TranscriptAI

TranscriptAI is a local and cloud-deployable audio transcription app with AI-generated notes, summaries, and action items.

Canonical backend

Use python_server_optimized.py as the only supported runtime path.

• Frontend: public/index.html, public/script.js, public/style.css
• Backend: python_server_optimized.py
• Legacy variants: server.js, python_server.py, and python_server_chunked.py

The legacy servers are kept for reference only. They are not the deployment target.

Local setup

Prerequisites

• Python 3.11.x recommended.
• An OpenRouter API key.
• ffmpeg is bundled through imageio-ffmpeg, so you do not need a separate system install for the default setup.

Install dependencies

cd /Users/jjohnson/Downloads/transcriptai
python3.11 -m pip install -r requirements.txt

Configure environment

Create a .env file with:

OPENROUTER_API_KEY=your_openrouter_key_here
TRANSCRIPTION_MODEL=openai/gpt-audio-mini
ANALYSIS_MODEL=openai/gpt-4.1-mini
PORT=50263

Run locally

python3.11 python_server_optimized.py

Open http://localhost:50263.

Tests

The Playwright tests expect the local server on port 50263.

npx playwright test

If you only want a quick smoke check, open the app and upload Fairfield Dr 16.m4a.

Render deployment

This repo includes a Render blueprint in render.yaml.

Service settings

• Runtime: Python
• Start command: gunicorn python_server_optimized:app --bind 0.0.0.0:$PORT
• Build command: pip install -r requirements.txt

Required Render secrets

• OPENROUTER_API_KEY
• Optional:
  • TRANSCRIPTION_MODEL (defaults to openai/gpt-audio-mini)
  • ANALYSIS_MODEL (defaults to openai/gpt-4.1-mini)
  • OPENROUTER_SITE_URL
  • OPENROUTER_APP_NAME

Deploy flow

1. Push the repo to a git host connected to Render.
2. Import the repo into Render using the blueprint in render.yaml.
3. Add OPENROUTER_API_KEY in the Render dashboard.
4. Deploy the web service.

Runtime files

• uploads/ holds temporary uploaded audio and can be cleared safely.
• test-results/, test-debug.png, and test-result.png are generated artifacts and are not part of the app.

Mac app notes

The desktop launcher and updater scripts already point at python_server_optimized.py.

• start_transcription_app.command
• TranscriptAI.applescript
• app_updater.py
• update_app.sh

File map

transcriptai/
├── python_server_optimized.py  # Canonical backend
├── public/
│   ├── index.html
│   ├── script.js
│   └── style.css
├── render.yaml                 # Render blueprint
├── requirements.txt            # Python dependencies
├── TranscriptAI.applescript
├── app_updater.py
├── update_app.sh
└── tests/