Identify speakers
Identify speakers labels each line of the transcript with the speaker it came from. It's off by default, and the app's own switch says why: it's slower with it on.
Like transcription itself, it runs entirely on your computer, and it's free on every plan.
How do I turn it on?
- Open Preferences (Cmd+, on Mac, Ctrl+, on Windows).
- Go to the Transcription pane.
- Turn on Identify speakers. New transcriptions will be labelled.
To label a video you've already transcribed, open Display → Re-transcribe… in the Transcript tab, expand Advanced settings, and turn speaker detection on for that run.
The speaker-detection assets download on demand, the same way a model does. If a build isn't available for your machine the switch says so rather than quietly doing nothing, and a transcript that can't be split keeps a single speaker instead of failing.
What do I get back?
Speakers, not identities. They arrive as anonymous labels — Speaker 1, Speaker 2 — with a colour each. Naming them is the useful work, and it's yours to do.
Open the speakers popover in the transcript toolbar to:
- Rename a speaker. The name follows through the transcript, the exports and the printed storyboard.
- Merge two speakers, when one person was split across two labels.
- Add a speaker that detection didn't find.
- Delete one, hide one from the view, or change its colour.
What if two people ended up under one speaker?
Reassign the words. Drag across the words that belong to someone else and assign them to the right speaker: StoryFolder splits the segment at word boundaries rather than reassigning the whole line.
This is the failure worth planning for. Two people talking in the same scene can land under a single automatic speaker, and no amount of renaming fixes that on its own — the words have to move.
How accurate is it?
Good enough to be worth correcting, not good enough to trust unread. StoryFolder runs a local speaker-diarization engine over the audio: it needs no Python, no PyTorch and no GPU, and it runs as its own process rather than inside the app.
No accuracy figure is published for it, and none should be inferred. Overlapping speech, similar voices and noisy location sound are where you'll be doing the correcting.
Do speaker names appear in exports?
Yes. Speaker names print in SRT and WebVTT subtitles once there's more than one speaker, and in the plain text, CSV and Fountain outputs. See Searching and exporting a transcript.
The Statistics tab also carries a transcript block: total lines, total words, and each speaker's share.