NVIDIA
NVIDIA
VAD multilingual Marblenet
Model
NVIDIA
NVIDIA
VAD multilingual Marblenet

MarbleNet VAD model with multilingual data

1 Version
1.10.0Selected
08/02/2022 8:33 PM UTC490 KB50 EpochsBatch Size: 256
Accuracy
KeyValue
auroc on ALL0.9112
datasetava speech
Model
KeyValue
Architecturemarblenet
Outputsframe-level voice activity prediction
Inputs16000 KHz Mono-channel Audio (wav files)