Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

NOTSOFAR

Benchmarks

Task NameDataset NameSOTA ResultTrend
Target-speaker Automatic Speech RecognitionNOTSOFAR Small-SDM 1
tcpWER15.8
14
Speaker DiarizationNOTSOFAR-1
DER8.2
9
Multi-speaker Automatic Speech RecognitionNotSoFar
CP-WER34.48
7
Turn-taking decisionNOTSOFAR NSF-1 (test)
Precision81
4
Showing 4 of 4 rows