Jojoe258Jojoe/30sAI_asr

30s AI is an application that is intended to transform unstructured to structured speech in real time and send the structured speech via a web socket to a customer service or support agent on a live call. It was developed by Tinyefuza Joe, Nabaccwa.M.Jema and Adam Katongole for the ASR hackathon Uganda 2026. # 30sAI_asr 30s AI is an application that transforms unstructured speech into structured text in real time and sends the transcription to a customer service or support agent over WebSocket. It was developed by Tinyefuza Joe, and Nabaccwa.M.Jema for the ASR hackathon Uganda 2026. ## What is in the repo - `services/modal_streaming_whisper.py` - real-time Whisper ASR service with `/ws`, `/transcribe`, and `/health` - `services/gpt2_service.py` - next-word prediction service - `services/tts_service.py` - text-to-speech service - `services/feedback_service.py` - stores corrections for later training - `client/index.html` - main browser UI - `client/browser_client.html` - minimal browser WebSocket client - `client/python_websocket_client.py` - CLI streaming client - `static/index.html` - static demo UI ## Prerequisites - Modal CLI installed and authenticated - Hugging Face secret configured for Whisper if required by your Modal account - Python 3.10+ for local client testing ## Deploy the endpoints Run these from the repository root: ```bash modal deploy services/modal_streaming_whisper.py modal deploy services/gpt2_service.py modal deploy services/tts_service.py modal deploy services/feedback_service.py ``` Or deploy everything with the helper script: ```bash ./scripts/deploy_all.sh ``` If this is your first time using Modal on the machine, run: ```bash modal setup ``` ## Run locally The Whisper service has a local entrypoint that prints the deployed URLs for the file upload and WebSocket endpoints: ```bash modal run services/modal_streaming_whisper.py ``` To open the browser UI locally, serve the repository over HTTP: ```bash python3 -m http.server 8000 ``` Then open `localhost` in your browser and paste the deployed service URLs into the Settings screen. ## Test the services ### 1. Test file transcription After running `modal run services/modal_streaming_whisper.py`, copy the printed file upload URL and send a WAV file …