-
Notifications
You must be signed in to change notification settings - Fork 759
Open
Description
VoxPopuli publishes pre-trained models of many different languages under CC BY-NC 4.0 license. We can add them to torchaudio.
non-fine-tuned weights
https://github.com/facebookresearch/voxpopuli#wav2vec-20
- es - base
- es - large
- fr - base
- fr - large
- it - base
- it - large
- ni - base
- ni - large
- sv - base
- sv - large
- 23 langs (10k subset) - base
- 23 langs (10k subset) - large
- 23 langs (100k subset) - base
- 23 langs (100k subset) - large
Fine-tuned ASR
https://github.com/facebookresearch/voxpopuli#asr-and-lm
- cs
- de Add wav2vec2 ASR German pretrained model from voxpopuli #1953
- en Add wav2vec2 ASR English pretrained model from voxpopuli #1956
- es Add wav2vec2 ASR Spanish pretrained model from voxpopuli #1924
- et
- fi
- fr Add wav2vec2 ASR French pretrained model from voxpopuli #1919
- hr
- hu
- it Add wav2vec2 ASR Italian pretrained model from voxpopuli #1954
- lt
- ni
- pl
- ro
- sk
- sl
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels