A self-hosted Whisper service and web frontend for live speech-to-text. Streaming transcription with speaker recognition, and a summary once the session ends. It runs on my own GPU with faster-whisper and pyannote, so nothing goes to the cloud.
Tool · 2026 · in development
MeetingScribe
Live speech-to-text with full speaker recognition.
- Role
- Solo developer
- Tech
- Python · CUDA · faster-whisper · pyannote · Claude/Gemini APIs