VPanjeta / whisper-websockets

Websockets implementation with OpenAI whisper for real time speech recognition

Whisper-Websockets

Implementation of OpenAI Whisper library with websockets for real time ASR

This is a rudimentary implementation of Whisper ASR over websockets for test purposes only. Implementation for a production level system would require major refactoring and code sanitisation.

Installtion

• Install libraries by sudo apt install ffmpeg portaudio19-dev
• Install python pacakges by pip3 install -r requirements.txt

Run

• Run the websocket server with python3 server.py which starts to listen for incoming socket connections.
• Connect to the server with the client using python3 client.py.
• To change config params, check the client.py file for config object.

ToDo

• Configure arument parser for client to take config.
• User configurable stream input chunk window.
• Support for multiple connections.
• Implement real time language translation over websockets.