VoiceStreamAI

Why we included this project

VoiceStreamAI is a self-hosted streaming transcription setup: a Python server takes audio over WebSocket, runs it through a voice-activity detector and a Whisper-based recognizer, and returns text in near real time instead of waiting for a whole clip to finish. The VAD and ASR components are wired together through factory and strategy patterns, so you can swap in a different detection model or transcription engine (Pyannote for detection, faster-whisper by default) and tune how audio chunks get buffered without rewriting the server. A bundled JavaScript client covers the browser side, and the Docker image simplifies GPU setup. If you want live captions, meeting notes, or voice controls running on your own hardware rather than through a commercial API, this is a solid reference implementation to build from.