Skip to content
Pascal Schindlmeier
DE
Back to all projects

Transcription & Speaker Diarization Pipeline

Problem
Interviews needed to be transcribed and attributed to individual speakers, which is extremely time-consuming to do by hand.
Approach
Built a pipeline combining WhisperX (transcription with word alignment) and pyannote.audio (speaker diarization), exporting a structured, searchable transcript.
Outcome
Significant time savings over manual transcription, publicly available.