Speaker diarization (speaker labels)

Diarization works out who spoke when in a recording and labels each stretch of speech with a speaker, such as Speaker 1 and Speaker 2.

  • Models and architecture
  • Updated

A plain transcript is one block of text. Diarization adds speaker labels so that a meeting, interview or call reads like a conversation. It can be a separate model run alongside the transcriber, or built into the speech model, as in Microsoft's VibeVoice-ASR and NVIDIA's Nemotron 3.5 Transcribe, which both return speaker labels.

The labels are anonymous: the model says that different people spoke, not who they are. Results get harder with overlapping speech, many speakers and poor audio, so listen to a sample of your own recordings before relying on it.

Related terms

  • Speech recognition (ASR, speech-to-text) Speech recognition, also called ASR or speech-to-text, turns spoken audio into written text.
  • Word error rate (WER) Word error rate is the standard score for speech recognition: the share of words in a transcript that are wrong, missing or extra compared with a correct reference.

Keep reading

All terms