Transcription App has two different ways to handle speakers. This tutorial explains both approaches and shows you how to use Speaker Recognition from the transcript toolbar.
Two speaker workflows
Speaker segmentation during upload or edit is configured on the media file itself. It requires Word-level timestamps to be enabled first. Use it when you want the transcript automatically broken into speaker turns right after transcription runs.
Speaker Recognition from the transcript toolbar is a later-stage tool you can run on an already-transcribed file. Use it when you already have a transcript and want the app to infer or refine speaker labels across it.
Speaker segmentation during upload
- When uploading or editing a media file, enable « Word-level timestamps ».
- Once word-level timestamps are enabled, the Speaker segmentation section becomes visible.
- Enable « Speaker segmentation ».
- Adjust Segmentation precision and Chunk length if needed.
- Complete the upload and transcribe the file normally.
Speaker Recognition workflow
- Open a completed transcript (the file must already be transcribed).
- In the transcript toolbar, click « Speaker Recognition ».
- Review the recognition settings.
- Run the analysis.
- Review the results when the analysis completes.
- Rename speakers manually if needed using the segment popup.
- Click the Save icon to save the transcript.
Tip: The app recommends labeling at least 3 segments per speaker before running Speaker Recognition if you want better matching. Running recognition resets existing speaker labels and applies new ones.
Other speaker tools in the transcript toolbar
- Text View — switches between segment view and speaker text view. In text view, TXT and DOC exports group consecutive same-speaker blocks.
- Merge Speakers — merges consecutive segments that share the same speaker label, after confirmation.
- Reset Speakers — clears all speaker labels and stored speaker-profile data, after confirmation.
Renaming speakers
- Click a segment to open the segment popup.
- Expand the Speaker section.
- Type the new speaker name.
- To rename every segment that currently shares the same old speaker label, use « Change all speakers ».
- Click « Save Segment ».
- Click the Save icon in the toolbar to save the transcript.
If Speaker Recognition is blocked
Speaker Recognition is blocked when the file is not transcribed yet, transcription is still running, or there is not enough usable transcript data. Wait for transcription to complete before running it.
Practical tips
- Decide on a speaker naming scheme early and stick to it (for example: Interviewer, Participant A, Participant B).
- Use « Change all speakers » to rename a label consistently across the whole transcript.
- Save often during heavy edits.