Voice-to-JIRA (MVP)

Turn short voice notes into JIRA-ready tickets in under 30 seconds.

✨ What it does (MVP)

• Upload .wav or .mp3
• Transcribe audio → text (placeholder function)
• Convert text → JIRA ticket fields (title, description, priority) with an LLM (placeholder)
• Copy/paste the ticket into JIRA (direct API push is a future step)

This repo ships with placeholders so you can run it locally without any paid APIs. When ready, swap transcribe_audio() and draft_ticket_with_llm() with Whisper/OpenAI/Gemini.

🏗 Tech Stack

• Python 3.10+
• Streamlit
• pydub, soundfile (optional helpers)

🚀 Quickstart

# 1) Create env
python -m venv .venv && source .venv/bin/activate

# 2) Install deps
pip install -r requirements.txt

# 3) (Optional) set environment variables for real APIs later
cp .env.example .env
# then edit .env with your keys

# 4) Run
streamlit run app.py

🔑 Environment Variables (optional for real integrations)

Create a .env file:

OPENAI_API_KEY=sk-...
GOOGLE_APPLICATION_CREDENTIALS=/path/to/creds.json
JIRA_BASE_URL=https://your-org.atlassian.net
JIRA_EMAIL=you@company.com
JIRA_API_TOKEN=...

🔌 Where to add real implementations

• src/transcription.py → replace transcribe_audio() with Whisper or GCP Speech-to-Text
• src/ticket_llm.py → replace draft_ticket_with_llm() with OpenAI/Gemini call + prompt
• src/jira_format.py → adjust output JSON/markdown for your JIRA template
• src/jira_push.py → implement direct JIRA API push (OAuth/API token)

🧪 Suggested test phrases

• "Add retry with exponential backoff for payment POST in checkout service. Priority medium. Assign to me."
• "Bug: App crashes when uploading PNG over 10MB on iOS 16. Steps to reproduce... set priority high."
• "Create spike to evaluate PostHog vs Amplitude for product analytics; due next sprint."

🛣 Roadmap

• v0.1: Upload → placeholder transcript → placeholder ticket
• v0.2: Whisper + real LLM
• v0.3: JIRA OAuth + push button
• v0.4: Standup mode (split one voice note into multiple tickets)

📄 License

MIT