NVIDIA
NVIDIA
STT Fr Conformer-Transducer Large
Model
NVIDIA
NVIDIA
STT Fr Conformer-Transducer Large

Conformer-Transducer-Large model for French Automatic Speech Recognition, Trained on French NeMo ASRSET

1 Version
1.5Selected
01/07/2022 7:13 AM UTC421.66 MB500 EpochsBatch Size: 8
ACCURACY (GREEDY)
KeyValue
MCV7.0 DEV6.85
MCV7.0 TEST7.95
MLS DEV5.05
MLS TEST4.10
MODEL
KeyValue
ARCHITECTURECITRINET
INPUTS16000 KHZ MONO-CHANNEL AUDIO (WAV FILES)
OUTPUTSTRANSCRIBED SPEECH