List your company
ResearchFormerly WizWand

Automatic Speech Recognition on Valibel corpus (test)

0WERTEXT SEQ2LABEL
-1.2567.22215.724.178Jun 18, 2024
Updated 8mo ago

Evaluation results

MethodLinks
TEXT SEQ2LABELInput=Text (Gold), Parameters=110M + 39.2M, Pre-training=CamemBERT2024.0600
TEXT GRAPHInput=Text (Gold), Parameters=110M + 41.4M, Pre-training=CamemBERT2024.0600
TEXT HOPSInput=Text (Gold), Parameters=110M + 100M, Pre-training=CamemBERT2024.0600
PIPELINE SEQ2LABELInput=Pipeline, Parameters=314M + 110M + 39.2M, Pre-training=Wav2vec2 + CamemBERT2024.0630.518.2
PIPELINE GRAPHInput=Pipeline, Parameters=314M + 110M + 41.4M, Pre-training=Wav2vec2 + CamemBERT2024.0630.518.2
PIPELINE HOPSInput=Pipeline, Parameters=314M + 110M + 100M, Pre-training=Wav2vec2 + CamemBERT2024.0630.518.2
AUDIO GRAPHInput=Audio, Parameters=315M + 34.9M, Pre-training=Wav2vec22024.0630.618.2
ORACLE SEQ2LABELInput=Oracle, Parameters=315M + 34.9M, Pre-training=Wav2vec22024.0630.918.6
AUDIO SEQ2LABELInput=Audio, Parameters=315M + 34.9M, Pre-training=Wav2vec22024.063118.4
ORACLE GRAPHInput=Oracle, Parameters=315M + 34.9M, Pre-training=Wav2vec22024.0631.419.2