Speaker diarization identifies and labels distinct speakers through the transcript — returned alongside the plain-text version, not instead of it.
Why a flat transcript often isn't enough
A single unbroken transcript of a deposition, a panel discussion, or a recorded interview is close to useless if you can't tell who said what. Reading back through audio to manually tag speakers doesn't scale past a handful of recordings.
What comes back
The transcript is segmented and labeled by distinct speaker — Speaker 1, Speaker 2, and so on — aligned to the same text you'd get from a plain transcription job, at the same per-hour pricing tier.
Where this breaks down — said plainly
- Speakers are labeled by number, not by name. Matching a label to an actual person is still a manual step on your end.
- Overlapping speech is the hardest case for any diarization system. Expect more correction needed where people talk over each other.
Related guides
- Transcription that doesn't store your audio in the cloud
- Caption training videos without uploading to YouTube
Where this fits on the site: this is the Who spoke when, in a recording service under Audio transcription & captions pricing — published tiers starting at $199 for up to 2 hours, no quote, no sales call. See Services & Pricing or go straight to Choose your job.