Redenlab's Automatic Audio Pre-processing Pipeline

Speaker Diarisation

Identify who spoke when, and tidy it up in minutes.

How it works

  1. Detect speakers: We automatically estimate how many distinct speakers are in your audio. (Manual override is coming in a future release.)
  2. Create segments: Using that count, the diarization engine marks time-stamped speaker boundaries and exports a TextGrid with tiers compatible with Praat.
  3. Review & edit: Open the audio with its TextGrid to adjust boundaries and label tiers with speaker names or roles.

What you get

  • Time-aligned speaker segments
  • Praat-ready TextGrid tiers
  • Editable labels for names/roles

Tips

  • Clearer recordings improve speaker separation.
  • After exporting, fine-tune cuts and labels directly in Praat for best results.