ContainersNVIDIADeepSeek-R1NVIDIALlama-3.1-Nemotron-70B-InstructNVIDIAPyTorch
CollectionsNVIDIAOmniverse Kit (FB)NVIDIADeepStream SDKNVIDIAOmniverse Kit App Streaming
ModelsNVIDIAStyleGAN3 pretrained modelsNVIDIAPeopleNetNVIDIATrafficCamNet
ResourcesNVIDIARiva Skills Quick Start
Helm ChartsNVIDIAGPU OperatorNVIDIANVIDIA NIM Operator
NVIDIA
NVIDIA
SSL En Conformer Large
Model
NVIDIA
NVIDIA
SSL En Conformer Large

Self-Supervised Learning (SSL) checkpoints for Conformer Large model. These are similar to w2v-Conformer model and can be fine-tuned for Automatic Speech Recognition (ASR).

2 Versions
1.10.1Selected
05/26/2022 11:26 PM UTC616.77 MB
Accuracy Librispeech-100h
KeyValue
Librispeech test-other6.17 %
Librispeech dev-other6.08 %
Librispeech dev-clean3.04 %
Librispeech test-clean3.17 %
Accuracy Librispeech-960h
KeyValue
Librispeech Test-other4.20 %
Librispeech Dev-other4.33 %
Librispeech Dev-clean1.91 %
Librispeech Test-clean1.96 %
Model
KeyValue
Encoder Dimension512
Number of Encoder Layers17
DatasetLibrilight
ARCHITECTUREConformer
INPUTS16000 KHZ MONO-CHANNEL AUDIO (WAV FILES)
Number of Weights120M
1.10.0
05/26/2022 8:56 PM UTC439.99 MB
Accuracy Librispeech-100h
KeyValue
LibrispeechTestOther6.39 %
LibrispeechDevClean3.25 %
LibrispeechDevOther6.64 %
LibrispeechTestClean3.19 %
Accuracy Librispeech-960h
KeyValue
LibrispeechTestOther4.50 %
LibrispeechDevClean1.92 %
LibrispeechDevOther4.38 %
LibrispeechTestClean2.02 %
Model
KeyValue
Encoder Dimension512
Number of Encoder Layers17
DatasetLibrilight
ARCHITECTUREConformer
INPUTS16000 KHZ MONO-CHANNEL AUDIO (WAV FILES)
Number of Weights120M