Local Notes Transcribe

A powerful local audio note transcription and processing system with dual model support (Whisper and Parakeet MLX) and fine-tuned LLM integration.

Features ✨

• Dual Transcription Models: Switch between Whisper (accuracy) and Parakeet MLX (speed)
• Fine-tuned LLM Integration: Uses your custom model via LM Studio for rich structured output
• Aesthetic HTML Cards: Beautiful, interactive cards displaying all extracted data
• Enhanced Database: Stores comprehensive structured data including action items, key points, entities
• Notes Browser: Web interface to browse and view all your processed notes
• Batch Processing: Process existing transcripts with the enhanced pipeline
• Flexible Configuration: Easy model switching via config file

Installation

# Clone the repository
git clone https://github.com/bilawalriaz/local-notes-transcribe.git
cd local-notes-transcribe

# Create and activate virtual environment
python -m venv packages
source packages/bin/activate  # On Windows: packages\Scripts\activate

# Install dependencies
pip install -r requirements.txt

Quick Start

1. Configure Your Environment

# Copy the configuration template
cp config_template.py config.py

# Edit config.py to choose your transcription model:
# TRANSCRIPTION_MODEL = "whisper"  # More accurate, slower
# TRANSCRIPTION_MODEL = "parakeet"  # Faster, Apple Silicon only

2. Start LM Studio (if using fine-tuned model)

• Load your fine-tuned model in LM Studio
• Ensure it's running on http://localhost:1234

3. Process Audio Notes

python enhanced_processor.py

Drop audio files into notes/input/ to process them automatically.

4. Browse Your Notes

python notes_browser.py

Access your notes at http://localhost:8000

Model Selection

Whisper

✅ More accurate for complex speech
✅ Better language detection
✅ Wider language support
❌ Slower transcription
❌ Larger memory footprint

Parakeet MLX

✅ Much faster transcription
✅ Direct audio file support
✅ Lower memory usage
✅ Better for quick notes
❌ Less accurate for complex speech
❌ Primarily English focused
❌ Requires Apple Silicon Mac

Configuration

See config_template.py for detailed configuration options including:

• Transcription model selection
• Model-specific settings
• LM Studio integration
• Directory paths

Changelog

See CHANGELOG.md for a list of changes and version history.

License

This project is licensed under the MIT License - see the LICENSE file for details.