Server is running!{Colors.RESET}") - print() - print(f" Web Interface: {url}") - print(f" API Docs: {url}/docs") - - if host == "0.0.0.0": - print() - print(" (Also accessible on your local network)") - - print() - print(" Press Ctrl+C to stop the server.") - print("=" * 60) - print() - - -def print_reinstall_hint(): - """Print a hint about how to reinstall.""" - print(f" {Colors.DIM}💡 Tip: To reinstall or upgrade, run:{Colors.RESET}") - print(f" {Colors.DIM} python start.py --reinstall{Colors.RESET}") - print() - - -# ============================================================================ -# COMMAND EXECUTION -# ============================================================================ - - -def run_command(cmd, cwd=None, check=True, capture=False, show_output=False): - """ - Run a shell command. - - Args: - cmd: Command string to execute - cwd: Working directory (optional) - check: If True, raise exception on non-zero exit - capture: If True, capture and return output - show_output: If True, show output in real-time - - Returns: - If capture=True: subprocess.CompletedProcess result - If capture=False: True on success, False on failure - """ - try: - if capture: - result = subprocess.run( - cmd, shell=True, cwd=cwd, capture_output=True, text=True, check=check - ) - return result - - if show_output or VERBOSE_MODE: - # Show output in real-time - result = subprocess.run(cmd, shell=True, cwd=cwd, check=check) - return result.returncode == 0 if not check else True - else: - # Suppress output - result = subprocess.run( - cmd, shell=True, cwd=cwd, capture_output=True, text=True, check=check - ) - return True - - except subprocess.CalledProcessError as e: - if check: - raise - return None if capture else False - except Exception as e: - if VERBOSE_MODE: - print_error(f"Command error: {e}") - return None if capture else False - - -def run_command_with_progress(cmd, cwd=None, description="Working"): - """ - Run a command with a progress indicator for long operations. - - Args: - cmd: Command string to execute - cwd: Working directory (optional) - description: Description to show during progress - - Returns: - True on success, False on failure - """ - if VERBOSE_MODE: - # In verbose mode, just show output directly - print_substep(f"Running: {cmd}", "info") - return run_command(cmd, cwd=cwd, show_output=True, check=False) - - # Start progress indicator in background - stop_progress = threading.Event() - - def progress_indicator(): - """Background thread to show progress spinner.""" - spinner = ["⠋", "⠙", "⠹", "⠸", "⠼", "⠴", "⠦", "⠧", "⠇", "⠏"] - idx = 0 - while not stop_progress.is_set(): - sys.stdout.write(f"\r {spinner[idx]} {description}...") - sys.stdout.flush() - idx = (idx + 1) % len(spinner) - time.sleep(0.1) - # Clear the progress line - sys.stdout.write("\r" + " " * 60 + "\r") - sys.stdout.flush() - - progress_thread = threading.Thread(target=progress_indicator, daemon=True) - progress_thread.start() - - try: - result = subprocess.run( - cmd, shell=True, cwd=cwd, capture_output=True, text=True - ) - - stop_progress.set() - progress_thread.join(timeout=1) - - if result.returncode != 0: - print_substep(f"Command failed with exit code {result.returncode}", "error") - if result.stderr: - # Show last part of error message - error_lines = result.stderr.strip().split("\n") - for line in error_lines[-5:]: - print(f" {line}") - return False - - return True - - except Exception as e: - stop_progress.set() - progress_thread.join(timeout=1) - print_error(f"Error running command: {e}") - return False - - -# ============================================================================ -# PLATFORM DETECTION -# ============================================================================ - - -def is_windows(): - """Check if running on Windows.""" - return platform.system() == "Windows" - - -def is_linux(): - """Check if running on Linux.""" - return platform.system() == "Linux" - - -def is_macos(): - """Check if running on macOS.""" - return platform.system() == "Darwin" - - -def get_platform_name(): - """Get human-readable platform name.""" - system = platform.system() - if system == "Windows": - return "Windows" - elif system == "Linux": - return "Linux" - elif system == "Darwin": - return "macOS" - else: - return system - - -# ============================================================================ -# PYTHON & VIRTUAL ENVIRONMENT FUNCTIONS -# ============================================================================ - - -def check_python_version(): - """ - Verify Python version is 3.10 or later. - Exits with error if version is too old. - """ - major = sys.version_info.major - minor = sys.version_info.minor - - if major < 3 or (major == 3 and minor < 10): - print_error(f"Python 3.10+ required, but found Python {major}.{minor}") - print() - print("Please install Python 3.10 or newer from:") - print(" https://www.python.org/downloads/") - print() - sys.exit(1) - - print_substep(f"Python {major}.{minor}.{sys.version_info.micro} detected", "done") - - -def get_venv_paths(root_dir): - """ - Get paths for virtual environment components. - - Args: - root_dir: Root directory of the project - - Returns: - Tuple of (venv_dir, venv_python, venv_pip) as Path objects - """ - venv_dir = root_dir / VENV_FOLDER - - if is_windows(): - venv_python = venv_dir / "Scripts" / "python.exe" - venv_pip = venv_dir / "Scripts" / "pip.exe" - else: - venv_python = venv_dir / "bin" / "python" - venv_pip = venv_dir / "bin" / "pip" - - return venv_dir, venv_python, venv_pip - - -def create_venv(venv_dir): - """ - Create a virtual environment. - - Args: - venv_dir: Path to create the virtual environment - - Returns: - True on success, False on failure - """ - print_substep("Creating virtual environment...") - - try: - result = subprocess.run( - [sys.executable, "-m", "venv", str(venv_dir)], - capture_output=True, - text=True, - ) - - if result.returncode != 0: - print_substep("Failed to create virtual environment", "error") - if result.stderr: - print(f" {result.stderr[:200]}") - return False - - print_substep("Virtual environment created", "done") - return True - - except Exception as e: - print_substep(f"Error creating venv: {e}", "error") - return False - - -def get_install_state(venv_dir): - """ - Check if installation is complete and get the install type. - - Args: - venv_dir: Path to virtual environment directory - - Returns: - Tuple of (is_installed: bool, install_type: str or None) - """ - install_complete_file = venv_dir / ".install_complete" - install_type_file = venv_dir / ".install_type" - - if not install_complete_file.exists(): - return False, None - - install_type = None - if install_type_file.exists(): - try: - install_type = install_type_file.read_text(encoding="utf-8").strip() - except Exception: - pass - - return True, install_type - - -def save_install_state(venv_dir, install_type): - """ - Save installation state files. - - Args: - venv_dir: Path to virtual environment directory - install_type: Type of installation (cpu, nvidia, nvidia-cu128, rocm) - """ - try: - # Save install type - install_type_file = venv_dir / ".install_type" - install_type_file.write_text(install_type, encoding="utf-8") - - # Save completion marker with timestamp - install_complete_file = venv_dir / ".install_complete" - timestamp = datetime.now().isoformat() - install_complete_file.write_text( - f"Installation completed at {timestamp}\n" f"Type: {install_type}\n", - encoding="utf-8", - ) - except Exception as e: - print_warning(f"Could not save install state: {e}") - - -def clear_install_complete(venv_dir): - """ - Clear only the install complete marker (for upgrades). - - Args: - venv_dir: Path to virtual environment directory - """ - install_complete_file = venv_dir / ".install_complete" - - try: - if install_complete_file.exists(): - install_complete_file.unlink() - except Exception as e: - print_warning(f"Could not clear install marker: {e}") - - -def robust_rmtree(path): - """ - Remove a directory tree with Windows-hardened error handling. - - Uses an onerror callback to strip read-only attributes (common on - extracted zip contents), retries on transient permission errors - (antivirus, Explorer indexing), and falls back to renaming the - directory aside if deletion fails entirely. - - Args: - path: Path to directory to remove - - Returns: - True if directory is gone (deleted or renamed aside), False if stuck - """ - path = Path(path) - if not path.exists(): - return True - - def handle_remove_readonly(func, fpath, exc): - """Clear read-only flag and retry the failed operation.""" - os.chmod(fpath, stat.S_IWRITE) - func(fpath) - - max_retries = 3 - retry_delay = 2 - - for attempt in range(max_retries): - try: - shutil.rmtree(path, onerror=handle_remove_readonly) - return True - except PermissionError: - if attempt < max_retries - 1: - print_substep( - f"Files locked, retrying in {retry_delay}s... " - f"(attempt {attempt + 1}/{max_retries})", - "warning", - ) - time.sleep(retry_delay) - except Exception: - break # Non-permission error, skip to rename fallback - - # Fallback: rename aside so we can proceed even if deletion fails - try: - aside_name = f"{path.name}.old.{int(time.time())}" - aside_path = path.parent / aside_name - path.rename(aside_path) - print_substep( - f"Could not delete folder; renamed to {aside_name}", - "warning", - ) - print_substep("You can safely delete that folder later.", "info") - return True - except Exception: - pass - - return False - - -def remove_venv(venv_dir): - """ - Remove an environment directory (venv or embedded) with robust error handling. - - Args: - venv_dir: Path to environment directory - - Returns: - True on success, False on failure - """ - if not venv_dir.exists(): - return True - - print_substep(f"Removing existing environment ({venv_dir.name})...") - - if robust_rmtree(venv_dir): - print_substep("Environment removed", "done") - return True - - print_error(f"Could not remove: {venv_dir}") - print_substep( - "Try closing any terminals, editors, or antivirus scanning this folder", - "info", - ) - if is_windows(): - print_substep(f'Or run: rmdir /s /q "{venv_dir.name}"', "info") - else: - print_substep(f'Or run: rm -rf "{venv_dir.name}"', "info") - return False - - -# ============================================================================ -# EMBEDDED PYTHON (WINDOWS FALLBACK) -# ============================================================================ - - -def get_embedded_python_paths(root_dir): - """ - Get paths for the embedded Python environment (Windows only). - - Args: - root_dir: Root directory of the project - - Returns: - Tuple of (embedded_dir, embedded_python, embedded_pip) as Path objects - """ - embedded_dir = root_dir / EMBEDDED_PYTHON_DIR - embedded_python = embedded_dir / "python.exe" - embedded_pip = embedded_dir / "Scripts" / "pip.exe" - return embedded_dir, embedded_python, embedded_pip - - -def is_embedded_python_available(root_dir): - """ - Check if embedded Python is already set up and functional. - - Args: - root_dir: Root directory of the project - - Returns: - True if embedded Python is ready to use - """ - embedded_dir, embedded_python, embedded_pip = get_embedded_python_paths(root_dir) - - if not embedded_python.exists() or not embedded_pip.exists(): - return False - - try: - result = subprocess.run( - [str(embedded_python), "--version"], - capture_output=True, - text=True, - timeout=10, - ) - return result.returncode == 0 - except Exception: - return False - - -def download_file(url, dest_path, description="Downloading"): - """ - Download a file from a URL with progress indication. - - Uses urlopen with an explicit per-operation timeout to prevent - indefinite hanging on flaky networks or corporate proxies. - Downloads to a temporary .part file first, then atomically moves - to dest_path so interrupted downloads never leave a valid-looking - but truncated file at the final path. - - Args: - url: URL to download from - dest_path: Local path to save the file - description: Description shown during download - - Returns: - True on success, False on failure - """ - print_substep(f"{description}...") - - dest_path = Path(dest_path) - part_path = dest_path.parent / (dest_path.name + ".part") - - try: - response = urllib.request.urlopen(url, timeout=30) - total_size = int(response.headers.get("Content-Length", 0)) - downloaded = 0 - last_percent = -1 - - with open(part_path, "wb") as f: - while True: - chunk = response.read(8192) - if not chunk: - break - f.write(chunk) - downloaded += len(chunk) - - if total_size > 0: - percent = min(100, int(downloaded * 100 / total_size)) - if percent != last_percent and percent % 5 == 0: - last_percent = percent - mb_done = downloaded / (1024 * 1024) - mb_total = total_size / (1024 * 1024) - sys.stdout.write( - f"\r {Colors.ICON_WORKING} {description}... " - f"{percent}% ({mb_done:.1f}/{mb_total:.1f} MB)" - ) - sys.stdout.flush() - else: - # No Content-Length: show bytes downloaded without percentage - mb_done = downloaded / (1024 * 1024) - if int(mb_done * 10) != last_percent: - last_percent = int(mb_done * 10) - sys.stdout.write( - f"\r {Colors.ICON_WORKING} {description}... " - f"{mb_done:.1f} MB" - ) - sys.stdout.flush() - - sys.stdout.write("\n") - sys.stdout.flush() - - # Atomic move: .part → final path - os.replace(str(part_path), str(dest_path)) - print_substep(f"{description} complete", "done") - return True - - except Exception as e: - sys.stdout.write("\n") - sys.stdout.flush() - print_substep(f"Download failed: {e}", "error") - print_substep(f"You can download manually from: {url}", "info") - return False - - finally: - # Clean up partial download on failure (no-op on success since - # os.replace already moved the .part file to dest_path) - try: - if part_path.exists(): - part_path.unlink() - except Exception: - pass - - -def verify_checksum(file_path, expected_sha256): - """ - Verify SHA-256 hash of a downloaded file. - - Args: - file_path: Path to the file to verify - expected_sha256: Expected hex digest string - - Returns: - True if hash matches, False otherwise - """ - sha256 = hashlib.sha256() - with open(file_path, "rb") as f: - for chunk in iter(lambda: f.read(8192), b""): - sha256.update(chunk) - actual = sha256.hexdigest() - if actual != expected_sha256: - print_substep("Checksum mismatch!", "error") - print_substep(f"Expected: {expected_sha256}", "info") - print_substep(f"Actual: {actual}", "info") - return False - return True - - -def patch_pth_file(embedded_dir): - """ - Patch the python3XX._pth file to enable pip and package imports. - - The embeddable Python distribution ships with a ._pth file that - restricts the import path. We need to uncomment 'import site' and - add 'Lib\\site-packages' so that pip-installed packages are importable. - - Note: pip usage with the embeddable distribution is "not supported" - per CPython docs, but works reliably with this patching approach. - The ._pth format has been stable since Python 3.5. Re-test if - bumping EMBEDDED_PYTHON_VERSION to a new minor release. - - Args: - embedded_dir: Path to the embedded Python directory - - Returns: - True on success, False on failure - """ - try: - # Find the ._pth file (e.g., python310._pth) - pth_files = list(embedded_dir.glob("python3*._pth")) - - if not pth_files: - print_substep("Could not find ._pth file in embedded Python", "error") - return False - - pth_file = pth_files[0] - content = pth_file.read_text(encoding="utf-8") - lines = content.splitlines() - - # Collect path entries, skipping comments, blanks, and the - # import site directive (which we'll re-add at the end in - # the canonical position: paths first, import site last). - path_lines = [] - has_site_packages = False - - for line in lines: - stripped = line.strip() - - # Skip import site (commented or not) — added back at the end - if stripped in ("#import site", "import site"): - continue - - # Skip blank lines and the stock comment - if not stripped or stripped.startswith("#"): - continue - - path_lines.append(stripped) - if "site-packages" in stripped: - has_site_packages = True - - # Add site-packages path if not already present - if not has_site_packages: - path_lines.append("Lib\\site-packages") - - # Add parent directory (project root) so that project modules - # like config.py, engine.py, utils.py are importable. - # The embedded Python dir is always /python_embedded/, - # so ".." resolves to the project root at runtime. - if ".." not in path_lines: - path_lines.append("..") - - # Canonical order: paths first, then import site last - path_lines.append("import site") - - pth_file.write_text("\n".join(path_lines) + "\n", encoding="utf-8") - - if VERBOSE_MODE: - print_substep(f"Patched {pth_file.name}", "done") - - return True - - except Exception as e: - print_substep(f"Failed to patch ._pth file: {e}", "error") - return False - - -def _create_dll_search_sitecustomize(embedded_dir): - """ - Create a sitecustomize.py in the embedded Python directory that configures - Windows DLL search paths at interpreter startup. - - This ensures native extensions (.pyd files) can find their dependent DLLs - regardless of how the embedded Python is launched (via start.py, manually, - or from a subprocess). The file is automatically loaded by site.py - (triggered by 'import site' in the ._pth file). - - No-op on non-Windows platforms. - - Args: - embedded_dir: Path to the embedded Python directory - """ - sitecustomize_path = Path(embedded_dir) / "sitecustomize.py" - - content = """\ -# Auto-generated by start.py -- DO NOT EDIT -# Configures DLL search paths for the embedded Python environment on Windows. -# This ensures native extensions (.pyd) can locate their dependent DLLs. -import os -import sys - -if sys.platform == "win32" and hasattr(os, "add_dll_directory"): - _exe_dir = os.path.dirname(sys.executable) - _sp_dir = os.path.join(_exe_dir, "Lib", "site-packages") - - for _d in [_exe_dir, _sp_dir]: - if os.path.isdir(_d): - try: - os.add_dll_directory(_d) - except OSError: - pass - - # Add *.libs directories (standard wheel pattern for vendored native DLLs, - # created by auditwheel/delvewheel when packaging binary extensions) - if os.path.isdir(_sp_dir): - for _entry in os.listdir(_sp_dir): - if _entry.endswith(".libs"): - _libs_path = os.path.join(_sp_dir, _entry) - if os.path.isdir(_libs_path): - try: - os.add_dll_directory(_libs_path) - except OSError: - pass -""" - - try: - sitecustomize_path.write_text(content, encoding="utf-8") - if VERBOSE_MODE: - print_substep("Created sitecustomize.py for DLL search paths", "done") - except Exception as e: - print_substep(f"Could not create sitecustomize.py: {e}", "warning") - - -def setup_embedded_python(root_dir): - """ - Download and configure an embedded Python 3.10 environment for Windows. - - This creates a fully self-contained Python installation inside the - project folder with pip bootstrapped and ready to install packages. - - Args: - root_dir: Root directory of the project - - Returns: - True on success, False on failure - """ - embedded_dir = root_dir / EMBEDDED_PYTHON_DIR - - # Check if already available - if is_embedded_python_available(root_dir): - print_substep( - f"Embedded Python {EMBEDDED_PYTHON_VERSION} already set up", "done" - ) - return True - - # Clean up any partial previous attempt - if embedded_dir.exists(): - if not robust_rmtree(embedded_dir): - print_substep("Could not clean up partial install", "error") - return False - - print_substep( - f"Setting up portable Python {EMBEDDED_PYTHON_VERSION} environment...", "info" - ) - - zip_path = root_dir / "_python_embedded.zip" - get_pip_path = root_dir / "_get-pip.py" - - try: - # Step 1: Download embeddable Python - if not download_file( - EMBEDDED_PYTHON_URL, - zip_path, - f"Downloading Python {EMBEDDED_PYTHON_VERSION} embeddable package", - ): - return False - - # Step 1b: Verify download integrity - if EMBEDDED_PYTHON_SHA256: - if not verify_checksum(zip_path, EMBEDDED_PYTHON_SHA256): - print_substep( - "Downloaded file may be corrupted. " - "Delete it and try again, or download manually.", - "error", - ) - return False - if VERBOSE_MODE: - print_substep("Checksum verified", "done") - - # Step 1c: Validate zip archive - if not zipfile.is_zipfile(str(zip_path)): - print_substep("Downloaded file is not a valid zip archive", "error") - print_substep("Your network may be returning an error page instead", "info") - return False - - # Step 2: Extract - print_substep("Extracting Python...") - try: - embedded_dir.mkdir(parents=True, exist_ok=True) - with zipfile.ZipFile(str(zip_path), "r") as zf: - zf.extractall(str(embedded_dir)) - print_substep("Python extracted", "done") - except Exception as e: - print_substep(f"Extraction failed: {e}", "error") - return False - - # Step 3: Patch ._pth file for pip and site-packages support - if not patch_pth_file(embedded_dir): - return False - - # Step 3b: Create sitecustomize.py for DLL search path configuration - _create_dll_search_sitecustomize(embedded_dir) - - # Step 4: Bootstrap pip - if not download_file(GET_PIP_URL, get_pip_path, "Downloading pip installer"): - return False - - embedded_python = embedded_dir / "python.exe" - print_substep("Installing pip...") - - pip_cmd = [str(embedded_python), str(get_pip_path)] - if VERBOSE_MODE: - result = subprocess.run(pip_cmd) - else: - result = subprocess.run(pip_cmd, capture_output=True, text=True) - - if result.returncode != 0: - print_substep("Failed to install pip", "error") - if not VERBOSE_MODE and hasattr(result, "stderr") and result.stderr: - error_lines = result.stderr.strip().split("\n")[-3:] - for line in error_lines: - print(f" {line}") - return False - - # Step 5: Verify pip is usable - embedded_pip = embedded_dir / "Scripts" / "pip.exe" - if not embedded_pip.exists(): - print_substep("pip was not created at expected location", "error") - return False - - print_substep("pip installed", "done") - - # Step 6: Install setuptools (provides pkg_resources, needed by perth and others) - # Modern get-pip.py no longer bundles setuptools, but many ML/AI packages - # (including resemble-perth) still import pkg_resources at runtime. - # Note: pkg_resources was removed in setuptools 81+ (targets Python 3.12+). - # On Python 3.10 pip resolves to a compatible older version automatically. - print_substep("Installing setuptools...") - setuptools_cmd = [ - str(embedded_python), - "-m", - "pip", - "install", - "--no-warn-script-location", - "setuptools", - ] - if VERBOSE_MODE: - st_result = subprocess.run(setuptools_cmd) - else: - st_result = subprocess.run(setuptools_cmd, capture_output=True, text=True) - - if st_result.returncode != 0: - print_substep( - "setuptools installation failed (pkg_resources may be unavailable)", - "warning", - ) - else: - print_substep("setuptools installed", "done") - - print() - print_substep( - f"Portable Python {EMBEDDED_PYTHON_VERSION} environment ready", "done" - ) - return True - - except Exception as e: - print_substep(f"Unexpected error during setup: {e}", "error") - return False - - finally: - # Clean up temporary downloads - for temp_file in [zip_path, get_pip_path]: - try: - if temp_file.exists(): - temp_file.unlink() - except Exception: - pass - - -def prompt_portable_install(reason="portability"): - """ - Offer Windows users the portable Python installation option. - - Args: - reason: "compatibility" (Python 3.11+) or "portability" (Python 3.10) - - Returns: - True if user wants portable mode, False for standard venv - """ - major = sys.version_info.major - minor = sys.version_info.minor - - print() - - if reason == "compatibility": - # Python 3.11+ — lead with compatibility problem, mention portability - print(f"{Colors.YELLOW}{'=' * 60}{Colors.RESET}") - print( - f" {Colors.YELLOW}{Colors.ICON_WARNING} Python {major}.{minor} detected" - f" — Portable Mode recommended{Colors.RESET}" - ) - print(f"{Colors.YELLOW}{'=' * 60}{Colors.RESET}") - print() - print(" On Windows, Python 3.11+ lacks pre-built binary packages") - print(" (wheels) for several key dependencies, including ONNX and") - print(" ONNXRuntime. This causes installation failures that are") - print(" difficult to resolve.") - print() - print( - f" {Colors.GREEN}Portable Mode solves this automatically{Colors.RESET}" - f" by using a compatible" - ) - print(" Python runtime, and as a bonus makes your entire installation") - print(" fully portable — copy it to a USB drive, share it as a zip") - print(" file, or move it anywhere.") - else: - # Python 3.10 — lead with portability benefits - print(f"{Colors.CYAN}{'=' * 60}{Colors.RESET}") - print(f" {Colors.CYAN}📦 Portable Mode Available{Colors.RESET}") - print(f"{Colors.CYAN}{'=' * 60}{Colors.RESET}") - print() - print(" The launcher can create a fully self-contained installation.") - print(" The entire project folder — including Python and all") - print(" dependencies — becomes completely portable:") - print() - print(" • Copy to a USB drive and run on another PC") - print(" • Zip the folder and share it with others") - print(" • Move it anywhere on your filesystem") - print(" • No Python installation needed on the target machine") - - print() - - # Option 1 — Portable (default) - print(f" {Colors.BOLD}[1] Install in Portable Mode (recommended){Colors.RESET}") - print(" Creates a self-contained Python environment inside this folder.") - print( - f" Your system Python {major}.{minor} remains" - f" {Colors.GREEN}completely untouched{Colors.RESET}." - ) - print() - - # Option 2 — Standard venv - print(" [2] Standard installation") - if reason == "compatibility": - print( - f" {Colors.DIM}Uses system Python {major}.{minor}" - f" with a virtual environment.{Colors.RESET}" - ) - print( - f" {Colors.DIM}May require Visual C++ Build Tools." - f" Not portable.{Colors.RESET}" - ) - else: - print( - f" {Colors.DIM}Uses a standard virtual environment." - f" Works fine but not portable.{Colors.RESET}" - ) - print() - - while True: - try: - choice = input(" Enter choice [1]: ").strip() - - if choice in ("", "1"): - return True - elif choice == "2": - print() - if reason == "compatibility": - print_substep( - f"Continuing with system Python {major}.{minor}", "warning" - ) - print_substep( - "If installation fails, re-run with:" - " python start.py --reinstall --portable", - "info", - ) - else: - print_substep("Using standard virtual environment", "info") - return False - else: - print_warning(f" Invalid choice '{choice}'. Please enter 1 or 2.") - print() - - except (EOFError, KeyboardInterrupt): - print() - print(" Aborted by user.") - sys.exit(2) - - -# ============================================================================ -# GPU DETECTION -# ============================================================================ - - -def detect_nvidia_gpu(): - """ - Detect NVIDIA GPU using nvidia-smi. - - Returns: - Tuple of (found: bool, gpu_name: str or None) - """ - try: - result = subprocess.run( - ["nvidia-smi", "--query-gpu=name", "--format=csv,noheader,nounits"], - capture_output=True, - text=True, - timeout=10, - ) - - if result.returncode == 0 and result.stdout.strip(): - gpu_name = result.stdout.strip().split("\n")[0] - return True, gpu_name - - return False, None - - except FileNotFoundError: - # nvidia-smi not found - return False, None - except subprocess.TimeoutExpired: - return False, None - except Exception: - return False, None - - -def detect_amd_gpu(): - """ - Detect AMD GPU using rocm-smi. - - Returns: - Tuple of (found: bool, gpu_name: str or None) - """ - try: - result = subprocess.run( - ["rocm-smi", "--showproductname"], - capture_output=True, - text=True, - timeout=10, - ) - - if result.returncode == 0 and result.stdout.strip(): - # Parse output to find GPU name - lines = result.stdout.strip().split("\n") - for line in lines: - if "Card series" in line or "GPU" in line: - # Extract the name part - parts = line.split(":") - if len(parts) > 1: - return True, parts[1].strip() - - # If we got output but couldn't parse name, still report found - return True, "AMD GPU (unknown model)" - - return False, None - - except FileNotFoundError: - # rocm-smi not found - return False, None - except subprocess.TimeoutExpired: - return False, None - except Exception: - return False, None - - -def detect_gpu(): - """ - Detect available GPUs. - - Returns: - Dictionary with detection results: - { - "nvidia": bool, - "nvidia_name": str or None, - "amd": bool, - "amd_name": str or None, - } - """ - nvidia_found, nvidia_name = detect_nvidia_gpu() - amd_found, amd_name = detect_amd_gpu() - - return { - "nvidia": nvidia_found, - "nvidia_name": nvidia_name, - "amd": amd_found, - "amd_name": amd_name, - } - - -# ============================================================================ -# INSTALLATION MENU -# ============================================================================ - - -def get_default_choice(gpu_info): - """ - Determine the default installation choice based on detected hardware. - - Args: - gpu_info: Dictionary from detect_gpu() - - Returns: - Installation type string (INSTALL_CPU, INSTALL_NVIDIA, etc.) - """ - if gpu_info["nvidia"]: - return INSTALL_NVIDIA - elif gpu_info["amd"] and is_linux(): - return INSTALL_ROCM - else: - return INSTALL_CPU - - -def show_installation_menu(gpu_info, default_choice): - """ - Display installation menu and get user choice. - - Args: - gpu_info: Dictionary from detect_gpu() - default_choice: Default installation type - - Returns: - Selected installation type string - """ - # Map install types to menu numbers - MENU_MAP = { - "1": INSTALL_CPU, - "2": INSTALL_NVIDIA, - "3": INSTALL_NVIDIA_CU128, - "4": INSTALL_ROCM, - } - - # Reverse map for showing default - REVERSE_MAP = {v: k for k, v in MENU_MAP.items()} - default_num = REVERSE_MAP[default_choice] - - # Print GPU detection results - print() - print("=" * 60) - print(" Hardware Detection") - print("=" * 60) - print() - - if gpu_info["nvidia"]: - print_success(f" NVIDIA GPU: Detected ({gpu_info['nvidia_name']})") - else: - print(f" NVIDIA GPU: {Colors.DIM}Not detected{Colors.RESET}") - - if gpu_info["amd"]: - print_success(f" AMD GPU: Detected ({gpu_info['amd_name']})") - else: - print(f" AMD GPU: {Colors.DIM}Not detected{Colors.RESET}") - - # Print menu - print() - print("=" * 60) - print(" Select Installation Type") - print("=" * 60) - print() - - # Menu options with descriptions - options = [ - ("1", "CPU Only", "No GPU acceleration - works on any system"), - ("2", "NVIDIA GPU (CUDA 12.1)", "Standard for RTX 20/30/40 series"), - ("3", "NVIDIA GPU (CUDA 12.8)", "For RTX 5090 / Blackwell GPUs only"), - ("4", "AMD GPU (ROCm 6.4)", "For AMD GPUs on Linux"), - ] - - for num, name, desc in options: - # Determine if this is the default - is_default = num == default_num - - # Check for special warnings - warning = "" - if num == "4" and is_windows(): - warning = f" {Colors.YELLOW}⚠️ Not supported on Windows{Colors.RESET}" - - # Build the option line - default_marker = f" {Colors.GREEN}[DEFAULT]{Colors.RESET}" if is_default else "" - - print(f" [{num}] {name}{default_marker}") - print(f" {Colors.DIM}{desc}{warning}{Colors.RESET}") - print() - - # Get user input - while True: - try: - prompt = f" Enter choice [{default_num}]: " - choice = input(prompt).strip() - - # Empty input = default - if not choice: - return default_choice - - # Validate input - if choice in MENU_MAP: - return MENU_MAP[choice] - - print_warning(f" Invalid choice '{choice}'. Please enter 1, 2, 3, or 4.") - print() - - except (EOFError, KeyboardInterrupt): - print() - print(" Aborted by user.") - sys.exit(2) - - -# ============================================================================ -# INSTALLATION FUNCTIONS -# ============================================================================ - - -def upgrade_pip(venv_python): - """ - Upgrade pip in the environment. - - Args: - venv_python: Path to the Python executable (venv or embedded) - """ - print_substep("Upgrading pip...") - - cmd = f'"{venv_python}" -m pip install --upgrade pip' - - # We force check=True here because having an old pip causes the - # dependency resolution errors you are seeing - try: - if VERBOSE_MODE: - subprocess.check_call(cmd, shell=True) - else: - subprocess.check_call( - cmd, shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL - ) - - print_substep("pip upgraded", "done") - return True - except subprocess.CalledProcessError: - print_substep("pip upgrade failed", "warning") - return False - - -def install_requirements(venv_pip, requirements_file, root_dir): - """ - Install dependencies from a requirements file. - - Args: - venv_pip: Path to pip executable in venv - requirements_file: Name of requirements file - root_dir: Root directory of the project - - Returns: - True on success, False on failure - """ - requirements_path = root_dir / requirements_file - - if not requirements_path.exists(): - print_error(f"Requirements file not found: {requirements_file}") - return False - - print_substep(f"Installing from {requirements_file}...") - - # Suppress pip warnings about scripts not on PATH (common with embedded Python) - cmd = f'"{venv_pip}" install --no-warn-script-location -r "{requirements_path}"' - - success = run_command_with_progress( - cmd, - cwd=str(root_dir), - description=f"Installing dependencies from {requirements_file}", - ) - - if success: - print_substep("Dependencies installed", "done") - else: - print_substep("Dependency installation failed", "error") - - return success - - -def install_chatterbox_no_deps(venv_pip): - """ - Install Chatterbox TTS without dependencies (for CUDA 12.8). - - Args: - venv_pip: Path to pip executable in venv - - Returns: - True on success, False on failure - """ - print_substep("Installing Chatterbox TTS (--no-deps to preserve PyTorch 2.8)...") - - cmd = f'"{venv_pip}" install --no-deps {CHATTERBOX_REPO}' - - success = run_command_with_progress(cmd, description="Installing Chatterbox TTS") - - if success: - print_substep("Chatterbox TTS installed", "done") - else: - print_substep("Chatterbox TTS installation failed", "error") - - return success - - -def perform_installation(venv_pip, install_type, root_dir): - """ - Perform installation based on selected type. - - Args: - venv_pip: Path to pip executable in venv - install_type: One of INSTALL_CPU, INSTALL_NVIDIA, INSTALL_NVIDIA_CU128, INSTALL_ROCM - root_dir: Root directory of the project - - Returns: - True on success, False on failure - """ - requirements_file = REQUIREMENTS_MAP.get(install_type) - - if not requirements_file: - print_error(f"Unknown installation type: {install_type}") - return False - - # Step 1: Install requirements - if not install_requirements(venv_pip, requirements_file, root_dir): - return False - - # Step 2: For CUDA 12.8, install chatterbox separately with --no-deps - if install_type == INSTALL_NVIDIA_CU128: - if not install_chatterbox_no_deps(venv_pip): - return False - - return True - - -def _patch_chatterbox_watermarker(env_dir, use_embedded): - """ - Patch installed chatterbox source files to make the Perth watermarker - gracefully optional. If perth fails to load or PerthImplicitWatermarker - is None, the server will skip watermarking instead of crashing. - - Uses a no-op watermarker class so that all call sites (apply_watermark) - continue to work without modification — they just pass audio through - unchanged. - - This patch is idempotent: re-running it on already-patched files is safe. - - Args: - env_dir: Path to environment directory (venv or python_embedded) - use_embedded: True if using embedded Python environment - """ - # Locate site-packages (differs between embedded, Windows venv, Linux/macOS venv) - if use_embedded or is_windows(): - site_packages = env_dir / "Lib" / "site-packages" - else: - # Linux/macOS venv: lib/python3.X/site-packages - lib_dir = env_dir / "lib" - site_packages = None - if lib_dir.exists(): - for child in sorted(lib_dir.iterdir()): - if child.name.startswith("python3") and child.is_dir(): - candidate = child / "site-packages" - if candidate.is_dir(): - site_packages = candidate - break - if site_packages is None: - print_substep( - "Could not locate site-packages, skipping watermarker patch", - "warning", - ) - return - - # Find chatterbox package directory (name varies by package version) - chatterbox_dir = None - for name in ["chatterbox", "chatterbox_tts"]: - candidate = site_packages / name - if candidate.is_dir(): - chatterbox_dir = candidate - break - - if chatterbox_dir is None: - if VERBOSE_MODE: - print_substep( - "Chatterbox package not found, skipping watermarker patch", "info" - ) - return - - SENTINEL = "# [patched by start.py: watermarker made optional]" - INIT_TARGET = "self.watermarker = perth.PerthImplicitWatermarker()" - target_files = ["tts.py", "tts_turbo.py", "mtl_tts.py", "vc.py"] - patched_count = 0 - - for filename in target_files: - filepath = chatterbox_dir / filename - if not filepath.exists(): - continue - - try: - content = filepath.read_text(encoding="utf-8") - except Exception as e: - print_substep(f"{filename}: could not read ({e}), skipping", "warning") - continue - - # Idempotency: skip if already patched - if SENTINEL in content: - if VERBOSE_MODE: - print_substep(f"{filename}: already patched", "info") - continue - - if INIT_TARGET not in content: - if VERBOSE_MODE: - print_substep(f"{filename}: target pattern not found, skipping", "info") - continue - - # Determine whether this file uses the logging module - has_logger = "import logging" in content or "getLogger" in content - if has_logger: - log_line = ( - "logger.warning(" - '"Perth watermarker unavailable ' - '\\u2014 audio will not be watermarked")' - ) - else: - log_line = ( - "print(" - '"Warning: Perth watermarker unavailable ' - '\\u2014 audio will not be watermarked")' - ) - - # Build the replacement block - lines = content.split("\n") - new_lines = [] - - for line in lines: - if INIT_TARGET in line and line.lstrip().startswith("self."): - indent = line[: len(line) - len(line.lstrip())] - new_lines.append(f"{indent}{SENTINEL}") - new_lines.append(f"{indent}try:") - new_lines.append( - f"{indent} self.watermarker = perth.PerthImplicitWatermarker()" - ) - new_lines.append(f"{indent}except Exception:") - new_lines.append(f"{indent} class _NoOpWatermarker:") - new_lines.append( - f"{indent} def apply_watermark(self, wav, *args, **kwargs):" - ) - new_lines.append(f"{indent} return wav") - new_lines.append(f"{indent} self.watermarker = _NoOpWatermarker()") - new_lines.append(f"{indent} {log_line}") - else: - new_lines.append(line) - - try: - filepath.write_text("\n".join(new_lines), encoding="utf-8") - print_substep(f"{filename}: watermarker made optional", "done") - patched_count += 1 - except Exception as e: - print_substep(f"{filename}: could not write ({e})", "warning") - - if patched_count > 0: - print_substep( - f"Patched {patched_count} file(s) for optional watermarking", "done" - ) - elif VERBOSE_MODE: - print_substep("No files needed watermarker patching", "info") - - -def verify_installation(venv_python): - """ - Verify critical dependencies are installed correctly. - - Args: - venv_python: Path to Python executable in venv - - Returns: - True if verification passed, False otherwise - """ - print_substep("Verifying installation...") - - # Python script to run inside the venv to test imports - test_script = """ -import sys -import json - -results = {} - -# Test PyTorch -try: - import torch - results["torch"] = { - "ok": True, - "version": torch.__version__, - "cuda_available": torch.cuda.is_available(), - "cuda_version": torch.version.cuda if torch.cuda.is_available() else None, - "gpu_count": torch.cuda.device_count() if torch.cuda.is_available() else 0, - "gpu_name": torch.cuda.get_device_name(0) if torch.cuda.is_available() and torch.cuda.device_count() > 0 else None, - } -except Exception as e: - results["torch"] = {"ok": False, "error": str(e)} - -# Test FastAPI -try: - import fastapi - results["fastapi"] = {"ok": True, "version": fastapi.__version__} -except Exception as e: - results["fastapi"] = {"ok": False, "error": str(e)} - -# Test Chatterbox -try: - # Try different import paths - try: - import chatterbox - results["chatterbox"] = {"ok": True} - except ImportError: - from chatterbox_tts import ChatterboxTTS - results["chatterbox"] = {"ok": True} -except Exception as e: - results["chatterbox"] = {"ok": False, "error": str(e)} - -# Test audio libraries -try: - import soundfile - import librosa - results["audio"] = {"ok": True} -except Exception as e: - results["audio"] = {"ok": False, "error": str(e)} - -print(json.dumps(results)) -""" - - try: - result = subprocess.run( - [str(venv_python), "-c", test_script], - capture_output=True, - text=True, - timeout=60, - ) - - if result.returncode != 0: - print_substep("Verification script returned non-zero", "warning") - if result.stderr: - # Show relevant error info - error_lines = result.stderr.strip().split("\n")[-3:] - for line in error_lines: - print(f" {line}") - return False - - # Parse JSON results - try: - results = json.loads(result.stdout) - except json.JSONDecodeError: - print_substep("Could not parse verification results", "warning") - return False - - all_ok = True - - # Report PyTorch status - torch_result = results.get("torch", {}) - if torch_result.get("ok"): - version_str = torch_result.get("version", "unknown") - - if torch_result.get("cuda_available"): - cuda_ver = torch_result.get("cuda_version", "unknown") - gpu_name = torch_result.get("gpu_name", "unknown") - print_substep(f"PyTorch {version_str} with CUDA {cuda_ver}", "done") - print_substep(f"GPU: {gpu_name}", "done") - else: - print_substep(f"PyTorch {version_str} (CPU mode)", "done") - else: - error = torch_result.get("error", "unknown error") - print_substep(f"PyTorch: {error}", "error") - all_ok = False - - # Report FastAPI status - fastapi_result = results.get("fastapi", {}) - if fastapi_result.get("ok"): - version = fastapi_result.get("version", "") - print_substep(f"FastAPI {version}", "done") - else: - error = fastapi_result.get("error", "unknown error") - print_substep(f"FastAPI: {error}", "error") - all_ok = False - - # Report Chatterbox status - chatterbox_result = results.get("chatterbox", {}) - if chatterbox_result.get("ok"): - print_substep("Chatterbox TTS", "done") - else: - error = chatterbox_result.get("error", "unknown error") - print_substep(f"Chatterbox: {error}", "error") - all_ok = False - - # Report audio libraries status - audio_result = results.get("audio", {}) - if audio_result.get("ok"): - print_substep("Audio libraries (soundfile, librosa)", "done") - else: - error = audio_result.get("error", "unknown error") - print_substep(f"Audio libraries: {error}", "error") - all_ok = False - - return all_ok - - except subprocess.TimeoutExpired: - print_substep("Verification timed out", "warning") - return False - except Exception as e: - print_substep(f"Verification error: {e}", "warning") - return False - - -# ============================================================================ -# SERVER MANAGEMENT -# ============================================================================ - - -def read_config(root_dir): - """ - Read host and port from config.yaml using simple parsing. - - Does not require PyYAML - uses regex-based parsing. - - Args: - root_dir: Root directory of the project - - Returns: - Tuple of (host: str, port: int) - """ - config_file = root_dir / CONFIG_FILE - - # Default values - host = "0.0.0.0" - port = 8004 - - if not config_file.exists(): - return host, port - - try: - content = config_file.read_text(encoding="utf-8") - - # Simple regex-based parsing for host and port - # This handles basic YAML structure without full parsing - - # Look for host setting - host_match = re.search( - r'^\s*host:\s*["\']?([^"\'#\n\r]+)["\']?', content, re.MULTILINE - ) - if host_match: - parsed_host = host_match.group(1).strip() - if parsed_host: - host = parsed_host - - # Look for port setting - port_match = re.search(r"^\s*port:\s*(\d+)", content, re.MULTILINE) - if port_match: - parsed_port = int(port_match.group(1)) - if 1 <= parsed_port <= 65535: - port = parsed_port - - except Exception as e: - # Silently use defaults on any error - if VERBOSE_MODE: - print_warning(f"Could not parse config.yaml: {e}") - - return host, port - - -def check_port_in_use(host, port): - """ - Check if a port is already in use. - - Args: - host: Host address - port: Port number - - Returns: - True if port is in use, False otherwise - """ - try: - sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM) - sock.settimeout(1) - - # Use localhost for checking if host is 0.0.0.0 - check_host = "127.0.0.1" if host == "0.0.0.0" else host - - result = sock.connect_ex((check_host, port)) - sock.close() - - return result == 0 - - except socket.error: - return False - - -def wait_for_server(host, port, timeout=SERVER_STARTUP_TIMEOUT): - """ - Wait for server to become ready by polling the port. - - Args: - host: Host address - port: Port number - timeout: Maximum seconds to wait - - Returns: - True if server is ready, False if timeout - """ - print_substep( - "Waiting for server to start (this may take a few minutes on first run)..." - ) - - start_time = time.time() - check_host = "127.0.0.1" if host == "0.0.0.0" else host - - # Progress indicator - sys.stdout.write(" ") - sys.stdout.flush() - - dots = 0 - last_dot_time = start_time - - while time.time() - start_time < timeout: - try: - sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM) - sock.settimeout(1) - result = sock.connect_ex((check_host, port)) - sock.close() - - if result == 0: - # Server is ready - sys.stdout.write("\n") - sys.stdout.flush() - elapsed = time.time() - start_time - print_substep(f"Server ready! (took {elapsed:.1f}s)", "done") - return True - - except socket.error: - pass - - # Show progress dots - current_time = time.time() - if current_time - last_dot_time >= 2: - sys.stdout.write(".") - sys.stdout.flush() - dots += 1 - last_dot_time = current_time - - # Line wrap every 30 dots - if dots % 30 == 0: - sys.stdout.write("\n ") - sys.stdout.flush() - - time.sleep(PORT_CHECK_INTERVAL) - - # Timeout reached - sys.stdout.write("\n") - sys.stdout.flush() - print_substep(f"Timeout after {timeout}s waiting for server", "error") - return False - - -def launch_server(venv_python, root_dir): - """ - Launch the server as a subprocess. - - Args: - venv_python: Path to Python executable in venv - root_dir: Root directory of the project - - Returns: - subprocess.Popen process object - """ - server_script = root_dir / SERVER_SCRIPT - - if not server_script.exists(): - print_error(f"{SERVER_SCRIPT} not found!") - return None - - print_substep(f"Starting {SERVER_SCRIPT}...") - - # Create subprocess - # On Windows, we don't want to create a new console window - kwargs = {} - if is_windows(): - # CREATE_NO_WINDOW flag - kwargs["creationflags"] = 0 - - # For embedded Python on Windows, ensure the interpreter directory is on - # PATH so that python310.dll, vcruntime140.dll, and other co-located DLLs - # are discoverable by Windows when loading native extensions (.pyd files). - # This complements the sitecustomize.py os.add_dll_directory() approach. - env = None - embedded_dir = root_dir / EMBEDDED_PYTHON_DIR - if ( - is_windows() - and embedded_dir.exists() - and str(venv_python).startswith(str(embedded_dir)) - ): - env = os.environ.copy() - env["PATH"] = ( - f"{embedded_dir}{os.pathsep}" - f"{embedded_dir / 'Scripts'}{os.pathsep}" - f"{env.get('PATH', '')}" - ) - if VERBOSE_MODE: - print_substep("Injected embedded Python into subprocess PATH", "info") - - process = subprocess.Popen( - [str(venv_python), str(server_script)], - cwd=str(root_dir), - env=env, - **kwargs, - ) - - return process - - -def cleanup_server(process): - """ - Clean up server process gracefully. - - Args: - process: subprocess.Popen process object - """ - if process is None: - return - - if process.poll() is not None: - # Process already terminated - return - - try: - # Try graceful termination first - process.terminate() - - try: - process.wait(timeout=5) - except subprocess.TimeoutExpired: - # Force kill if graceful shutdown fails - print_substep("Force stopping server...", "warning") - process.kill() - try: - process.wait(timeout=3) - except subprocess.TimeoutExpired: - # Give up - process may be orphaned - pass - - except Exception as e: - # Process might already be gone - if VERBOSE_MODE: - print_warning(f"Error during cleanup: {e}") - - -# ============================================================================ -# ARGUMENT PARSER -# ============================================================================ - - -def parse_args(): - """ - Parse command-line arguments. - - Returns: - argparse.Namespace with parsed arguments - """ - parser = argparse.ArgumentParser( - description="Chatterbox TTS Server - Launcher", - formatter_class=argparse.RawDescriptionHelpFormatter, - epilog=""" -Examples: - python start.py # Normal start (shows menu if first run) - python start.py --reinstall # Remove and reinstall (shows menu) - python start.py --upgrade # Upgrade keeping current hardware choice - python start.py --nvidia # Install/start with NVIDIA CUDA 12.1 - python start.py --nvidia-cu128 # Install/start with NVIDIA CUDA 12.8 - python start.py --cpu # Install/start with CPU only - python start.py --rocm # Install/start with AMD ROCm - python start.py --portable # Use portable mode (Windows) - python start.py --no-portable # Use standard venv (Windows) - python start.py -v # Verbose mode (show all output) -""", - ) - - # Reinstall/upgrade options - action_group = parser.add_argument_group("Actions") - action_group.add_argument( - "--reinstall", - "-r", - action="store_true", - help="Remove existing installation and reinstall fresh (prompts for hardware selection)", - ) - action_group.add_argument( - "--upgrade", - "-u", - action="store_true", - help="Upgrade to latest version (keeps current hardware selection)", - ) - - # Direct installation type selection - install_group = parser.add_argument_group("Installation Type (skip menu)") - install_group.add_argument( - "--cpu", action="store_true", help="Install CPU-only version" - ) - install_group.add_argument( - "--nvidia", action="store_true", help="Install NVIDIA CUDA 12.1 version" - ) - install_group.add_argument( - "--nvidia-cu128", - action="store_true", - help="Install NVIDIA CUDA 12.8 version (for RTX 5090/Blackwell)", - ) - install_group.add_argument( - "--rocm", action="store_true", help="Install AMD ROCm version (Linux only)" - ) - - # Environment mode (Windows) - env_group = parser.add_argument_group("Environment Mode (Windows)") - env_group.add_argument( - "--portable", - action="store_true", - help="Use portable Python environment (Windows only, skip prompt)", - ) - env_group.add_argument( - "--no-portable", - action="store_true", - help="Use standard virtual environment instead of portable (skip prompt)", - ) - - # Other options - other_group = parser.add_argument_group("Other Options") - other_group.add_argument( - "--verbose", "-v", action="store_true", help="Show detailed installation output" - ) - - return parser.parse_args() - - -def get_install_type_from_args(args): - """ - Get installation type from command-line arguments. - - Args: - args: Parsed arguments from argparse - - Returns: - Installation type string or None if not specified - """ - if args.cpu: - return INSTALL_CPU - elif args.nvidia: - return INSTALL_NVIDIA - elif args.nvidia_cu128: - return INSTALL_NVIDIA_CU128 - elif args.rocm: - return INSTALL_ROCM - - return None - - -# ============================================================================ -# MAIN FUNCTION -# ============================================================================ - - -def main(): - """Main entry point for the launcher.""" - global VERBOSE_MODE - - # Parse command-line arguments - args = parse_args() - if args.verbose: - VERBOSE_MODE = True - - # Get root directory (where this script is located) - root_dir = Path(__file__).parent.absolute() - - # Print banner - print_banner() - - # Validate portable mode flags - if args.portable and args.no_portable: - print_error("Cannot use --portable and --no-portable together.") - sys.exit(1) - - if args.portable and not is_windows(): - print_error("Portable mode is only available on Windows.") - print("On Linux and macOS, the standard virtual environment is used.") - sys.exit(1) - - # Total steps for progress display - total_steps = 6 - - # ======================================================================== - # Step 1: Check Python version - # ======================================================================== - print_step(1, total_steps, "Checking Python installation...") - check_python_version() - - # Portable mode decision (Windows only) - # Determines whether to use the self-contained embedded Python environment - # or a standard virtual environment. On Linux/macOS, always uses venv. - use_embedded = False - - if not is_windows(): - use_embedded = False - - elif args.upgrade: - # During upgrade, preserve existing environment type - if args.portable or args.no_portable: - print_substep( - "--portable/--no-portable ignored during --upgrade. " - "Use --reinstall to switch modes.", - "warning", - ) - if is_embedded_python_available(root_dir): - use_embedded = True - print_substep( - f"Using existing portable Python {EMBEDDED_PYTHON_VERSION}", "done" - ) - else: - use_embedded = False - - elif args.no_portable: - use_embedded = False - if sys.version_info >= (3, 11): - print_substep( - f"Using system Python {sys.version_info.major}.{sys.version_info.minor}" - f" (--no-portable). Build tools may be required.", - "warning", - ) - - elif args.portable: - use_embedded = True - print_substep("Portable Mode selected via --portable flag", "info") - - elif TEST_EMBEDDED_PYTHON_PATH: - # In test mode: automatically use portable Python (no prompt) - use_embedded = True - print() - print(f"{Colors.YELLOW}{'=' * 60}{Colors.RESET}") - print( - f" {Colors.YELLOW}{Colors.ICON_INFO} TEST MODE: Simulating Python 3.11+" - ) - print( - f" {Colors.YELLOW} → Forcing portable Python environment{Colors.RESET}" - ) - print(f"{Colors.YELLOW}{'=' * 60}{Colors.RESET}") - print() - - elif not args.reinstall and is_embedded_python_available(root_dir): - # Existing portable environment found — reuse it - use_embedded = True - print_substep( - f"Using existing portable Python {EMBEDDED_PYTHON_VERSION}", "done" - ) - - elif sys.version_info >= (3, 11): - use_embedded = prompt_portable_install(reason="compatibility") - - else: - # Python 3.10 on Windows — offer portability - use_embedded = prompt_portable_install(reason="portability") - - # Set up portable Python if chosen but not yet available - if use_embedded and not is_embedded_python_available(root_dir): - print() - if not setup_embedded_python(root_dir): - print() - print_error( - "Could not set up portable Python environment. " - "Falling back to system Python." - ) - if sys.version_info >= (3, 11): - print_substep( - "You may need Visual C++ Build Tools for a successful install", - "warning", - ) - use_embedded = False - - # ======================================================================== - # Step 2: Setup paths - # ======================================================================== - print_step(2, total_steps, "Setting up environment...") - - if use_embedded: - venv_dir, venv_python, venv_pip = get_embedded_python_paths(root_dir) - print_substep(f"Project directory: {root_dir}", "info") - print_substep(f"Python environment: {venv_dir} (portable)", "info") - else: - venv_dir, venv_python, venv_pip = get_venv_paths(root_dir) - print_substep(f"Project directory: {root_dir}", "info") - print_substep(f"Virtual environment: {venv_dir}", "info") - - # ======================================================================== - # Step 3: Handle reinstall/upgrade flags - # ======================================================================== - existing_type = None - - if args.reinstall: - print_step(3, total_steps, "Preparing fresh reinstall...") - - # Remove all possible environment directories (venv and/or embedded) - for env_name in [VENV_FOLDER, EMBEDDED_PYTHON_DIR]: - env_path = root_dir / env_name - if env_path.exists(): - if not remove_venv(env_path): - print_error(f"Could not remove {env_name}/.") - print_substep( - f"Please manually delete the '{env_name}' folder and try again.", - "info", - ) - sys.exit(1) - - # Re-setup embedded Python if that path was chosen - if use_embedded: - if not setup_embedded_python(root_dir): - print_error("Failed to set up embedded Python after reinstall!") - print_substep( - "Try again, or run without --reinstall to use system Python.", - "info", - ) - sys.exit(1) - # Refresh paths after re-creation - venv_dir, venv_python, venv_pip = get_embedded_python_paths(root_dir) - - print_substep("Ready for fresh installation", "done") - - elif args.upgrade: - print_step(3, total_steps, "Preparing upgrade...") - is_installed, existing_type = get_install_state(venv_dir) - - if is_installed and existing_type: - print_substep( - f"Current installation: {INSTALL_NAMES.get(existing_type, existing_type)}", - "info", - ) - print_substep( - "Upgrading will reinstall dependencies with the same hardware selection", - "info", - ) - # Clear only the install complete marker - clear_install_complete(venv_dir) - else: - print_substep( - "No existing installation found, will perform fresh install", "warning" - ) - - else: - print_step(3, total_steps, "Checking existing installation...") - is_installed, existing_type = get_install_state(venv_dir) - - if is_installed: - type_name = INSTALL_NAMES.get(existing_type, existing_type) - print_substep(f"Found existing {type_name} installation", "done") - - # Warn Windows users on Python 3.11+ who are using system Python - if is_windows() and sys.version_info >= (3, 11) and not use_embedded: - major = sys.version_info.major - minor = sys.version_info.minor - print() - print_warning( - f" Note: You're running Python {major}.{minor} on Windows." - ) - print_warning( - " If you experience CUDA or dependency issues, your Python" - ) - print_warning(" version may be the cause. Consider reinstalling with") - print_warning( - " portable mode: python start.py --reinstall --portable" - ) - else: - print_substep("No existing installation found", "info") - - # ======================================================================== - # Step 4: Installation flow (if needed) - # ======================================================================== - is_installed, current_type = get_install_state(venv_dir) - - if not is_installed: - print_step(4, total_steps, "Installing Chatterbox TTS Server...") - - # Create environment if it doesn't exist - if not venv_dir.exists(): - if use_embedded: - # Re-setup embedded Python (e.g., after a partial failure) - if not setup_embedded_python(root_dir): - print_error("Failed to set up embedded Python environment!") - sys.exit(1) - venv_dir, venv_python, venv_pip = get_embedded_python_paths(root_dir) - else: - if not create_venv(venv_dir): - print_error("Failed to create virtual environment!") - print() - print("Try creating it manually:") - print(f" python -m venv {VENV_FOLDER}") - print() - sys.exit(1) - - # Determine installation type - install_type = None - - # First check CLI flags - install_type = get_install_type_from_args(args) - - # If upgrading, use the existing type - if install_type is None and existing_type: - install_type = existing_type - print_substep( - f"Using existing hardware selection: {INSTALL_NAMES.get(install_type, install_type)}", - "info", - ) - - # If still no type, show menu - if install_type is None: - print() - print_substep("Detecting available hardware...", "info") - gpu_info = detect_gpu() - default_choice = get_default_choice(gpu_info) - install_type = show_installation_menu(gpu_info, default_choice) - - # Show selected type - type_name = INSTALL_NAMES.get(install_type, install_type) - print() - print_substep(f"Selected: {type_name}", "done") - - # ROCm warning on Windows - if install_type == INSTALL_ROCM and is_windows(): - print() - print_warning("=" * 60) - print_warning(" ⚠️ WARNING: ROCm is not supported on Windows!") - print_warning("=" * 60) - print() - print_warning(" ROCm (AMD GPU acceleration) only works on Linux.") - print_warning(" Installation will proceed, but GPU acceleration") - print_warning(" will NOT work. The server will run on CPU only.") - print() - - try: - response = input(" Continue anyway? (y/n) [n]: ").strip().lower() - if response != "y": - print() - print(" Installation cancelled.") - print(" Tip: Use --nvidia for NVIDIA GPUs or --cpu for CPU-only.") - sys.exit(2) - except (EOFError, KeyboardInterrupt): - print() - print(" Cancelled.") - sys.exit(2) - - print() - - # Upgrade pip - print() - upgrade_pip(venv_python) - - # Perform installation - print() - success = perform_installation(venv_pip, install_type, root_dir) - - if not success: - print() - print_error("=" * 60) - print_error(" Installation failed!") - print_error("=" * 60) - print() - print("Troubleshooting tips:") - print() - print(" 1. Check your internet connection") - print(" 2. Try running with --verbose for more details:") - print(" python start.py --reinstall --verbose") - print() - print(" 3. Check if you have enough disk space") - print() - print(" 4. Try installing manually:") - requirements_file = REQUIREMENTS_MAP.get(install_type, "requirements.txt") - print(f" pip install -r {requirements_file}") - if install_type == INSTALL_NVIDIA_CU128: - print(f" pip install --no-deps {CHATTERBOX_REPO}") - print() - sys.exit(1) - - # Patch chatterbox to make watermarker gracefully optional - print() - print_substep("Applying post-install patches...") - _patch_chatterbox_watermarker(venv_dir, use_embedded) - - # Verify installation - print() - verification_ok = verify_installation(venv_python) - - if not verification_ok: - print() - print_warning("Installation verification had some issues.") - print_warning("The server may still work. Attempting to continue...") - - # Save installation state - save_install_state(venv_dir, install_type) - - print() - print_success("=" * 60) - print_success(" Installation complete!") - print_success("=" * 60) - - else: - print_step(4, total_steps, "Using existing installation...") - type_name = INSTALL_NAMES.get(current_type, current_type or "unknown") - print_substep(f"Installation type: {type_name}", "done") - - # ======================================================================== - # Step 5: Read configuration - # ======================================================================== - print_step(5, total_steps, "Loading configuration...") - - host, port = read_config(root_dir) - print_substep(f"Server will run on {host}:{port}", "done") - - # Check if port is already in use - if check_port_in_use(host, port): - print() - print_error("=" * 60) - print_error(f" Port {port} is already in use!") - print_error("=" * 60) - print() - print("Another instance may be running, or another program is using this port.") - print() - print("Options:") - print(f" 1. Stop the other process using port {port}") - print(f" 2. Change the port in {CONFIG_FILE}") - print() - sys.exit(1) - - # ======================================================================== - # Step 6: Launch server - # ======================================================================== - print_step(6, total_steps, "Launching Chatterbox TTS Server...") - - server_process = launch_server(venv_python, root_dir) - - if server_process is None: - print_error("Failed to launch server!") - sys.exit(1) - - # Wait for server to become ready - server_ready = wait_for_server(host, port) - - if not server_ready: - print() - print_error("=" * 60) - print_error(" Server failed to start!") - print_error("=" * 60) - print() - print("The server did not become ready within the timeout period.") - print() - print("Common causes:") - print(" - Missing CUDA drivers (for GPU installation)") - print(" - Insufficient memory (model requires ~8GB+ VRAM)") - print(" - Network issues downloading the model") - print(" - Port conflict") - print() - print("Check the server output above for error messages.") - print() - print("Try running with verbose mode for more details:") - print(" python start.py --verbose") - print() - - cleanup_server(server_process) - sys.exit(1) - - # Show success status - print_status_box(host, port) - print_reinstall_hint() - - # ======================================================================== - # Keep running until interrupted - # ======================================================================== - try: - while True: - # Check if server process is still running - exit_code = server_process.poll() - - if exit_code is not None: - # Server has exited - print() - if exit_code == 0: - print_substep("Server stopped normally", "done") - else: - print_substep(f"Server exited with code {exit_code}", "warning") - break - - time.sleep(1) - - except KeyboardInterrupt: - print() - print() - print("-" * 40) - print("Shutting down Chatterbox TTS Server...") - print("-" * 40) - - cleanup_server(server_process) - - print() - print("Server stopped. Goodbye!") - print() - sys.exit(0) - - # Clean up - cleanup_server(server_process) - - # Exit with server's exit code - exit_code = server_process.returncode if server_process.returncode else 0 - sys.exit(exit_code) - - -# ============================================================================ -# ENTRY POINT -# ============================================================================ - -if __name__ == "__main__": - try: - main() - except KeyboardInterrupt: - print() - print() - print("Interrupted by user.") - sys.exit(2) - except Exception as e: - print() - print_error(f"Unexpected error: {e}") - print() - if VERBOSE_MODE: - import traceback - - traceback.print_exc() - else: - print("Run with --verbose for more details.") - print() - sys.exit(1) +#!/usr/bin/env python3 +""" +Chatterbox TTS Server - Cross-Platform Launcher Script +======================================================= + +A user-friendly launcher with automatic setup, virtual environment +management, hardware detection, dependency installation, and server startup. + +Features: +- Cross-platform support (Windows, Linux, macOS) +- Automatic GPU detection (NVIDIA, AMD) +- Interactive hardware selection menu +- Virtual environment management +- Dependency installation with progress indication +- Server startup with health checking +- Reinstall/upgrade support + +Usage: + Windows: Double-click start.bat or run: python start.py + Linux: Run: ./start.sh or: python3 start.py + +Options: + --reinstall, -r Remove existing installation and reinstall fresh + --upgrade, -u Upgrade to latest version (keeps hardware selection) + --cpu Install CPU version (skip menu) + --nvidia Install NVIDIA CUDA 12.1 version (skip menu) + --nvidia-cu128 Install NVIDIA CUDA 12.8 version (skip menu) + --rocm Install AMD ROCm version (skip menu) + --portable Use portable Python environment (Windows, skip prompt) + --no-portable Use standard virtual environment (Windows, skip prompt) + --verbose, -v Show detailed installation output + --help, -h Show this help message + +Requirements: + - Python 3.10 or later + - Internet connection for downloading dependencies +""" + +import argparse +import hashlib +import json +import os +import platform +import re +import shutil +import socket +import stat +import subprocess +import sys +import threading +import time +import urllib.request +import zipfile +from datetime import datetime +from pathlib import Path + +# ============================================================================ +# CONFIGURATION +# ============================================================================ + +# TESTING FLAG: Set to True to simulate Python 3.11+ on Windows +# (forces embedded Python fallback even if actual Python version is <3.11) +# This is useful for testing the embedded Python path without installing Python 3.11+ +TEST_EMBEDDED_PYTHON_PATH = False + +# Virtual environment settings +VENV_FOLDER = "venv" +SERVER_SCRIPT = "server.py" +CONFIG_FILE = "config.yaml" + +# Embedded Python settings (Windows fallback for Python 3.11+) +EMBEDDED_PYTHON_DIR = "python_embedded" +EMBEDDED_PYTHON_VERSION = "3.10.11" +EMBEDDED_PYTHON_URL = ( + f"https://www.python.org/ftp/python/{EMBEDDED_PYTHON_VERSION}/" + f"python-{EMBEDDED_PYTHON_VERSION}-embed-amd64.zip" +) +GET_PIP_URL = "https://bootstrap.pypa.io/get-pip.py" + +# SHA-256 hash of the embeddable zip for integrity verification. +# Set to "" to skip verification (not recommended for production). +# To compute: download the file from EMBEDDED_PYTHON_URL, then run: +# python -c "import hashlib; print(hashlib.sha256(open('python-3.10.11-embed-amd64.zip','rb').read()).hexdigest())" +EMBEDDED_PYTHON_SHA256 = "" + +# Installation type identifiers +INSTALL_CPU = "cpu" +INSTALL_NVIDIA = "nvidia" +INSTALL_NVIDIA_CU128 = "nvidia-cu128" +INSTALL_ROCM = "rocm" + +# Requirements file mapping +REQUIREMENTS_MAP = { + INSTALL_CPU: "requirements.txt", + INSTALL_NVIDIA: "requirements-nvidia.txt", + INSTALL_NVIDIA_CU128: "requirements-nvidia-cu128.txt", + INSTALL_ROCM: "requirements-rocm.txt", +} + +# Human-readable names for installation types +INSTALL_NAMES = { + INSTALL_CPU: "CPU Only", + INSTALL_NVIDIA: "NVIDIA GPU (CUDA 12.1)", + INSTALL_NVIDIA_CU128: "NVIDIA GPU (CUDA 12.8 / Blackwell)", + INSTALL_ROCM: "AMD GPU (ROCm 6.4)", +} + +# Chatterbox fork URL (used for CUDA 12.8 installation) +CHATTERBOX_REPO = "git+https://github.com/devnen/chatterbox-v2.git@master" + +# Timeout settings (seconds) +# First run downloads large model files (~2GB). Subsequent starts are much faster. +SERVER_STARTUP_TIMEOUT = 1800 +PORT_CHECK_INTERVAL = 0.5 + +# Global verbose mode flag (set from args) +VERBOSE_MODE = True + + +# ============================================================================ +# ANSI COLOR SUPPORT +# ============================================================================ + + +class Colors: + """ANSI color codes for cross-platform colored terminal output.""" + + CYAN = "\033[96m" + GREEN = "\033[92m" + YELLOW = "\033[93m" + RED = "\033[91m" + BOLD = "\033[1m" + DIM = "\033[2m" + RESET = "\033[0m" + + # Status icons + ICON_SUCCESS = "✓" + ICON_ERROR = "✗" + ICON_WARNING = "⚠" + ICON_INFO = "→" + ICON_WORKING = "●" + + @staticmethod + def is_windows(): + """Check if running on Windows.""" + return platform.system() == "Windows" + + @staticmethod + def is_linux(): + """Check if running on Linux.""" + return platform.system() == "Linux" + + @staticmethod + def is_macos(): + """Check if running on macOS.""" + return platform.system() == "Darwin" + + @classmethod + def enable_windows_colors(cls): + """Enable ANSI color support on Windows 10+.""" + if cls.is_windows(): + try: + import ctypes + + kernel32 = ctypes.windll.kernel32 + # Enable ANSI escape sequences on Windows 10+ + kernel32.SetConsoleMode(kernel32.GetStdHandle(-11), 7) + except Exception: + # If this fails, colors just won't work (non-fatal) + pass + + +# Enable Windows colors at module load time +Colors.enable_windows_colors() + + +# ============================================================================ +# PRINT HELPER FUNCTIONS +# ============================================================================ + + +def print_banner(): + """Print the startup banner.""" + print() + print("=" * 60) + print(" Chatterbox TTS Server - Launcher") + print("=" * 60) + print() + + +def print_header(text): + """Print a section header.""" + print(f"\n{Colors.CYAN}{text}{Colors.RESET}") + + +def print_step(step, total, message): + """Print a numbered step.""" + print(f"\n[{step}/{total}] {message}") + + +def print_substep(message, status="info"): + """ + Print a sub-step with status indicator. + + Args: + message: The message to print + status: One of "done", "error", "warning", "info" + """ + icons = { + "done": (Colors.GREEN, Colors.ICON_SUCCESS), + "error": (Colors.RED, Colors.ICON_ERROR), + "warning": (Colors.YELLOW, Colors.ICON_WARNING), + "info": (Colors.RESET, Colors.ICON_INFO), + } + + color, icon = icons.get(status, (Colors.RESET, Colors.ICON_INFO)) + print(f" {color}{icon}{Colors.RESET} {message}") + + +def print_success(text): + """Print a success message in green.""" + print(f"{Colors.GREEN}{text}{Colors.RESET}") + + +def print_warning(text): + """Print a warning message in yellow.""" + print(f"{Colors.YELLOW}{text}{Colors.RESET}") + + +def print_error(text): + """Print an error message in red.""" + print(f"{Colors.RED}{text}{Colors.RESET}") + + +def print_status_box(host, port): + """Print the final status box with server information.""" + display_host = "localhost" if host == "0.0.0.0" else host + url = f"http://{display_host}:{port}" + + print() + print("=" * 60) + print(f" {Colors.GREEN}🎙️ Chatterbox TTS Server is running!{Colors.RESET}") + print() + print(f" Web Interface: {url}") + print(f" API Docs: {url}/docs") + + if host == "0.0.0.0": + print() + print(" (Also accessible on your local network)") + + print() + print(" Press Ctrl+C to stop the server.") + print("=" * 60) + print() + + +def print_reinstall_hint(): + """Print a hint about how to reinstall.""" + print(f" {Colors.DIM}💡 Tip: To reinstall or upgrade, run:{Colors.RESET}") + print(f" {Colors.DIM} python start.py --reinstall{Colors.RESET}") + print() + + +# ============================================================================ +# COMMAND EXECUTION +# ============================================================================ + + +def run_command(cmd, cwd=None, check=True, capture=False, show_output=False): + """ + Run a shell command. + + Args: + cmd: Command string to execute + cwd: Working directory (optional) + check: If True, raise exception on non-zero exit + capture: If True, capture and return output + show_output: If True, show output in real-time + + Returns: + If capture=True: subprocess.CompletedProcess result + If capture=False: True on success, False on failure + """ + try: + if capture: + result = subprocess.run( + cmd, shell=True, cwd=cwd, capture_output=True, text=True, check=check + ) + return result + + if show_output or VERBOSE_MODE: + # Show output in real-time + result = subprocess.run(cmd, shell=True, cwd=cwd, check=check) + return result.returncode == 0 if not check else True + else: + # Suppress output + result = subprocess.run( + cmd, shell=True, cwd=cwd, capture_output=True, text=True, check=check + ) + return True + + except subprocess.CalledProcessError as e: + if check: + raise + return None if capture else False + except Exception as e: + if VERBOSE_MODE: + print_error(f"Command error: {e}") + return None if capture else False + + +def run_command_with_progress(cmd, cwd=None, description="Working"): + """ + Run a command with a progress indicator for long operations. + + Args: + cmd: Command string to execute + cwd: Working directory (optional) + description: Description to show during progress + + Returns: + True on success, False on failure + """ + if VERBOSE_MODE: + # In verbose mode, just show output directly + print_substep(f"Running: {cmd}", "info") + return run_command(cmd, cwd=cwd, show_output=True, check=False) + + # Start progress indicator in background + stop_progress = threading.Event() + + def progress_indicator(): + """Background thread to show progress spinner.""" + spinner = ["⠋", "⠙", "⠹", "⠸", "⠼", "⠴", "⠦", "⠧", "⠇", "⠏"] + idx = 0 + while not stop_progress.is_set(): + sys.stdout.write(f"\r {spinner[idx]} {description}...") + sys.stdout.flush() + idx = (idx + 1) % len(spinner) + time.sleep(0.1) + # Clear the progress line + sys.stdout.write("\r" + " " * 60 + "\r") + sys.stdout.flush() + + progress_thread = threading.Thread(target=progress_indicator, daemon=True) + progress_thread.start() + + try: + result = subprocess.run( + cmd, shell=True, cwd=cwd, capture_output=True, text=True + ) + + stop_progress.set() + progress_thread.join(timeout=1) + + if result.returncode != 0: + print_substep(f"Command failed with exit code {result.returncode}", "error") + if result.stderr: + # Show last part of error message + error_lines = result.stderr.strip().split("\n") + for line in error_lines[-5:]: + print(f" {line}") + return False + + return True + + except Exception as e: + stop_progress.set() + progress_thread.join(timeout=1) + print_error(f"Error running command: {e}") + return False + + +# ============================================================================ +# PLATFORM DETECTION +# ============================================================================ + + +def is_windows(): + """Check if running on Windows.""" + return platform.system() == "Windows" + + +def is_linux(): + """Check if running on Linux.""" + return platform.system() == "Linux" + + +def is_macos(): + """Check if running on macOS.""" + return platform.system() == "Darwin" + + +def get_platform_name(): + """Get human-readable platform name.""" + system = platform.system() + if system == "Windows": + return "Windows" + elif system == "Linux": + return "Linux" + elif system == "Darwin": + return "macOS" + else: + return system + + +# ============================================================================ +# PYTHON & VIRTUAL ENVIRONMENT FUNCTIONS +# ============================================================================ + + +def check_python_version(): + """ + Verify Python version is 3.10 or later. + Exits with error if version is too old. + """ + major = sys.version_info.major + minor = sys.version_info.minor + + if major < 3 or (major == 3 and minor < 10): + print_error(f"Python 3.10+ required, but found Python {major}.{minor}") + print() + print("Please install Python 3.10 or newer from:") + print(" https://www.python.org/downloads/") + print() + sys.exit(1) + + print_substep(f"Python {major}.{minor}.{sys.version_info.micro} detected", "done") + + +def get_venv_paths(root_dir): + """ + Get paths for virtual environment components. + + Args: + root_dir: Root directory of the project + + Returns: + Tuple of (venv_dir, venv_python, venv_pip) as Path objects + """ + venv_dir = root_dir / VENV_FOLDER + + if is_windows(): + venv_python = venv_dir / "Scripts" / "python.exe" + venv_pip = venv_dir / "Scripts" / "pip.exe" + else: + venv_python = venv_dir / "bin" / "python" + venv_pip = venv_dir / "bin" / "pip" + + return venv_dir, venv_python, venv_pip + + +def create_venv(venv_dir): + """ + Create a virtual environment. + + Args: + venv_dir: Path to create the virtual environment + + Returns: + True on success, False on failure + """ + print_substep("Creating virtual environment...") + + try: + result = subprocess.run( + [sys.executable, "-m", "venv", str(venv_dir)], + capture_output=True, + text=True, + ) + + if result.returncode != 0: + print_substep("Failed to create virtual environment", "error") + if result.stderr: + print(f" {result.stderr[:200]}") + return False + + print_substep("Virtual environment created", "done") + return True + + except Exception as e: + print_substep(f"Error creating venv: {e}", "error") + return False + + +def get_install_state(venv_dir): + """ + Check if installation is complete and get the install type. + + Args: + venv_dir: Path to virtual environment directory + + Returns: + Tuple of (is_installed: bool, install_type: str or None) + """ + install_complete_file = venv_dir / ".install_complete" + install_type_file = venv_dir / ".install_type" + + if not install_complete_file.exists(): + return False, None + + install_type = None + if install_type_file.exists(): + try: + install_type = install_type_file.read_text(encoding="utf-8").strip() + except Exception: + pass + + return True, install_type + + +def save_install_state(venv_dir, install_type): + """ + Save installation state files. + + Args: + venv_dir: Path to virtual environment directory + install_type: Type of installation (cpu, nvidia, nvidia-cu128, rocm) + """ + try: + # Save install type + install_type_file = venv_dir / ".install_type" + install_type_file.write_text(install_type, encoding="utf-8") + + # Save completion marker with timestamp + install_complete_file = venv_dir / ".install_complete" + timestamp = datetime.now().isoformat() + install_complete_file.write_text( + f"Installation completed at {timestamp}\n" f"Type: {install_type}\n", + encoding="utf-8", + ) + except Exception as e: + print_warning(f"Could not save install state: {e}") + + +def clear_install_complete(venv_dir): + """ + Clear only the install complete marker (for upgrades). + + Args: + venv_dir: Path to virtual environment directory + """ + install_complete_file = venv_dir / ".install_complete" + + try: + if install_complete_file.exists(): + install_complete_file.unlink() + except Exception as e: + print_warning(f"Could not clear install marker: {e}") + + +def robust_rmtree(path): + """ + Remove a directory tree with Windows-hardened error handling. + + Uses an onerror callback to strip read-only attributes (common on + extracted zip contents), retries on transient permission errors + (antivirus, Explorer indexing), and falls back to renaming the + directory aside if deletion fails entirely. + + Args: + path: Path to directory to remove + + Returns: + True if directory is gone (deleted or renamed aside), False if stuck + """ + path = Path(path) + if not path.exists(): + return True + + def handle_remove_readonly(func, fpath, exc): + """Clear read-only flag and retry the failed operation.""" + os.chmod(fpath, stat.S_IWRITE) + func(fpath) + + max_retries = 3 + retry_delay = 2 + + for attempt in range(max_retries): + try: + shutil.rmtree(path, onerror=handle_remove_readonly) + return True + except PermissionError: + if attempt < max_retries - 1: + print_substep( + f"Files locked, retrying in {retry_delay}s... " + f"(attempt {attempt + 1}/{max_retries})", + "warning", + ) + time.sleep(retry_delay) + except Exception: + break # Non-permission error, skip to rename fallback + + # Fallback: rename aside so we can proceed even if deletion fails + try: + aside_name = f"{path.name}.old.{int(time.time())}" + aside_path = path.parent / aside_name + path.rename(aside_path) + print_substep( + f"Could not delete folder; renamed to {aside_name}", + "warning", + ) + print_substep("You can safely delete that folder later.", "info") + return True + except Exception: + pass + + return False + + +def remove_venv(venv_dir): + """ + Remove an environment directory (venv or embedded) with robust error handling. + + Args: + venv_dir: Path to environment directory + + Returns: + True on success, False on failure + """ + if not venv_dir.exists(): + return True + + print_substep(f"Removing existing environment ({venv_dir.name})...") + + if robust_rmtree(venv_dir): + print_substep("Environment removed", "done") + return True + + print_error(f"Could not remove: {venv_dir}") + print_substep( + "Try closing any terminals, editors, or antivirus scanning this folder", + "info", + ) + if is_windows(): + print_substep(f'Or run: rmdir /s /q "{venv_dir.name}"', "info") + else: + print_substep(f'Or run: rm -rf "{venv_dir.name}"', "info") + return False + + +# ============================================================================ +# EMBEDDED PYTHON (WINDOWS FALLBACK) +# ============================================================================ + + +def get_embedded_python_paths(root_dir): + """ + Get paths for the embedded Python environment (Windows only). + + Args: + root_dir: Root directory of the project + + Returns: + Tuple of (embedded_dir, embedded_python, embedded_pip) as Path objects + """ + embedded_dir = root_dir / EMBEDDED_PYTHON_DIR + embedded_python = embedded_dir / "python.exe" + embedded_pip = embedded_dir / "Scripts" / "pip.exe" + return embedded_dir, embedded_python, embedded_pip + + +def is_embedded_python_available(root_dir): + """ + Check if embedded Python is already set up and functional. + + Args: + root_dir: Root directory of the project + + Returns: + True if embedded Python is ready to use + """ + embedded_dir, embedded_python, embedded_pip = get_embedded_python_paths(root_dir) + + if not embedded_python.exists() or not embedded_pip.exists(): + return False + + try: + result = subprocess.run( + [str(embedded_python), "--version"], + capture_output=True, + text=True, + timeout=10, + ) + return result.returncode == 0 + except Exception: + return False + + +def download_file(url, dest_path, description="Downloading"): + """ + Download a file from a URL with progress indication. + + Uses urlopen with an explicit per-operation timeout to prevent + indefinite hanging on flaky networks or corporate proxies. + Downloads to a temporary .part file first, then atomically moves + to dest_path so interrupted downloads never leave a valid-looking + but truncated file at the final path. + + Args: + url: URL to download from + dest_path: Local path to save the file + description: Description shown during download + + Returns: + True on success, False on failure + """ + print_substep(f"{description}...") + + dest_path = Path(dest_path) + part_path = dest_path.parent / (dest_path.name + ".part") + + try: + response = urllib.request.urlopen(url, timeout=30) + total_size = int(response.headers.get("Content-Length", 0)) + downloaded = 0 + last_percent = -1 + + with open(part_path, "wb") as f: + while True: + chunk = response.read(8192) + if not chunk: + break + f.write(chunk) + downloaded += len(chunk) + + if total_size > 0: + percent = min(100, int(downloaded * 100 / total_size)) + if percent != last_percent and percent % 5 == 0: + last_percent = percent + mb_done = downloaded / (1024 * 1024) + mb_total = total_size / (1024 * 1024) + sys.stdout.write( + f"\r {Colors.ICON_WORKING} {description}... " + f"{percent}% ({mb_done:.1f}/{mb_total:.1f} MB)" + ) + sys.stdout.flush() + else: + # No Content-Length: show bytes downloaded without percentage + mb_done = downloaded / (1024 * 1024) + if int(mb_done * 10) != last_percent: + last_percent = int(mb_done * 10) + sys.stdout.write( + f"\r {Colors.ICON_WORKING} {description}... " + f"{mb_done:.1f} MB" + ) + sys.stdout.flush() + + sys.stdout.write("\n") + sys.stdout.flush() + + # Atomic move: .part → final path + os.replace(str(part_path), str(dest_path)) + print_substep(f"{description} complete", "done") + return True + + except Exception as e: + sys.stdout.write("\n") + sys.stdout.flush() + print_substep(f"Download failed: {e}", "error") + print_substep(f"You can download manually from: {url}", "info") + return False + + finally: + # Clean up partial download on failure (no-op on success since + # os.replace already moved the .part file to dest_path) + try: + if part_path.exists(): + part_path.unlink() + except Exception: + pass + + +def verify_checksum(file_path, expected_sha256): + """ + Verify SHA-256 hash of a downloaded file. + + Args: + file_path: Path to the file to verify + expected_sha256: Expected hex digest string + + Returns: + True if hash matches, False otherwise + """ + sha256 = hashlib.sha256() + with open(file_path, "rb") as f: + for chunk in iter(lambda: f.read(8192), b""): + sha256.update(chunk) + actual = sha256.hexdigest() + if actual != expected_sha256: + print_substep("Checksum mismatch!", "error") + print_substep(f"Expected: {expected_sha256}", "info") + print_substep(f"Actual: {actual}", "info") + return False + return True + + +def patch_pth_file(embedded_dir): + """ + Patch the python3XX._pth file to enable pip and package imports. + + The embeddable Python distribution ships with a ._pth file that + restricts the import path. We need to uncomment 'import site' and + add 'Lib\\site-packages' so that pip-installed packages are importable. + + Note: pip usage with the embeddable distribution is "not supported" + per CPython docs, but works reliably with this patching approach. + The ._pth format has been stable since Python 3.5. Re-test if + bumping EMBEDDED_PYTHON_VERSION to a new minor release. + + Args: + embedded_dir: Path to the embedded Python directory + + Returns: + True on success, False on failure + """ + try: + # Find the ._pth file (e.g., python310._pth) + pth_files = list(embedded_dir.glob("python3*._pth")) + + if not pth_files: + print_substep("Could not find ._pth file in embedded Python", "error") + return False + + pth_file = pth_files[0] + content = pth_file.read_text(encoding="utf-8") + lines = content.splitlines() + + # Collect path entries, skipping comments, blanks, and the + # import site directive (which we'll re-add at the end in + # the canonical position: paths first, import site last). + path_lines = [] + has_site_packages = False + + for line in lines: + stripped = line.strip() + + # Skip import site (commented or not) — added back at the end + if stripped in ("#import site", "import site"): + continue + + # Skip blank lines and the stock comment + if not stripped or stripped.startswith("#"): + continue + + path_lines.append(stripped) + if "site-packages" in stripped: + has_site_packages = True + + # Add site-packages path if not already present + if not has_site_packages: + path_lines.append("Lib\\site-packages") + + # Add parent directory (project root) so that project modules + # like config.py, engine.py, utils.py are importable. + # The embedded Python dir is always /python_embedded/, + # so ".." resolves to the project root at runtime. + if ".." not in path_lines: + path_lines.append("..") + + # Canonical order: paths first, then import site last + path_lines.append("import site") + + pth_file.write_text("\n".join(path_lines) + "\n", encoding="utf-8") + + if VERBOSE_MODE: + print_substep(f"Patched {pth_file.name}", "done") + + return True + + except Exception as e: + print_substep(f"Failed to patch ._pth file: {e}", "error") + return False + + +def _create_dll_search_sitecustomize(embedded_dir): + """ + Create a sitecustomize.py in the embedded Python directory that configures + Windows DLL search paths at interpreter startup. + + This ensures native extensions (.pyd files) can find their dependent DLLs + regardless of how the embedded Python is launched (via start.py, manually, + or from a subprocess). The file is automatically loaded by site.py + (triggered by 'import site' in the ._pth file). + + No-op on non-Windows platforms. + + Args: + embedded_dir: Path to the embedded Python directory + """ + sitecustomize_path = Path(embedded_dir) / "sitecustomize.py" + + content = """\ +# Auto-generated by start.py -- DO NOT EDIT +# Configures DLL search paths for the embedded Python environment on Windows. +# This ensures native extensions (.pyd) can locate their dependent DLLs. +import os +import sys + +if sys.platform == "win32" and hasattr(os, "add_dll_directory"): + _exe_dir = os.path.dirname(sys.executable) + _sp_dir = os.path.join(_exe_dir, "Lib", "site-packages") + + for _d in [_exe_dir, _sp_dir]: + if os.path.isdir(_d): + try: + os.add_dll_directory(_d) + except OSError: + pass + + # Add *.libs directories (standard wheel pattern for vendored native DLLs, + # created by auditwheel/delvewheel when packaging binary extensions) + if os.path.isdir(_sp_dir): + for _entry in os.listdir(_sp_dir): + if _entry.endswith(".libs"): + _libs_path = os.path.join(_sp_dir, _entry) + if os.path.isdir(_libs_path): + try: + os.add_dll_directory(_libs_path) + except OSError: + pass +""" + + try: + sitecustomize_path.write_text(content, encoding="utf-8") + if VERBOSE_MODE: + print_substep("Created sitecustomize.py for DLL search paths", "done") + except Exception as e: + print_substep(f"Could not create sitecustomize.py: {e}", "warning") + + +def setup_embedded_python(root_dir): + """ + Download and configure an embedded Python 3.10 environment for Windows. + + This creates a fully self-contained Python installation inside the + project folder with pip bootstrapped and ready to install packages. + + Args: + root_dir: Root directory of the project + + Returns: + True on success, False on failure + """ + embedded_dir = root_dir / EMBEDDED_PYTHON_DIR + + # Check if already available + if is_embedded_python_available(root_dir): + print_substep( + f"Embedded Python {EMBEDDED_PYTHON_VERSION} already set up", "done" + ) + return True + + # Clean up any partial previous attempt + if embedded_dir.exists(): + if not robust_rmtree(embedded_dir): + print_substep("Could not clean up partial install", "error") + return False + + print_substep( + f"Setting up portable Python {EMBEDDED_PYTHON_VERSION} environment...", "info" + ) + + zip_path = root_dir / "_python_embedded.zip" + get_pip_path = root_dir / "_get-pip.py" + + try: + # Step 1: Download embeddable Python + if not download_file( + EMBEDDED_PYTHON_URL, + zip_path, + f"Downloading Python {EMBEDDED_PYTHON_VERSION} embeddable package", + ): + return False + + # Step 1b: Verify download integrity + if EMBEDDED_PYTHON_SHA256: + if not verify_checksum(zip_path, EMBEDDED_PYTHON_SHA256): + print_substep( + "Downloaded file may be corrupted. " + "Delete it and try again, or download manually.", + "error", + ) + return False + if VERBOSE_MODE: + print_substep("Checksum verified", "done") + + # Step 1c: Validate zip archive + if not zipfile.is_zipfile(str(zip_path)): + print_substep("Downloaded file is not a valid zip archive", "error") + print_substep("Your network may be returning an error page instead", "info") + return False + + # Step 2: Extract + print_substep("Extracting Python...") + try: + embedded_dir.mkdir(parents=True, exist_ok=True) + with zipfile.ZipFile(str(zip_path), "r") as zf: + zf.extractall(str(embedded_dir)) + print_substep("Python extracted", "done") + except Exception as e: + print_substep(f"Extraction failed: {e}", "error") + return False + + # Step 3: Patch ._pth file for pip and site-packages support + if not patch_pth_file(embedded_dir): + return False + + # Step 3b: Create sitecustomize.py for DLL search path configuration + _create_dll_search_sitecustomize(embedded_dir) + + # Step 4: Bootstrap pip + if not download_file(GET_PIP_URL, get_pip_path, "Downloading pip installer"): + return False + + embedded_python = embedded_dir / "python.exe" + print_substep("Installing pip...") + + pip_cmd = [str(embedded_python), str(get_pip_path)] + if VERBOSE_MODE: + result = subprocess.run(pip_cmd) + else: + result = subprocess.run(pip_cmd, capture_output=True, text=True) + + if result.returncode != 0: + print_substep("Failed to install pip", "error") + if not VERBOSE_MODE and hasattr(result, "stderr") and result.stderr: + error_lines = result.stderr.strip().split("\n")[-3:] + for line in error_lines: + print(f" {line}") + return False + + # Step 5: Verify pip is usable + embedded_pip = embedded_dir / "Scripts" / "pip.exe" + if not embedded_pip.exists(): + print_substep("pip was not created at expected location", "error") + return False + + print_substep("pip installed", "done") + + # Step 6: Install setuptools (provides pkg_resources, needed by perth and others) + # Modern get-pip.py no longer bundles setuptools, but many ML/AI packages + # (including resemble-perth) still import pkg_resources at runtime. + # Note: pkg_resources was removed in setuptools 81+ (targets Python 3.12+). + # On Python 3.10 pip resolves to a compatible older version automatically. + print_substep("Installing setuptools...") + setuptools_cmd = [ + str(embedded_python), + "-m", + "pip", + "install", + "--no-warn-script-location", + "setuptools", + ] + if VERBOSE_MODE: + st_result = subprocess.run(setuptools_cmd) + else: + st_result = subprocess.run(setuptools_cmd, capture_output=True, text=True) + + if st_result.returncode != 0: + print_substep( + "setuptools installation failed (pkg_resources may be unavailable)", + "warning", + ) + else: + print_substep("setuptools installed", "done") + + print() + print_substep( + f"Portable Python {EMBEDDED_PYTHON_VERSION} environment ready", "done" + ) + return True + + except Exception as e: + print_substep(f"Unexpected error during setup: {e}", "error") + return False + + finally: + # Clean up temporary downloads + for temp_file in [zip_path, get_pip_path]: + try: + if temp_file.exists(): + temp_file.unlink() + except Exception: + pass + + +def prompt_portable_install(reason="portability"): + """ + Offer Windows users the portable Python installation option. + + Args: + reason: "compatibility" (Python 3.11+) or "portability" (Python 3.10) + + Returns: + True if user wants portable mode, False for standard venv + """ + major = sys.version_info.major + minor = sys.version_info.minor + + print() + + if reason == "compatibility": + # Python 3.11+ — lead with compatibility problem, mention portability + print(f"{Colors.YELLOW}{'=' * 60}{Colors.RESET}") + print( + f" {Colors.YELLOW}{Colors.ICON_WARNING} Python {major}.{minor} detected" + f" — Portable Mode recommended{Colors.RESET}" + ) + print(f"{Colors.YELLOW}{'=' * 60}{Colors.RESET}") + print() + print(" On Windows, Python 3.11+ lacks pre-built binary packages") + print(" (wheels) for several key dependencies, including ONNX and") + print(" ONNXRuntime. This causes installation failures that are") + print(" difficult to resolve.") + print() + print( + f" {Colors.GREEN}Portable Mode solves this automatically{Colors.RESET}" + f" by using a compatible" + ) + print(" Python runtime, and as a bonus makes your entire installation") + print(" fully portable — copy it to a USB drive, share it as a zip") + print(" file, or move it anywhere.") + else: + # Python 3.10 — lead with portability benefits + print(f"{Colors.CYAN}{'=' * 60}{Colors.RESET}") + print(f" {Colors.CYAN}📦 Portable Mode Available{Colors.RESET}") + print(f"{Colors.CYAN}{'=' * 60}{Colors.RESET}") + print() + print(" The launcher can create a fully self-contained installation.") + print(" The entire project folder — including Python and all") + print(" dependencies — becomes completely portable:") + print() + print(" • Copy to a USB drive and run on another PC") + print(" • Zip the folder and share it with others") + print(" • Move it anywhere on your filesystem") + print(" • No Python installation needed on the target machine") + + print() + + # Option 1 — Portable (default) + print(f" {Colors.BOLD}[1] Install in Portable Mode (recommended){Colors.RESET}") + print(" Creates a self-contained Python environment inside this folder.") + print( + f" Your system Python {major}.{minor} remains" + f" {Colors.GREEN}completely untouched{Colors.RESET}." + ) + print() + + # Option 2 — Standard venv + print(" [2] Standard installation") + if reason == "compatibility": + print( + f" {Colors.DIM}Uses system Python {major}.{minor}" + f" with a virtual environment.{Colors.RESET}" + ) + print( + f" {Colors.DIM}May require Visual C++ Build Tools." + f" Not portable.{Colors.RESET}" + ) + else: + print( + f" {Colors.DIM}Uses a standard virtual environment." + f" Works fine but not portable.{Colors.RESET}" + ) + print() + + while True: + try: + choice = input(" Enter choice [1]: ").strip() + + if choice in ("", "1"): + return True + elif choice == "2": + print() + if reason == "compatibility": + print_substep( + f"Continuing with system Python {major}.{minor}", "warning" + ) + print_substep( + "If installation fails, re-run with:" + " python start.py --reinstall --portable", + "info", + ) + else: + print_substep("Using standard virtual environment", "info") + return False + else: + print_warning(f" Invalid choice '{choice}'. Please enter 1 or 2.") + print() + + except (EOFError, KeyboardInterrupt): + print() + print(" Aborted by user.") + sys.exit(2) + + +# ============================================================================ +# GPU DETECTION +# ============================================================================ + + +def detect_nvidia_gpu(): + """ + Detect NVIDIA GPU using nvidia-smi. + + Returns: + Tuple of (found: bool, gpu_name: str or None) + """ + try: + result = subprocess.run( + ["nvidia-smi", "--query-gpu=name", "--format=csv,noheader,nounits"], + capture_output=True, + text=True, + timeout=10, + ) + + if result.returncode == 0 and result.stdout.strip(): + gpu_name = result.stdout.strip().split("\n")[0] + return True, gpu_name + + return False, None + + except FileNotFoundError: + # nvidia-smi not found + return False, None + except subprocess.TimeoutExpired: + return False, None + except Exception: + return False, None + + +def detect_amd_gpu(): + """ + Detect AMD GPU using rocm-smi. + + Returns: + Tuple of (found: bool, gpu_name: str or None) + """ + try: + result = subprocess.run( + ["rocm-smi", "--showproductname"], + capture_output=True, + text=True, + timeout=10, + ) + + if result.returncode == 0 and result.stdout.strip(): + # Parse output to find GPU name + lines = result.stdout.strip().split("\n") + for line in lines: + if "Card series" in line or "GPU" in line: + # Extract the name part + parts = line.split(":") + if len(parts) > 1: + return True, parts[1].strip() + + # If we got output but couldn't parse name, still report found + return True, "AMD GPU (unknown model)" + + return False, None + + except FileNotFoundError: + # rocm-smi not found + return False, None + except subprocess.TimeoutExpired: + return False, None + except Exception: + return False, None + + +def detect_gpu(): + """ + Detect available GPUs. + + Returns: + Dictionary with detection results: + { + "nvidia": bool, + "nvidia_name": str or None, + "amd": bool, + "amd_name": str or None, + } + """ + nvidia_found, nvidia_name = detect_nvidia_gpu() + amd_found, amd_name = detect_amd_gpu() + + return { + "nvidia": nvidia_found, + "nvidia_name": nvidia_name, + "amd": amd_found, + "amd_name": amd_name, + } + + +# ============================================================================ +# INSTALLATION MENU +# ============================================================================ + + +def get_default_choice(gpu_info): + """ + Determine the default installation choice based on detected hardware. + + Args: + gpu_info: Dictionary from detect_gpu() + + Returns: + Installation type string (INSTALL_CPU, INSTALL_NVIDIA, etc.) + """ + if gpu_info["nvidia"]: + return INSTALL_NVIDIA + elif gpu_info["amd"] and is_linux(): + return INSTALL_ROCM + else: + return INSTALL_CPU + + +def show_installation_menu(gpu_info, default_choice): + """ + Display installation menu and get user choice. + + Args: + gpu_info: Dictionary from detect_gpu() + default_choice: Default installation type + + Returns: + Selected installation type string + """ + # Map install types to menu numbers + MENU_MAP = { + "1": INSTALL_CPU, + "2": INSTALL_NVIDIA, + "3": INSTALL_NVIDIA_CU128, + "4": INSTALL_ROCM, + } + + # Reverse map for showing default + REVERSE_MAP = {v: k for k, v in MENU_MAP.items()} + default_num = REVERSE_MAP[default_choice] + + # Print GPU detection results + print() + print("=" * 60) + print(" Hardware Detection") + print("=" * 60) + print() + + if gpu_info["nvidia"]: + print_success(f" NVIDIA GPU: Detected ({gpu_info['nvidia_name']})") + else: + print(f" NVIDIA GPU: {Colors.DIM}Not detected{Colors.RESET}") + + if gpu_info["amd"]: + print_success(f" AMD GPU: Detected ({gpu_info['amd_name']})") + else: + print(f" AMD GPU: {Colors.DIM}Not detected{Colors.RESET}") + + # Print menu + print() + print("=" * 60) + print(" Select Installation Type") + print("=" * 60) + print() + + # Menu options with descriptions + options = [ + ("1", "CPU Only", "No GPU acceleration - works on any system"), + ("2", "NVIDIA GPU (CUDA 12.1)", "Standard for RTX 20/30/40 series"), + ("3", "NVIDIA GPU (CUDA 12.8)", "For RTX 5090 / Blackwell GPUs only"), + ("4", "AMD GPU (ROCm 6.4)", "For AMD GPUs on Linux"), + ] + + for num, name, desc in options: + # Determine if this is the default + is_default = num == default_num + + # Check for special warnings + warning = "" + if num == "4" and is_windows(): + warning = f" {Colors.YELLOW}⚠️ Not supported on Windows{Colors.RESET}" + + # Build the option line + default_marker = f" {Colors.GREEN}[DEFAULT]{Colors.RESET}" if is_default else "" + + print(f" [{num}] {name}{default_marker}") + print(f" {Colors.DIM}{desc}{warning}{Colors.RESET}") + print() + + # Get user input + while True: + try: + prompt = f" Enter choice [{default_num}]: " + choice = input(prompt).strip() + + # Empty input = default + if not choice: + return default_choice + + # Validate input + if choice in MENU_MAP: + return MENU_MAP[choice] + + print_warning(f" Invalid choice '{choice}'. Please enter 1, 2, 3, or 4.") + print() + + except (EOFError, KeyboardInterrupt): + print() + print(" Aborted by user.") + sys.exit(2) + + +# ============================================================================ +# INSTALLATION FUNCTIONS +# ============================================================================ + + +def upgrade_pip(venv_python): + """ + Upgrade pip in the environment. + + Args: + venv_python: Path to the Python executable (venv or embedded) + """ + print_substep("Upgrading pip...") + + cmd = f'"{venv_python}" -m pip install --upgrade pip' + + # We force check=True here because having an old pip causes the + # dependency resolution errors you are seeing + try: + if VERBOSE_MODE: + subprocess.check_call(cmd, shell=True) + else: + subprocess.check_call( + cmd, shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL + ) + + print_substep("pip upgraded", "done") + return True + except subprocess.CalledProcessError: + print_substep("pip upgrade failed", "warning") + return False + + +def install_requirements(venv_pip, requirements_file, root_dir): + """ + Install dependencies from a requirements file. + + Args: + venv_pip: Path to pip executable in venv + requirements_file: Name of requirements file + root_dir: Root directory of the project + + Returns: + True on success, False on failure + """ + requirements_path = root_dir / requirements_file + + if not requirements_path.exists(): + print_error(f"Requirements file not found: {requirements_file}") + return False + + print_substep(f"Installing from {requirements_file}...") + + # Suppress pip warnings about scripts not on PATH (common with embedded Python) + cmd = f'"{venv_pip}" install --no-warn-script-location -r "{requirements_path}"' + + success = run_command_with_progress( + cmd, + cwd=str(root_dir), + description=f"Installing dependencies from {requirements_file}", + ) + + if success: + print_substep("Dependencies installed", "done") + else: + print_substep("Dependency installation failed", "error") + + return success + + +def install_chatterbox_no_deps(venv_pip): + """ + Install Chatterbox TTS without dependencies (for CUDA 12.8). + + Args: + venv_pip: Path to pip executable in venv + + Returns: + True on success, False on failure + """ + print_substep("Installing Chatterbox TTS (--no-deps to preserve PyTorch 2.8)...") + + cmd = f'"{venv_pip}" install --no-deps {CHATTERBOX_REPO}' + + success = run_command_with_progress(cmd, description="Installing Chatterbox TTS") + + if success: + print_substep("Chatterbox TTS installed", "done") + else: + print_substep("Chatterbox TTS installation failed", "error") + + return success + + +def perform_installation(venv_pip, install_type, root_dir): + """ + Perform installation based on selected type. + + Args: + venv_pip: Path to pip executable in venv + install_type: One of INSTALL_CPU, INSTALL_NVIDIA, INSTALL_NVIDIA_CU128, INSTALL_ROCM + root_dir: Root directory of the project + + Returns: + True on success, False on failure + """ + requirements_file = REQUIREMENTS_MAP.get(install_type) + + if not requirements_file: + print_error(f"Unknown installation type: {install_type}") + return False + + # Step 1: Install requirements + if not install_requirements(venv_pip, requirements_file, root_dir): + return False + + # Step 2: For CUDA 12.8, install chatterbox separately with --no-deps + if install_type == INSTALL_NVIDIA_CU128: + if not install_chatterbox_no_deps(venv_pip): + return False + + return True + + +def _patch_chatterbox_watermarker(env_dir, use_embedded): + """ + Patch installed chatterbox source files to make the Perth watermarker + gracefully optional. If perth fails to load or PerthImplicitWatermarker + is None, the server will skip watermarking instead of crashing. + + Uses a no-op watermarker class so that all call sites (apply_watermark) + continue to work without modification — they just pass audio through + unchanged. + + This patch is idempotent: re-running it on already-patched files is safe. + + Args: + env_dir: Path to environment directory (venv or python_embedded) + use_embedded: True if using embedded Python environment + """ + # Locate site-packages (differs between embedded, Windows venv, Linux/macOS venv) + if use_embedded or is_windows(): + site_packages = env_dir / "Lib" / "site-packages" + else: + # Linux/macOS venv: lib/python3.X/site-packages + lib_dir = env_dir / "lib" + site_packages = None + if lib_dir.exists(): + for child in sorted(lib_dir.iterdir()): + if child.name.startswith("python3") and child.is_dir(): + candidate = child / "site-packages" + if candidate.is_dir(): + site_packages = candidate + break + if site_packages is None: + print_substep( + "Could not locate site-packages, skipping watermarker patch", + "warning", + ) + return + + # Find chatterbox package directory (name varies by package version) + chatterbox_dir = None + for name in ["chatterbox", "chatterbox_tts"]: + candidate = site_packages / name + if candidate.is_dir(): + chatterbox_dir = candidate + break + + if chatterbox_dir is None: + if VERBOSE_MODE: + print_substep( + "Chatterbox package not found, skipping watermarker patch", "info" + ) + return + + SENTINEL = "# [patched by start.py: watermarker made optional]" + INIT_TARGET = "self.watermarker = perth.PerthImplicitWatermarker()" + target_files = ["tts.py", "tts_turbo.py", "mtl_tts.py", "vc.py"] + patched_count = 0 + + for filename in target_files: + filepath = chatterbox_dir / filename + if not filepath.exists(): + continue + + try: + content = filepath.read_text(encoding="utf-8") + except Exception as e: + print_substep(f"{filename}: could not read ({e}), skipping", "warning") + continue + + # Idempotency: skip if already patched + if SENTINEL in content: + if VERBOSE_MODE: + print_substep(f"{filename}: already patched", "info") + continue + + if INIT_TARGET not in content: + if VERBOSE_MODE: + print_substep(f"{filename}: target pattern not found, skipping", "info") + continue + + # Determine whether this file uses the logging module + has_logger = "import logging" in content or "getLogger" in content + if has_logger: + log_line = ( + "logger.warning(" + '"Perth watermarker unavailable ' + '\\u2014 audio will not be watermarked")' + ) + else: + log_line = ( + "print(" + '"Warning: Perth watermarker unavailable ' + '\\u2014 audio will not be watermarked")' + ) + + # Build the replacement block + lines = content.split("\n") + new_lines = [] + + for line in lines: + if INIT_TARGET in line and line.lstrip().startswith("self."): + indent = line[: len(line) - len(line.lstrip())] + new_lines.append(f"{indent}{SENTINEL}") + new_lines.append(f"{indent}try:") + new_lines.append( + f"{indent} self.watermarker = perth.PerthImplicitWatermarker()" + ) + new_lines.append(f"{indent}except Exception:") + new_lines.append(f"{indent} class _NoOpWatermarker:") + new_lines.append( + f"{indent} def apply_watermark(self, wav, *args, **kwargs):" + ) + new_lines.append(f"{indent} return wav") + new_lines.append(f"{indent} self.watermarker = _NoOpWatermarker()") + new_lines.append(f"{indent} {log_line}") + else: + new_lines.append(line) + + try: + filepath.write_text("\n".join(new_lines), encoding="utf-8") + print_substep(f"{filename}: watermarker made optional", "done") + patched_count += 1 + except Exception as e: + print_substep(f"{filename}: could not write ({e})", "warning") + + if patched_count > 0: + print_substep( + f"Patched {patched_count} file(s) for optional watermarking", "done" + ) + elif VERBOSE_MODE: + print_substep("No files needed watermarker patching", "info") + + +def verify_installation(venv_python): + """ + Verify critical dependencies are installed correctly. + + Args: + venv_python: Path to Python executable in venv + + Returns: + True if verification passed, False otherwise + """ + print_substep("Verifying installation...") + + # Python script to run inside the venv to test imports + test_script = """ +import sys +import json + +results = {} + +# Test PyTorch +try: + import torch + results["torch"] = { + "ok": True, + "version": torch.__version__, + "cuda_available": torch.cuda.is_available(), + "cuda_version": torch.version.cuda if torch.cuda.is_available() else None, + "gpu_count": torch.cuda.device_count() if torch.cuda.is_available() else 0, + "gpu_name": torch.cuda.get_device_name(0) if torch.cuda.is_available() and torch.cuda.device_count() > 0 else None, + } +except Exception as e: + results["torch"] = {"ok": False, "error": str(e)} + +# Test FastAPI +try: + import fastapi + results["fastapi"] = {"ok": True, "version": fastapi.__version__} +except Exception as e: + results["fastapi"] = {"ok": False, "error": str(e)} + +# Test Chatterbox +try: + # Try different import paths + try: + import chatterbox + results["chatterbox"] = {"ok": True} + except ImportError: + from chatterbox_tts import ChatterboxTTS + results["chatterbox"] = {"ok": True} +except Exception as e: + results["chatterbox"] = {"ok": False, "error": str(e)} + +# Test audio libraries +try: + import soundfile + import librosa + results["audio"] = {"ok": True} +except Exception as e: + results["audio"] = {"ok": False, "error": str(e)} + +print(json.dumps(results)) +""" + + try: + result = subprocess.run( + [str(venv_python), "-c", test_script], + capture_output=True, + text=True, + timeout=60, + ) + + if result.returncode != 0: + print_substep("Verification script returned non-zero", "warning") + if result.stderr: + # Show relevant error info + error_lines = result.stderr.strip().split("\n")[-3:] + for line in error_lines: + print(f" {line}") + return False + + # Parse JSON results + try: + results = json.loads(result.stdout) + except json.JSONDecodeError: + print_substep("Could not parse verification results", "warning") + return False + + all_ok = True + + # Report PyTorch status + torch_result = results.get("torch", {}) + if torch_result.get("ok"): + version_str = torch_result.get("version", "unknown") + + if torch_result.get("cuda_available"): + cuda_ver = torch_result.get("cuda_version", "unknown") + gpu_name = torch_result.get("gpu_name", "unknown") + print_substep(f"PyTorch {version_str} with CUDA {cuda_ver}", "done") + print_substep(f"GPU: {gpu_name}", "done") + else: + print_substep(f"PyTorch {version_str} (CPU mode)", "done") + else: + error = torch_result.get("error", "unknown error") + print_substep(f"PyTorch: {error}", "error") + all_ok = False + + # Report FastAPI status + fastapi_result = results.get("fastapi", {}) + if fastapi_result.get("ok"): + version = fastapi_result.get("version", "") + print_substep(f"FastAPI {version}", "done") + else: + error = fastapi_result.get("error", "unknown error") + print_substep(f"FastAPI: {error}", "error") + all_ok = False + + # Report Chatterbox status + chatterbox_result = results.get("chatterbox", {}) + if chatterbox_result.get("ok"): + print_substep("Chatterbox TTS", "done") + else: + error = chatterbox_result.get("error", "unknown error") + print_substep(f"Chatterbox: {error}", "error") + all_ok = False + + # Report audio libraries status + audio_result = results.get("audio", {}) + if audio_result.get("ok"): + print_substep("Audio libraries (soundfile, librosa)", "done") + else: + error = audio_result.get("error", "unknown error") + print_substep(f"Audio libraries: {error}", "error") + all_ok = False + + return all_ok + + except subprocess.TimeoutExpired: + print_substep("Verification timed out", "warning") + return False + except Exception as e: + print_substep(f"Verification error: {e}", "warning") + return False + + +# ============================================================================ +# SERVER MANAGEMENT +# ============================================================================ + + +def read_config(root_dir): + """ + Read host and port from config.yaml using simple parsing. + + Does not require PyYAML - uses regex-based parsing. + + Args: + root_dir: Root directory of the project + + Returns: + Tuple of (host: str, port: int) + """ + config_file = root_dir / CONFIG_FILE + + # Default values + host = "0.0.0.0" + port = 8004 + + if not config_file.exists(): + return host, port + + try: + content = config_file.read_text(encoding="utf-8") + + # Simple regex-based parsing for host and port + # This handles basic YAML structure without full parsing + + # Look for host setting + host_match = re.search( + r'^\s*host:\s*["\']?([^"\'#\n\r]+)["\']?', content, re.MULTILINE + ) + if host_match: + parsed_host = host_match.group(1).strip() + if parsed_host: + host = parsed_host + + # Look for port setting + port_match = re.search(r"^\s*port:\s*(\d+)", content, re.MULTILINE) + if port_match: + parsed_port = int(port_match.group(1)) + if 1 <= parsed_port <= 65535: + port = parsed_port + + except Exception as e: + # Silently use defaults on any error + if VERBOSE_MODE: + print_warning(f"Could not parse config.yaml: {e}") + + return host, port + + +def check_port_in_use(host, port): + """ + Check if a port is already in use. + + Args: + host: Host address + port: Port number + + Returns: + True if port is in use, False otherwise + """ + try: + sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM) + sock.settimeout(1) + + # Use localhost for checking if host is 0.0.0.0 + check_host = "127.0.0.1" if host == "0.0.0.0" else host + + result = sock.connect_ex((check_host, port)) + sock.close() + + return result == 0 + + except socket.error: + return False + + +def wait_for_server(host, port, timeout=SERVER_STARTUP_TIMEOUT): + """ + Wait for server to become ready by polling the port. + + Args: + host: Host address + port: Port number + timeout: Maximum seconds to wait + + Returns: + True if server is ready, False if timeout + """ + print_substep( + "Waiting for server to start (this may take a few minutes on first run)..." + ) + + start_time = time.time() + check_host = "127.0.0.1" if host == "0.0.0.0" else host + + # Progress indicator + sys.stdout.write(" ") + sys.stdout.flush() + + dots = 0 + last_dot_time = start_time + + while time.time() - start_time < timeout: + try: + sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM) + sock.settimeout(1) + result = sock.connect_ex((check_host, port)) + sock.close() + + if result == 0: + # Server is ready + sys.stdout.write("\n") + sys.stdout.flush() + elapsed = time.time() - start_time + print_substep(f"Server ready! (took {elapsed:.1f}s)", "done") + return True + + except socket.error: + pass + + # Show progress dots + current_time = time.time() + if current_time - last_dot_time >= 2: + sys.stdout.write(".") + sys.stdout.flush() + dots += 1 + last_dot_time = current_time + + # Line wrap every 30 dots + if dots % 30 == 0: + sys.stdout.write("\n ") + sys.stdout.flush() + + time.sleep(PORT_CHECK_INTERVAL) + + # Timeout reached + sys.stdout.write("\n") + sys.stdout.flush() + print_substep(f"Timeout after {timeout}s waiting for server", "error") + return False + + +def launch_server(venv_python, root_dir): + """ + Launch the server as a subprocess. + + Args: + venv_python: Path to Python executable in venv + root_dir: Root directory of the project + + Returns: + subprocess.Popen process object + """ + server_script = root_dir / SERVER_SCRIPT + + if not server_script.exists(): + print_error(f"{SERVER_SCRIPT} not found!") + return None + + print_substep(f"Starting {SERVER_SCRIPT}...") + + # Create subprocess + # On Windows, we don't want to create a new console window + kwargs = {} + if is_windows(): + # CREATE_NO_WINDOW flag + kwargs["creationflags"] = 0 + + # For embedded Python on Windows, ensure the interpreter directory is on + # PATH so that python310.dll, vcruntime140.dll, and other co-located DLLs + # are discoverable by Windows when loading native extensions (.pyd files). + # This complements the sitecustomize.py os.add_dll_directory() approach. + env = None + embedded_dir = root_dir / EMBEDDED_PYTHON_DIR + if ( + is_windows() + and embedded_dir.exists() + and str(venv_python).startswith(str(embedded_dir)) + ): + env = os.environ.copy() + env["PATH"] = ( + f"{embedded_dir}{os.pathsep}" + f"{embedded_dir / 'Scripts'}{os.pathsep}" + f"{env.get('PATH', '')}" + ) + if VERBOSE_MODE: + print_substep("Injected embedded Python into subprocess PATH", "info") + + process = subprocess.Popen( + [str(venv_python), str(server_script)], + cwd=str(root_dir), + env=env, + **kwargs, + ) + + return process + + +def cleanup_server(process): + """ + Clean up server process gracefully. + + Args: + process: subprocess.Popen process object + """ + if process is None: + return + + if process.poll() is not None: + # Process already terminated + return + + try: + # Try graceful termination first + process.terminate() + + try: + process.wait(timeout=5) + except subprocess.TimeoutExpired: + # Force kill if graceful shutdown fails + print_substep("Force stopping server...", "warning") + process.kill() + try: + process.wait(timeout=3) + except subprocess.TimeoutExpired: + # Give up - process may be orphaned + pass + + except Exception as e: + # Process might already be gone + if VERBOSE_MODE: + print_warning(f"Error during cleanup: {e}") + + +# ============================================================================ +# ARGUMENT PARSER +# ============================================================================ + + +def parse_args(): + """ + Parse command-line arguments. + + Returns: + argparse.Namespace with parsed arguments + """ + parser = argparse.ArgumentParser( + description="Chatterbox TTS Server - Launcher", + formatter_class=argparse.RawDescriptionHelpFormatter, + epilog=""" +Examples: + python start.py # Normal start (shows menu if first run) + python start.py --reinstall # Remove and reinstall (shows menu) + python start.py --upgrade # Upgrade keeping current hardware choice + python start.py --nvidia # Install/start with NVIDIA CUDA 12.1 + python start.py --nvidia-cu128 # Install/start with NVIDIA CUDA 12.8 + python start.py --cpu # Install/start with CPU only + python start.py --rocm # Install/start with AMD ROCm + python start.py --portable # Use portable mode (Windows) + python start.py --no-portable # Use standard venv (Windows) + python start.py -v # Verbose mode (show all output) +""", + ) + + # Reinstall/upgrade options + action_group = parser.add_argument_group("Actions") + action_group.add_argument( + "--reinstall", + "-r", + action="store_true", + help="Remove existing installation and reinstall fresh (prompts for hardware selection)", + ) + action_group.add_argument( + "--upgrade", + "-u", + action="store_true", + help="Upgrade to latest version (keeps current hardware selection)", + ) + + # Direct installation type selection + install_group = parser.add_argument_group("Installation Type (skip menu)") + install_group.add_argument( + "--cpu", action="store_true", help="Install CPU-only version" + ) + install_group.add_argument( + "--nvidia", action="store_true", help="Install NVIDIA CUDA 12.1 version" + ) + install_group.add_argument( + "--nvidia-cu128", + action="store_true", + help="Install NVIDIA CUDA 12.8 version (for RTX 5090/Blackwell)", + ) + install_group.add_argument( + "--rocm", action="store_true", help="Install AMD ROCm version (Linux only)" + ) + + # Environment mode (Windows) + env_group = parser.add_argument_group("Environment Mode (Windows)") + env_group.add_argument( + "--portable", + action="store_true", + help="Use portable Python environment (Windows only, skip prompt)", + ) + env_group.add_argument( + "--no-portable", + action="store_true", + help="Use standard virtual environment instead of portable (skip prompt)", + ) + + # Other options + other_group = parser.add_argument_group("Other Options") + other_group.add_argument( + "--verbose", "-v", action="store_true", help="Show detailed installation output" + ) + + return parser.parse_args() + + +def get_install_type_from_args(args): + """ + Get installation type from command-line arguments. + + Args: + args: Parsed arguments from argparse + + Returns: + Installation type string or None if not specified + """ + if args.cpu: + return INSTALL_CPU + elif args.nvidia: + return INSTALL_NVIDIA + elif args.nvidia_cu128: + return INSTALL_NVIDIA_CU128 + elif args.rocm: + return INSTALL_ROCM + + return None + + +# ============================================================================ +# MAIN FUNCTION +# ============================================================================ + + +def main(): + """Main entry point for the launcher.""" + global VERBOSE_MODE + + # Parse command-line arguments + args = parse_args() + if args.verbose: + VERBOSE_MODE = True + + # Get root directory (where this script is located) + root_dir = Path(__file__).parent.absolute() + + # Print banner + print_banner() + + # Validate portable mode flags + if args.portable and args.no_portable: + print_error("Cannot use --portable and --no-portable together.") + sys.exit(1) + + if args.portable and not is_windows(): + print_error("Portable mode is only available on Windows.") + print("On Linux and macOS, the standard virtual environment is used.") + sys.exit(1) + + # Total steps for progress display + total_steps = 6 + + # ======================================================================== + # Step 1: Check Python version + # ======================================================================== + print_step(1, total_steps, "Checking Python installation...") + check_python_version() + + # Portable mode decision (Windows only) + # Determines whether to use the self-contained embedded Python environment + # or a standard virtual environment. On Linux/macOS, always uses venv. + use_embedded = False + + if not is_windows(): + use_embedded = False + + elif args.upgrade: + # During upgrade, preserve existing environment type + if args.portable or args.no_portable: + print_substep( + "--portable/--no-portable ignored during --upgrade. " + "Use --reinstall to switch modes.", + "warning", + ) + if is_embedded_python_available(root_dir): + use_embedded = True + print_substep( + f"Using existing portable Python {EMBEDDED_PYTHON_VERSION}", "done" + ) + else: + use_embedded = False + + elif args.no_portable: + use_embedded = False + if sys.version_info >= (3, 11): + print_substep( + f"Using system Python {sys.version_info.major}.{sys.version_info.minor}" + f" (--no-portable). Build tools may be required.", + "warning", + ) + + elif args.portable: + use_embedded = True + print_substep("Portable Mode selected via --portable flag", "info") + + elif TEST_EMBEDDED_PYTHON_PATH: + # In test mode: automatically use portable Python (no prompt) + use_embedded = True + print() + print(f"{Colors.YELLOW}{'=' * 60}{Colors.RESET}") + print( + f" {Colors.YELLOW}{Colors.ICON_INFO} TEST MODE: Simulating Python 3.11+" + ) + print( + f" {Colors.YELLOW} → Forcing portable Python environment{Colors.RESET}" + ) + print(f"{Colors.YELLOW}{'=' * 60}{Colors.RESET}") + print() + + elif not args.reinstall and is_embedded_python_available(root_dir): + # Existing portable environment found — reuse it + use_embedded = True + print_substep( + f"Using existing portable Python {EMBEDDED_PYTHON_VERSION}", "done" + ) + + elif sys.version_info >= (3, 11): + use_embedded = prompt_portable_install(reason="compatibility") + + else: + # Python 3.10 on Windows — offer portability + use_embedded = prompt_portable_install(reason="portability") + + # Set up portable Python if chosen but not yet available + if use_embedded and not is_embedded_python_available(root_dir): + print() + if not setup_embedded_python(root_dir): + print() + print_error( + "Could not set up portable Python environment. " + "Falling back to system Python." + ) + if sys.version_info >= (3, 11): + print_substep( + "You may need Visual C++ Build Tools for a successful install", + "warning", + ) + use_embedded = False + + # ======================================================================== + # Step 2: Setup paths + # ======================================================================== + print_step(2, total_steps, "Setting up environment...") + + if use_embedded: + venv_dir, venv_python, venv_pip = get_embedded_python_paths(root_dir) + print_substep(f"Project directory: {root_dir}", "info") + print_substep(f"Python environment: {venv_dir} (portable)", "info") + else: + venv_dir, venv_python, venv_pip = get_venv_paths(root_dir) + print_substep(f"Project directory: {root_dir}", "info") + print_substep(f"Virtual environment: {venv_dir}", "info") + + # ======================================================================== + # Step 3: Handle reinstall/upgrade flags + # ======================================================================== + existing_type = None + + if args.reinstall: + print_step(3, total_steps, "Preparing fresh reinstall...") + + # Remove all possible environment directories (venv and/or embedded) + for env_name in [VENV_FOLDER, EMBEDDED_PYTHON_DIR]: + env_path = root_dir / env_name + if env_path.exists(): + if not remove_venv(env_path): + print_error(f"Could not remove {env_name}/.") + print_substep( + f"Please manually delete the '{env_name}' folder and try again.", + "info", + ) + sys.exit(1) + + # Re-setup embedded Python if that path was chosen + if use_embedded: + if not setup_embedded_python(root_dir): + print_error("Failed to set up embedded Python after reinstall!") + print_substep( + "Try again, or run without --reinstall to use system Python.", + "info", + ) + sys.exit(1) + # Refresh paths after re-creation + venv_dir, venv_python, venv_pip = get_embedded_python_paths(root_dir) + + print_substep("Ready for fresh installation", "done") + + elif args.upgrade: + print_step(3, total_steps, "Preparing upgrade...") + is_installed, existing_type = get_install_state(venv_dir) + + if is_installed and existing_type: + print_substep( + f"Current installation: {INSTALL_NAMES.get(existing_type, existing_type)}", + "info", + ) + print_substep( + "Upgrading will reinstall dependencies with the same hardware selection", + "info", + ) + # Clear only the install complete marker + clear_install_complete(venv_dir) + else: + print_substep( + "No existing installation found, will perform fresh install", "warning" + ) + + else: + print_step(3, total_steps, "Checking existing installation...") + is_installed, existing_type = get_install_state(venv_dir) + + if is_installed: + type_name = INSTALL_NAMES.get(existing_type, existing_type) + print_substep(f"Found existing {type_name} installation", "done") + + # Warn Windows users on Python 3.11+ who are using system Python + if is_windows() and sys.version_info >= (3, 11) and not use_embedded: + major = sys.version_info.major + minor = sys.version_info.minor + print() + print_warning( + f" Note: You're running Python {major}.{minor} on Windows." + ) + print_warning( + " If you experience CUDA or dependency issues, your Python" + ) + print_warning(" version may be the cause. Consider reinstalling with") + print_warning( + " portable mode: python start.py --reinstall --portable" + ) + else: + print_substep("No existing installation found", "info") + + # ======================================================================== + # Step 4: Installation flow (if needed) + # ======================================================================== + is_installed, current_type = get_install_state(venv_dir) + + if not is_installed: + print_step(4, total_steps, "Installing Chatterbox TTS Server...") + + # Create environment if it doesn't exist + if not venv_dir.exists(): + if use_embedded: + # Re-setup embedded Python (e.g., after a partial failure) + if not setup_embedded_python(root_dir): + print_error("Failed to set up embedded Python environment!") + sys.exit(1) + venv_dir, venv_python, venv_pip = get_embedded_python_paths(root_dir) + else: + if not create_venv(venv_dir): + print_error("Failed to create virtual environment!") + print() + print("Try creating it manually:") + print(f" python -m venv {VENV_FOLDER}") + print() + sys.exit(1) + + # Determine installation type + install_type = None + + # First check CLI flags + install_type = get_install_type_from_args(args) + + # If upgrading, use the existing type + if install_type is None and existing_type: + install_type = existing_type + print_substep( + f"Using existing hardware selection: {INSTALL_NAMES.get(install_type, install_type)}", + "info", + ) + + # If still no type, show menu + if install_type is None: + print() + print_substep("Detecting available hardware...", "info") + gpu_info = detect_gpu() + default_choice = get_default_choice(gpu_info) + install_type = show_installation_menu(gpu_info, default_choice) + + # Show selected type + type_name = INSTALL_NAMES.get(install_type, install_type) + print() + print_substep(f"Selected: {type_name}", "done") + + # ROCm warning on Windows + if install_type == INSTALL_ROCM and is_windows(): + print() + print_warning("=" * 60) + print_warning(" ⚠️ WARNING: ROCm is not supported on Windows!") + print_warning("=" * 60) + print() + print_warning(" ROCm (AMD GPU acceleration) only works on Linux.") + print_warning(" Installation will proceed, but GPU acceleration") + print_warning(" will NOT work. The server will run on CPU only.") + print() + + try: + response = input(" Continue anyway? (y/n) [n]: ").strip().lower() + if response != "y": + print() + print(" Installation cancelled.") + print(" Tip: Use --nvidia for NVIDIA GPUs or --cpu for CPU-only.") + sys.exit(2) + except (EOFError, KeyboardInterrupt): + print() + print(" Cancelled.") + sys.exit(2) + + print() + + # Upgrade pip + print() + upgrade_pip(venv_python) + + # Perform installation + print() + success = perform_installation(venv_pip, install_type, root_dir) + + if not success: + print() + print_error("=" * 60) + print_error(" Installation failed!") + print_error("=" * 60) + print() + print("Troubleshooting tips:") + print() + print(" 1. Check your internet connection") + print(" 2. Try running with --verbose for more details:") + print(" python start.py --reinstall --verbose") + print() + print(" 3. Check if you have enough disk space") + print() + print(" 4. Try installing manually:") + requirements_file = REQUIREMENTS_MAP.get(install_type, "requirements.txt") + print(f" pip install -r {requirements_file}") + if install_type == INSTALL_NVIDIA_CU128: + print(f" pip install --no-deps {CHATTERBOX_REPO}") + print() + sys.exit(1) + + # Patch chatterbox to make watermarker gracefully optional + print() + print_substep("Applying post-install patches...") + _patch_chatterbox_watermarker(venv_dir, use_embedded) + + # Verify installation + print() + verification_ok = verify_installation(venv_python) + + if not verification_ok: + print() + print_warning("Installation verification had some issues.") + print_warning("The server may still work. Attempting to continue...") + + # Save installation state + save_install_state(venv_dir, install_type) + + print() + print_success("=" * 60) + print_success(" Installation complete!") + print_success("=" * 60) + + else: + print_step(4, total_steps, "Using existing installation...") + type_name = INSTALL_NAMES.get(current_type, current_type or "unknown") + print_substep(f"Installation type: {type_name}", "done") + + # ======================================================================== + # Step 5: Read configuration + # ======================================================================== + print_step(5, total_steps, "Loading configuration...") + + host, port = read_config(root_dir) + print_substep(f"Server will run on {host}:{port}", "done") + + # Check if port is already in use + if check_port_in_use(host, port): + print() + print_error("=" * 60) + print_error(f" Port {port} is already in use!") + print_error("=" * 60) + print() + print("Another instance may be running, or another program is using this port.") + print() + print("Options:") + print(f" 1. Stop the other process using port {port}") + print(f" 2. Change the port in {CONFIG_FILE}") + print() + sys.exit(1) + + # ======================================================================== + # Step 6: Launch server + # ======================================================================== + print_step(6, total_steps, "Launching Chatterbox TTS Server...") + + server_process = launch_server(venv_python, root_dir) + + if server_process is None: + print_error("Failed to launch server!") + sys.exit(1) + + # Wait for server to become ready + server_ready = wait_for_server(host, port) + + if not server_ready: + print() + print_error("=" * 60) + print_error(" Server failed to start!") + print_error("=" * 60) + print() + print("The server did not become ready within the timeout period.") + print() + print("Common causes:") + print(" - Missing CUDA drivers (for GPU installation)") + print(" - Insufficient memory (model requires ~8GB+ VRAM)") + print(" - Network issues downloading the model") + print(" - Port conflict") + print() + print("Check the server output above for error messages.") + print() + print("Try running with verbose mode for more details:") + print(" python start.py --verbose") + print() + + cleanup_server(server_process) + sys.exit(1) + + # Show success status + print_status_box(host, port) + print_reinstall_hint() + + # ======================================================================== + # Keep running until interrupted + # ======================================================================== + try: + while True: + # Check if server process is still running + exit_code = server_process.poll() + + if exit_code is not None: + # Server has exited + print() + if exit_code == 0: + print_substep("Server stopped normally", "done") + else: + print_substep(f"Server exited with code {exit_code}", "warning") + break + + time.sleep(1) + + except KeyboardInterrupt: + print() + print() + print("-" * 40) + print("Shutting down Chatterbox TTS Server...") + print("-" * 40) + + cleanup_server(server_process) + + print() + print("Server stopped. Goodbye!") + print() + sys.exit(0) + + # Clean up + cleanup_server(server_process) + + # Exit with server's exit code + exit_code = server_process.returncode if server_process.returncode else 0 + sys.exit(exit_code) + + +# ============================================================================ +# ENTRY POINT +# ============================================================================ + +if __name__ == "__main__": + try: + main() + except KeyboardInterrupt: + print() + print() + print("Interrupted by user.") + sys.exit(2) + except Exception as e: + print() + print_error(f"Unexpected error: {e}") + print() + if VERBOSE_MODE: + import traceback + + traceback.print_exc() + else: + print("Run with --verbose for more details.") + print() + sys.exit(1) diff --git a/ui/index.html b/ui/index.html index 5a0cb21..77dd6a3 100644 --- a/ui/index.html +++ b/ui/index.html @@ -134,11 +134,19 @@

Generate Speech

Generate Speech - +
+ + +
`; - // Create close button const closeButton = document.createElement('button'); closeButton.type = 'button'; closeButton.className = 'notification__close'; @@ -221,12 +224,54 @@ document.addEventListener('DOMContentLoaded', async function () { return `${minutes}:${secs}`; } + function visualizeStream(canvasElement) { + if (!analyser || !canvasElement) return; + + const canvasCtx = canvasElement.getContext("2d"); + const bufferLength = analyser.frequencyBinCount; + const dataArray = new Uint8Array(bufferLength); + + const draw = () => { + if (!isGenerating) { + cancelAnimationFrame(visualizerAnimationId); + return; + } + visualizerAnimationId = requestAnimationFrame(draw); + + analyser.getByteTimeDomainData(dataArray); + + canvasCtx.fillStyle = document.documentElement.classList.contains('dark') ? '#1e293b' : '#f1f5f9'; + canvasCtx.fillRect(0, 0, canvasElement.width, canvasElement.height); + + canvasCtx.lineWidth = 2; + canvasCtx.strokeStyle = document.documentElement.classList.contains('dark') ? '#6366f1' : '#4f46e5'; + canvasCtx.beginPath(); + + const sliceWidth = canvasElement.width * 1.0 / bufferLength; + let x = 0; + + for (let i = 0; i < bufferLength; i++) { + const v = dataArray[i] / 128.0; + const y = v * canvasElement.height / 2; + + if (i === 0) canvasCtx.moveTo(x, y); + else canvasCtx.lineTo(x, y); + + x += sliceWidth; + } + + canvasCtx.lineTo(canvasElement.width, canvasElement.height / 2); + canvasCtx.stroke(); + }; + + draw(); + } + // --- Theme Management --- function applyTheme(theme) { const isDark = theme === 'dark'; document.documentElement.classList.toggle('dark', isDark); - // WaveSurfer color update if (wavesurfer) { wavesurfer.setOptions({ waveColor: isDark ? '#6366f1' : '#a5b4fc', @@ -279,7 +324,6 @@ document.addEventListener('DOMContentLoaded', async function () { } function debouncedSaveState() { - // Do not save anything until the entire UI has finished its initial setup. if (!uiReady || !listenersAttached) { return; } clearTimeout(saveStateTimeout); saveStateTimeout = setTimeout(saveCurrentUiState, DEBOUNCE_DELAY_MS); @@ -298,7 +342,7 @@ document.addEventListener('DOMContentLoaded', async function () { } } - // --- Model Management Functions (New Features) --- + // --- Model Management Functions --- function updateModelUI(modelInfo) { if (!modelInfo) { @@ -308,11 +352,9 @@ document.addEventListener('DOMContentLoaded', async function () { currentModelInfo = modelInfo; - // Update model indicator badge if (modelIndicator && modelBadge) { modelIndicator.classList.remove('hidden'); - // Use simplified modifier classes if (modelInfo.type === 'turbo') { modelBadge.className = 'model-badge turbo'; modelBadgeText.textContent = '⚡ Turbo'; @@ -325,7 +367,6 @@ document.addEventListener('DOMContentLoaded', async function () { } } - // Update model status indicator if (modelStatusIndicator && modelStatusText) { if (modelInfo.loaded) { modelStatusIndicator.className = 'status-dot success'; @@ -338,7 +379,6 @@ document.addEventListener('DOMContentLoaded', async function () { } } - // Update model selector dropdown to match loaded model if (modelSelect && !modelChangesPending) { let selectorValue = 'chatterbox'; if (modelInfo.type === 'turbo') { @@ -350,11 +390,9 @@ document.addEventListener('DOMContentLoaded', async function () { selectedModelSelector = selectorValue; } - // Show/hide model-specific UI sections const exaggerationGroup = document.getElementById('exaggeration-group'); const cfgWeightGroup = document.getElementById('cfg-weight-group'); - // Show/hide paralinguistic tags section (Turbo only) if (paralinguisticTagsSection) { if (modelInfo.type === 'turbo' && modelInfo.supports_paralinguistic_tags) { paralinguisticTagsSection.classList.remove('hidden'); @@ -363,7 +401,6 @@ document.addEventListener('DOMContentLoaded', async function () { } } - // Hide exaggeration and CFG for turbo model if (modelInfo.type === 'turbo') { exaggerationGroup?.classList.add('hidden'); cfgWeightGroup?.classList.add('hidden'); @@ -372,10 +409,7 @@ document.addEventListener('DOMContentLoaded', async function () { cfgWeightGroup?.classList.remove('hidden'); } - // Refresh presets to filter based on current model type populatePresets(); - - // Update language options based on model type updateLanguageOptions(modelInfo.type); console.log('Model UI updated:', modelInfo); @@ -388,23 +422,18 @@ document.addEventListener('DOMContentLoaded', async function () { const isMultilingual = modelType === 'multilingual'; const languages = isMultilingual ? LANGUAGES_MULTILINGUAL : LANGUAGES_ENGLISH_ONLY; - // Save current selection before switching away from Multilingual if (!isMultilingual && currentValue && currentValue !== 'en') { lastMultilingualLanguage = currentValue; } - // Show/hide language selector based on model type - // Only show for multilingual model (or if config says to show it) if (isMultilingual) { languageSelectContainer.classList.remove('hidden'); } else { languageSelectContainer.classList.add('hidden'); } - // Clear existing options languageSelect.innerHTML = ''; - // Populate with appropriate languages languages.forEach(lang => { const option = document.createElement('option'); option.value = lang.code; @@ -412,9 +441,7 @@ document.addEventListener('DOMContentLoaded', async function () { languageSelect.appendChild(option); }); - // Restore appropriate selection if (isMultilingual) { - // Restore last Multilingual language selection languageSelect.value = lastMultilingualLanguage; } else { languageSelect.value = 'en'; @@ -429,7 +456,6 @@ document.addEventListener('DOMContentLoaded', async function () { const textBefore = textArea.value.substring(0, startPos); const textAfter = textArea.value.substring(endPos); - // Insert tag with a space after if not at end and next char isn't a space let insertText = tag; if (textAfter.length > 0 && textAfter[0] !== ' ') { insertText = tag + ' '; @@ -437,17 +463,14 @@ document.addEventListener('DOMContentLoaded', async function () { textArea.value = textBefore + insertText + textAfter; - // Update cursor position to after the inserted tag const newCursorPos = startPos + insertText.length; textArea.setSelectionRange(newCursorPos, newCursorPos); textArea.focus(); - // Update character count if (charCount) { charCount.textContent = textArea.value.length; } - // Trigger state save debouncedSaveState(); } @@ -465,12 +488,10 @@ document.addEventListener('DOMContentLoaded', async function () { if (newSelector !== currentSelector) { modelChangesPending = true; - // Show the apply button if (applyModelBtn) { applyModelBtn.classList.remove('hidden'); } - // Update status indicator and text to show pending state if (modelStatusIndicator) { modelStatusIndicator.className = 'status-dot warning'; } @@ -481,12 +502,10 @@ document.addEventListener('DOMContentLoaded', async function () { } else { modelChangesPending = false; - // Hide the apply button if (applyModelBtn) { applyModelBtn.classList.add('hidden'); } - // Restore status from current model info updateModelUI(currentModelInfo); } } @@ -497,7 +516,6 @@ document.addEventListener('DOMContentLoaded', async function () { const newSelector = modelSelect.value; - // Update status if (modelStatusText) { modelStatusText.textContent = 'Saving configuration...'; } @@ -513,7 +531,6 @@ document.addEventListener('DOMContentLoaded', async function () { } try { - // Save the model selector to config const response = await fetch(`${API_BASE_URL}/save_settings`, { method: 'POST', headers: { 'Content-Type': 'application/json' }, @@ -531,7 +548,6 @@ document.addEventListener('DOMContentLoaded', async function () { showNotification('Model configuration saved. Initiating server restart...', 'info'); - // Trigger server restart const restartResponse = await fetch(`${API_BASE_URL}/restart_server`, { method: 'POST' }); @@ -543,7 +559,6 @@ document.addEventListener('DOMContentLoaded', async function () { 10000 ); - // Attempt to reload after delay setTimeout(() => { window.location.reload(); }, 5000); @@ -559,7 +574,6 @@ document.addEventListener('DOMContentLoaded', async function () { console.error('Error applying model change:', error); showNotification(`Error: ${error.message}`, 'error'); - // Re-enable button if (applyModelBtn) { applyModelBtn.disabled = false; applyModelBtn.innerHTML = ` @@ -589,7 +603,7 @@ document.addEventListener('DOMContentLoaded', async function () { if (languageSelectContainer && currentConfig?.ui?.show_language_select === false) { languageSelectContainer.classList.add('hidden'); } - updateSpeedFactorWarning(); // Initial check for speed factor warning + updateSpeedFactorWarning(); const initialGenResult = currentConfig.initial_gen_result; if (initialGenResult && initialGenResult.outputUrl) { initializeWaveSurfer(initialGenResult.outputUrl, initialGenResult); @@ -613,7 +627,6 @@ document.addEventListener('DOMContentLoaded', async function () { hideGenerationWarning = currentUiState.hide_generation_warning || false; currentVoiceMode = currentUiState.last_voice_mode || 'predefined'; - // NEW: Handle model info from initial data if (data.model_info) { updateModelUI(data.model_info); } @@ -627,16 +640,13 @@ document.addEventListener('DOMContentLoaded', async function () { currentConfig = { ui: { title: "Chatterbox TTS Server (Error Mode)" }, generation_defaults: {}, ui_state: {} }; currentUiState = currentConfig.ui_state; } - initializeApplication(); // Attempt to init in a degraded state + initializeApplication(); } finally { - // --- PHASE 2: Attach listeners and enable UI readiness --- - // This pushes the listener attachment to the end of the event queue, - // ensuring all initialization events have fired harmlessly before we start listening. setTimeout(() => { attachStateSavingListeners(); listenersAttached = true; uiReady = true; - }, 50); // A 50ms delay is more robust than 0ms for complex UIs. + }, 50); } } @@ -646,12 +656,10 @@ document.addEventListener('DOMContentLoaded', async function () { if (charCount) charCount.textContent = textArea.value.length; } - // Handle Voice Mode Selection const modeRadioToSelect = document.querySelector(`input[name="voice_mode"][value="${currentVoiceMode}"]`); if (modeRadioToSelect) { modeRadioToSelect.checked = true; - // FIX: Manually fire the change event so the .selected class updates visually modeRadioToSelect.dispatchEvent(new Event('change')); } else { const defaultRadio = document.querySelector('input[name="voice_mode"][value="predefined"]'); @@ -688,27 +696,18 @@ document.addEventListener('DOMContentLoaded', async function () { if (hideChunkWarningCheckbox) hideChunkWarningCheckbox.checked = hideChunkWarning; if (hideGenerationWarningCheckbox) hideGenerationWarningCheckbox.checked = hideGenerationWarning; - // --- PRESET RESTORATION LOGIC --- - - // 1. Restore the name from state variable if (currentUiState.last_preset_name) { currentPresetName = currentUiState.last_preset_name; } - // 2. Logic to apply preset (if empty) OR just highlight button (if text exists) if (textArea && !textArea.value && appPresets && appPresets.length > 0) { - // Case A: No text entered. We want to load a preset fully. - // Priority: Saved preset > "Standard Narration" > First available const savedPreset = appPresets.find(p => p.name === currentPresetName); const defaultPreset = savedPreset || appPresets.find(p => p.name === "Standard Narration") || appPresets[0]; if (defaultPreset) { - // Apply values AND visuals, no notification, no save applyPreset(defaultPreset, false, false); } } else if (currentPresetName) { - // Case B: Text already exists (restored from last_text). - // We don't want to overwrite parameters, but we want to show which preset button was active. updatePresetVisuals(currentPresetName); } } @@ -734,7 +733,7 @@ document.addEventListener('DOMContentLoaded', async function () { const valueDisplay = document.getElementById(valueDisplayId); slider.addEventListener('input', () => { if (valueDisplay) valueDisplay.textContent = slider.value; - if (slider.id === 'speed-factor') updateSpeedFactorWarning(); // Update warning on input + if (slider.id === 'speed-factor') updateSpeedFactorWarning(); }); slider.addEventListener('change', debouncedSaveState); } @@ -742,7 +741,6 @@ document.addEventListener('DOMContentLoaded', async function () { if (languageSelect) languageSelect.addEventListener('change', debouncedSaveState); if (outputFormatSelect) outputFormatSelect.addEventListener('change', debouncedSaveState); - // NEW: Model management listeners if (modelSelect) { modelSelect.addEventListener('change', handleModelSelectChange); } @@ -751,7 +749,6 @@ document.addEventListener('DOMContentLoaded', async function () { applyModelBtn.addEventListener('click', applyModelChange); } - // NEW: Tag button listeners tagButtons.forEach(button => { button.addEventListener('click', (e) => { const tag = e.currentTarget.getAttribute('data-tag'); @@ -762,7 +759,6 @@ document.addEventListener('DOMContentLoaded', async function () { }); } - // --- Dynamic UI Population --- function populatePredefinedVoices(voicesData = initialPredefinedVoices) { if (!predefinedVoiceSelect) return; const currentSelectedValue = predefinedVoiceSelect.value; @@ -808,11 +804,8 @@ document.addEventListener('DOMContentLoaded', async function () { function updatePresetVisuals(name) { currentPresetName = name; - - // Find all preset buttons const buttons = document.querySelectorAll('.preset-btn'); buttons.forEach(btn => { - // We will add data-name to buttons in the next step if (btn.dataset.name === name) { btn.classList.add('selected'); } else { @@ -824,8 +817,6 @@ document.addEventListener('DOMContentLoaded', async function () { function populatePresets() { if (!presetsContainer || !appPresets) return; - // Filter presets based on current model - // Hide "Turbo" presets when Chatterbox-Original is loaded let filteredPresets = appPresets; if (currentModelInfo && currentModelInfo.type !== 'turbo') { filteredPresets = appPresets.filter(preset => @@ -833,7 +824,6 @@ document.addEventListener('DOMContentLoaded', async function () { ); } - // Clear container presetsContainer.innerHTML = ''; if (filteredPresets.length === 0) { @@ -914,7 +904,6 @@ document.addEventListener('DOMContentLoaded', async function () { } } - // --- Voice Mode and Options Visibility --- function toggleVoiceOptionsDisplay() { const selectedMode = document.querySelector('input[name="voice_mode"]:checked')?.value; currentVoiceMode = selectedMode; @@ -932,34 +921,35 @@ document.addEventListener('DOMContentLoaded', async function () { } if (splitTextToggle) toggleChunkControlsVisibility(); - // --- Audio Player (WaveSurfer) --- function initializeWaveSurfer(audioUrl, resultDetails = {}) { if (wavesurfer) { - wavesurfer.unAll(); // Remove all event listeners before destroying + wavesurfer.unAll(); wavesurfer.destroy(); wavesurfer = null; } - if (currentAudioBlobUrl) { + if (currentAudioBlobUrl && !resultDetails.isStreaming) { URL.revokeObjectURL(currentAudioBlobUrl); currentAudioBlobUrl = null; } - currentAudioBlobUrl = audioUrl; + + const isStreaming = resultDetails.isStreaming || false; - // Ensure the container is clean or re-created audioPlayerContainer.innerHTML = `
-

Generated Audio

-
+

${isStreaming ? 'Streaming Audio...' : 'Generated Audio'}

+
+ ${isStreaming ? '' : ''} +
- - + @@ -968,87 +958,103 @@ document.addEventListener('DOMContentLoaded', async function () {
- Mode: -- - - • Gen Time: --s + Mode: ${resultDetails.submittedVoiceMode || currentVoiceMode} + • Gen Time: ${resultDetails.genTime || '--'}s • Duration: --:--
`; - // Re-select elements after recreating them - const waveformDiv = audioPlayerContainer.querySelector('#waveform'); const playBtn = audioPlayerContainer.querySelector('#play-btn'); const downloadLink = audioPlayerContainer.querySelector('#download-link'); - const playerModeSpan = audioPlayerContainer.querySelector('#player-voice-mode'); - const playerFileSpan = audioPlayerContainer.querySelector('#player-voice-file-details'); - const playerGenTimeSpan = audioPlayerContainer.querySelector('#player-gen-time'); const audioDurationSpan = audioPlayerContainer.querySelector('#audio-duration'); - const audioFilename = resultDetails.filename || (typeof audioUrl === 'string' ? audioUrl.split('/').pop() : 'tts_output.wav'); - if (downloadLink) { - downloadLink.href = audioUrl; - downloadLink.download = audioFilename; - const downloadTextSpan = downloadLink.querySelector('span'); // Target the span for text update - if (downloadTextSpan) { - downloadTextSpan.textContent = `Download ${audioFilename.split('.').pop().toUpperCase()}`; - } - } - if (playerModeSpan) playerModeSpan.textContent = resultDetails.submittedVoiceMode || currentVoiceMode || '--'; - if (playerFileSpan) { - let fileDetail = ''; - if ((resultDetails.submittedVoiceMode || currentVoiceMode) === 'clone' && resultDetails.submittedCloneFile) { - fileDetail = `(${resultDetails.submittedCloneFile})`; - } else if ((resultDetails.submittedVoiceMode || currentVoiceMode) === 'predefined' && resultDetails.submittedPredefinedVoice) { - fileDetail = `(${resultDetails.submittedPredefinedVoice})`; - } - playerFileSpan.innerHTML = fileDetail; + const playIcon = `Play`; + const pauseIcon = `Pause`; + + if (isStreaming) { + playBtn.onclick = () => { + if (audioCtx.state === 'running') { + audioCtx.suspend(); + playBtn.innerHTML = playIcon; + } else { + audioCtx.resume(); + playBtn.innerHTML = pauseIcon; + } + }; + return; // Don't init Wavesurfer yet } - if (playerGenTimeSpan) playerGenTimeSpan.textContent = resultDetails.genTime ? `${resultDetails.genTime}s` : '--s'; - const playIconSVG = `Play`; - const pauseIconSVG = `Pause`; + // Standard Init for completed audio const isDark = document.documentElement.classList.contains('dark'); - wavesurfer = WaveSurfer.create({ - container: waveformDiv, waveColor: isDark ? '#6366f1' : '#a5b4fc', progressColor: isDark ? '#4f46e5' : '#6366f1', - cursorColor: isDark ? '#cbd5e1' : '#475569', barWidth: 3, barRadius: 3, cursorWidth: 1, height: 80, barGap: 2, - responsive: true, url: audioUrl, mediaControls: false, normalize: true, + container: '#waveform', + waveColor: isDark ? '#6366f1' : '#a5b4fc', + progressColor: isDark ? '#4f46e5' : '#6366f1', + cursorColor: isDark ? '#cbd5e1' : '#475569', + barWidth: 3, + barRadius: 3, + height: 80, + responsive: true, + url: audioUrl, + normalize: true, }); wavesurfer.on('ready', () => { const duration = wavesurfer.getDuration(); if (audioDurationSpan) audioDurationSpan.textContent = formatTime(duration); - if (playBtn) { - playBtn.disabled = false; - playBtn.innerHTML = playIconSVG; - } - if (downloadLink) { - downloadLink.classList.remove('disabled'); - downloadLink.removeAttribute('aria-disabled'); - } - }); - wavesurfer.on('play', () => { if (playBtn) playBtn.innerHTML = pauseIconSVG; }); - wavesurfer.on('pause', () => { if (playBtn) playBtn.innerHTML = playIconSVG; }); - wavesurfer.on('finish', () => { if (playBtn) playBtn.innerHTML = playIconSVG; wavesurfer.seekTo(0); }); - wavesurfer.on('error', (err) => { - console.error("WaveSurfer error:", err); - showNotification(`Error loading audio waveform: ${err.message || err}`, 'error'); - if (waveformDiv) waveformDiv.innerHTML = `

Could not load waveform.

`; - if (playBtn) playBtn.disabled = true; + playBtn.disabled = false; + playBtn.innerHTML = playIcon; + downloadLink.classList.remove('disabled'); + downloadLink.href = audioUrl; }); - if (playBtn) { - playBtn.onclick = () => { - if (wavesurfer) { - wavesurfer.playPause(); - } - }; - } + wavesurfer.on('play', () => { playBtn.innerHTML = pauseIcon; }); + wavesurfer.on('pause', () => { playBtn.innerHTML = playIcon; }); + wavesurfer.on('finish', () => { playBtn.innerHTML = playIcon; wavesurfer.seekTo(0); }); + + playBtn.onclick = () => wavesurfer.playPause(); + setTimeout(() => audioPlayerContainer.scrollIntoView({ behavior: 'smooth', block: 'nearest' }), 150); } + function streamAudio(arrayBuffer, sampleRate = 24000) { + if (!audioCtx) { + audioCtx = new (window.AudioContext || window.webkitAudioContext)({ + sampleRate: sampleRate + }); + audioNextStartTime = audioCtx.currentTime; + + analyser = audioCtx.createAnalyser(); + analyser.fftSize = 256; // Smaller for a "bar" look or 2048 for smooth + } else if (audioCtx.state === 'suspended') { + // Context stays suspended if user pressed pause + } + + const int16Data = new Int16Array(arrayBuffer); + const float32Data = new Float32Array(int16Data.length); + + for (let i = 0; i < int16Data.length; i++) { + float32Data[i] = int16Data[i] < 0 ? int16Data[i] / 32768 : int16Data[i] / 32767; + } + + const audioBuffer = audioCtx.createBuffer(1, float32Data.length, sampleRate); + audioBuffer.copyToChannel(float32Data, 0, 0); + + const source = audioCtx.createBufferSource(); + source.buffer = audioBuffer; + + source.connect(analyser); + analyser.connect(audioCtx.destination); + + // Contiguous scheduling + const start = Math.max(audioNextStartTime, audioCtx.currentTime); + source.start(start); + + audioNextStartTime = start + audioBuffer.duration; + } + // --- TTS Generation Logic --- function getTTSFormData() { const jsonData = { @@ -1062,7 +1068,7 @@ document.addEventListener('DOMContentLoaded', async function () { voice_mode: currentVoiceMode, split_text: splitTextToggle.checked, chunk_size: parseInt(chunkSizeSlider.value, 10), - output_format: outputFormatSelect.value || 'mp3' + output_format: 'wav' }; if (currentVoiceMode === 'predefined' && predefinedVoiceSelect.value !== 'none') { jsonData.predefined_voice_id = predefinedVoiceSelect.value; @@ -1073,37 +1079,154 @@ document.addEventListener('DOMContentLoaded', async function () { } async function submitTTSRequest() { + const streamToggle = document.getElementById('stream-audio-toggle'); + const useStreaming = streamToggle ? streamToggle.checked : false; + isGenerating = true; + + if (audioCtx) { + audioCtx.close(); + audioCtx = null; + } + analyser = null; + if (visualizerAnimationId) cancelAnimationFrame(visualizerAnimationId); + leftoverChunkBytes = new Uint8Array(0); + audioNextStartTime = 0; + showLoadingOverlay(); const startTime = performance.now(); const jsonData = getTTSFormData(); + + let fullAudioParts = []; + try { - const response = await fetch(`${API_BASE_URL}/tts`, { + const endpoint = useStreaming ? '/tts/stream' : '/tts'; + + if (useStreaming) { + // Initialize player UI immediately in "Live" mode + initializeWaveSurfer(null, { + isStreaming: true, + submittedVoiceMode: jsonData.voice_mode + }); + + const canvas = document.getElementById('stream-canvas'); + if (canvas) { + canvas.width = canvas.parentElement.offsetWidth; + canvas.height = 80; + } + } + + const response = await fetch(`${API_BASE_URL}${endpoint}`, { method: 'POST', headers: { 'Content-Type': 'application/json' }, body: JSON.stringify(jsonData) }); + if (!response.ok) { const errorResult = await response.json().catch(() => ({ detail: `HTTP error ${response.status}` })); throw new Error(errorResult.detail || 'TTS generation failed.'); } - const audioBlob = await response.blob(); - const endTime = performance.now(); - const genTime = ((endTime - startTime) / 1000).toFixed(2); - const filenameFromServer = response.headers.get('Content-Disposition')?.split('filename=')[1]?.replace(/"/g, '') || 'generated_audio.wav'; - const resultDetails = { - outputUrl: URL.createObjectURL(audioBlob), filename: filenameFromServer, genTime: genTime, - submittedVoiceMode: jsonData.voice_mode, submittedPredefinedVoice: jsonData.predefined_voice_id, - submittedCloneFile: jsonData.reference_audio_filename - }; - initializeWaveSurfer(resultDetails.outputUrl, resultDetails); - showNotification('Audio generated successfully!', 'success'); + + if (useStreaming) { + const reader = response.body.getReader(); + let receivedFirstByte = false; + let wavHeaderRemoved = false; + let sampleRate = 24000; + + while (true) { + const { done, value } = await reader.read(); + if (done) break; + + if (!receivedFirstByte) { + receivedFirstByte = true; + hideLoadingOverlay(); + const canvas = document.getElementById('stream-canvas'); + if (canvas) visualizeStream(canvas); + } + + // Byte-alignment protection (Fixes static/gibberish) + const combinedBuffer = new Uint8Array(leftoverChunkBytes.length + value.length); + combinedBuffer.set(leftoverChunkBytes); + combinedBuffer.set(value, leftoverChunkBytes.length); + + const remainder = combinedBuffer.length % 2; + let chunkToProcess = combinedBuffer; + + if (remainder !== 0) { + chunkToProcess = combinedBuffer.slice(0, combinedBuffer.length - 1); + leftoverChunkBytes = combinedBuffer.slice(combinedBuffer.length - 1); + } else { + leftoverChunkBytes = new Uint8Array(0); + } + + fullAudioParts.push(value); + + let audioDataForCtx = chunkToProcess; + + if (!wavHeaderRemoved && audioDataForCtx.length >= 44) { + const view = new DataView(audioDataForCtx.buffer, audioDataForCtx.byteOffset, audioDataForCtx.byteLength); + sampleRate = view.getUint32(24, true); + audioDataForCtx = audioDataForCtx.slice(44); + wavHeaderRemoved = true; + } else if (!wavHeaderRemoved) { + continue; + } + + if (audioDataForCtx.length > 0) { + streamAudio(audioDataForCtx.buffer, sampleRate); + } + } + + if (visualizerAnimationId) cancelAnimationFrame(visualizerAnimationId); + + // Convert accumulated stream to permanent blob for standard player + const blob = new Blob(fullAudioParts, { type: 'audio/wav' }); + const endTime = performance.now(); + const genTime = ((endTime - startTime) / 1000).toFixed(2); + + const resultDetails = { + outputUrl: URL.createObjectURL(blob), + filename: `tts_stream_${Date.now()}.wav`, + genTime: genTime, + submittedVoiceMode: jsonData.voice_mode, + submittedPredefinedVoice: jsonData.predefined_voice_id, + submittedCloneFile: jsonData.reference_audio_filename + }; + + // Swap the Live canvas for a real Wavesurfer waveform + initializeWaveSurfer(resultDetails.outputUrl, resultDetails); + showNotification('Stream completed successfully!', 'success', 2000); + + } else { + // ... Existing Batch Logic ... + const audioBlob = await response.blob(); + const endTime = performance.now(); + const genTime = ((endTime - startTime) / 1000).toFixed(2); + const filenameFromServer = response.headers.get('Content-Disposition')?.split('filename=')[1]?.replace(/"/g, '') || 'generated_audio.wav'; + + const resultDetails = { + outputUrl: URL.createObjectURL(audioBlob), + filename: filenameFromServer, + genTime: genTime, + submittedVoiceMode: jsonData.voice_mode, + submittedPredefinedVoice: jsonData.predefined_voice_id, + submittedCloneFile: jsonData.reference_audio_filename + }; + initializeWaveSurfer(resultDetails.outputUrl, resultDetails); + showNotification('Audio generated successfully!', 'success'); + hideLoadingOverlay(); + } + } catch (error) { console.error('TTS Generation Error:', error); - showNotification(error.message || 'An unknown error occurred during TTS generation.', 'error'); + showNotification(error.message || 'An unknown error occurred.', 'error'); + hideLoadingOverlay(); + if (visualizerAnimationId) cancelAnimationFrame(visualizerAnimationId); + if (audioPlayerContainer.innerHTML.includes('Streaming Audio')) { + audioPlayerContainer.innerHTML = ''; + } } finally { isGenerating = false; - hideLoadingOverlay(); } } @@ -1118,17 +1241,12 @@ document.addEventListener('DOMContentLoaded', async function () { submitTTSRequest(); } - // --- Attach main generation event to the button's CLICK, not the form's SUBMIT --- - // This is a more robust method that prevents accidental submissions during page load. if (generateBtn) { generateBtn.addEventListener('click', function (event) { + if (audioCtx && audioCtx.state === 'suspended') { + audioCtx.resume(); + } - console.log('Generate button clicked!'); - console.log('Current voice mode:', currentVoiceMode); - console.log('Is generating:', isGenerating); - console.log('Text content:', textArea ? textArea.value.trim() : 'NO TEXTAREA'); - - // We still prevent default in case the button has any default browser actions. event.preventDefault(); if (isGenerating) { @@ -1149,17 +1267,13 @@ document.addEventListener('DOMContentLoaded', async function () { return; } - // Check for the generation quality warning. if (!hideGenerationWarning) { showGenerationWarningModal(); - return; // Stop here and let the modal handler take over. + return; } - // If the warning is hidden, proceed to the final checks. proceedWithSubmissionChecks(); }); - } else { - console.log('Generate button not found!'); } // --- Modal Handling --- @@ -1209,10 +1323,11 @@ document.addEventListener('DOMContentLoaded', async function () { }); function showLoadingOverlay() { if (loadingOverlay && generateBtn && loadingCancelBtn) { - loadingMessage.textContent = 'Generating audio...'; - loadingStatusText.textContent = 'Please wait. This may take some time.'; + loadingMessage.textContent = 'Generating...'; + loadingStatusText.textContent = 'Waiting for stream to start...'; loadingOverlay.style.display = 'flex'; - loadingOverlay.classList.remove('hidden', 'opacity-0'); loadingOverlay.dataset.state = 'open'; + loadingOverlay.classList.remove('hidden', 'opacity-0'); + loadingOverlay.dataset.state = 'open'; generateBtn.disabled = true; loadingCancelBtn.disabled = false; } } @@ -1231,11 +1346,16 @@ document.addEventListener('DOMContentLoaded', async function () { function displayServerConfiguration() { if (!serverConfigForm || !currentConfig || Object.keys(currentConfig).length === 0) return; const fieldsToDisplay = { - "server.host": currentConfig.server?.host, "server.port": currentConfig.server?.port, - "tts_engine.device": currentConfig.tts_engine?.device, "tts_engine.default_voice_id": currentConfig.tts_engine?.default_voice_id, - "paths.model_cache": currentConfig.paths?.model_cache, "tts_engine.predefined_voices_path": currentConfig.tts_engine?.predefined_voices_path, - "tts_engine.reference_audio_path": currentConfig.tts_engine?.reference_audio_path, "paths.output": currentConfig.paths?.output, - "audio_output.format": currentConfig.audio_output?.format, "audio_output.sample_rate": currentConfig.audio_output?.sample_rate + "server.host": currentConfig.server?.host, + "server.port": currentConfig.server?.port, + "tts_engine.device": currentConfig.tts_engine?.device, + "tts_engine.default_voice_id": currentConfig.tts_engine?.default_voice_id, + "paths.model_cache": currentConfig.paths?.model_cache, + "tts_engine.predefined_voices_path": currentConfig.tts_engine?.predefined_voices_path, + "tts_engine.reference_audio_path": currentConfig.tts_engine?.reference_audio_path, + "paths.output": currentConfig.paths?.output, + "audio_output.format": currentConfig.audio_output?.format, + "audio_output.sample_rate": currentConfig.audio_output?.sample_rate }; for (const name in fieldsToDisplay) { const input = serverConfigForm.querySelector(`input[name="${name}"]`); @@ -1368,7 +1488,12 @@ document.addEventListener('DOMContentLoaded', async function () { if (!files || files.length === 0) return; const originalButtonHTML = buttonToAnimate ? buttonToAnimate.innerHTML : ''; if (buttonToAnimate) { - buttonToAnimate.innerHTML = `Uploading...`; + buttonToAnimate.innerHTML = ` + + + + + Uploading...`; buttonToAnimate.disabled = true; } const uploadNotification = showNotification(`Uploading ${files.length} file(s)...`, 'info', 0); @@ -1476,7 +1601,5 @@ document.addEventListener('DOMContentLoaded', async function () { }); } - // Call fetchInitialData at the end of setup to kick everything off. - // Note: This calls initializeApplication internally. await fetchInitialData(); });