Peer-reviewed research
End-to-end speaker segmentation for overlap-aware resegmentation
Hervé Bredin and Antoine Laurent, Interspeech 2021, 2021
Read the originalRead on October 17, 2026
- What it is
- A 2021 paper on a single neural model that finds speech, speaker turns and overlapped speech at once, instead of three separate steps. It works on short audio chunks at a fine time resolution.
- How much weight it carries
- Peer-reviewed at Interspeech 2021. It reports results on several corpora, and its time resolution describes the model's output frames, not the precision of every transcript timestamp.
Facts taken from this source
- Bredin and Laurent's 2021 segmentation model works on 5-second audio chunks at a temporal resolution of every 16 milliseconds.
“our model operates on short audio chunks (5 seconds) but at a much higher temporal resolution (every 16ms)” (Abstract)