Transcriber

Transcripts you can correct, revisit and reuse.

Audio
FastAPI
Revision history
Review workflows

I built Transcriber as a recording library with synchronized playback, editable transcripts and version history.

Review alongside the recording

Users can move between the transcript and the corresponding audio, correct text and timing, and adjust speaker labels. I integrated ElevenLabs Scribe for transcription and pyannote for voice identification.

A voice library lets users register samples to recognize participants in future recordings and find their contributions in the existing archive. Matches are proposed for review and correction.

Transcriber’s synchronized audio and transcript editor — Demonstration data.A selected view of the editor: timed words, speaker identities and corrections remain connected to the recording.

Keep corrections and their history

Version history preserves the original transcript and subsequent corrections. Reprocessing a recording lets users compare a new transcript without overwriting their review work.

Reuse reviewed transcripts

Transcripts can be exported as text, Markdown, subtitles or structured data. An API and MCP connections let other applications and AI assistants work with recordings and specific transcript revisions, keeping the material they use connected to its source.