Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Speaker Extraction research benchmarks · SOTA2 Research

Speaker Extraction

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

VoxCeleb2 mix Clean Visual

11.61SI-SNR (dB)

17

Jun 11, 2026

Three-speaker dataset (synthetic)

2.38PESQ

15

Feb 26, 2026

MMSpeech 3mix-noisy (test)

11.94SI-SDRi

6

Apr 28, 2026

MMSpeech 2mix-noisy (test)

12.16SI-SDRi

6

Apr 28, 2026

7.1SI-SNRi (dB)

6

Feb 26, 2026

YGD-3mix w/o missing subset

14.9SI-SNRi (dB)

6

Feb 26, 2026

YGD-3mix (test)

11.7SI-SNRi (dB)

6

Feb 26, 2026

10.7SI-SNRi (dB)

6

Feb 26, 2026

YGD-2mix w/o missing subset

15.1SI-SNRi (dB)

6

Feb 26, 2026

YGD-2mix (test)

13.2SI-SNRi (dB)

6

Feb 26, 2026

Popular datasets

YGD MMSpeech VoxCeleb Three-speaker dataset

Follow for updates

Join SOTA2 Contact