QuentinFuxa/WhisperLiveKit: Real-time & local speech-to-text, translation, and speaker diarization. With server & web UI.

WhisperLiveKit provides real-time speech-to-text transcription with speaker identification that runs entirely locally. Built on state-of-the-art research including SimulStreaming and WhisperStreaming, it offers ultra-low latency transcription with voice activity detection. The toolkit includes a ready-to-use backend server, web UI, and supports multiple concurrent users with optional speaker diarization capabilities.