Speaker Diarization

Turkish equivalent: Konuşmacı bölümlemeDomain: Speech Processing

The task of determining when different speakers are active in an audio recording, often summarized as answering who spoke when.

Boundaries

Speaker diarization answers who spoke when, not necessarily who the real-world person is; speaker verification or identification is a separate biometric task.

  • Speaker Embedding
  • Voice Activity Detection
  • Clustering
  • Overlapped Speech