Fine-tuned Wav2Vec2 XLS-R 1B model for ASR in French
This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on French using the train and validation splits of Common Voice 11.0, Multilingual LibriSpeech, Voxpopuli, Multilingual TEDx, MediaSpeech, and African Accented French on 16kHz sampled speech audio. When using the model make sure that your speech input is also sampled at 16Khz.
Genrally we advise to use bofenghuang/asr-wav2vec2-ctc-french because it has the smaller model size and the better performance.
- Downloads last month
- 7
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Datasets used to train bofenghuang/asr-wav2vec2-xls-r-1b-ctc-french
Evaluation results
- Test WER on Common Voice 11.0self-reported14.800
- Test WER (+LM) on Common Voice 11.0self-reported12.610
- Test WER on Multilingual LibriSpeech (MLS)self-reported9.390
- Test WER (+LM) on Multilingual LibriSpeech (MLS)self-reported8.060
- Test WER on VoxPopuliself-reported11.800
- Test WER (+LM) on VoxPopuliself-reported9.940
- Test WER on African Accented Frenchself-reported22.980
- Test WER (+LM) on African Accented Frenchself-reported20.730
- Test WER on Robust Speech Event - Dev Dataself-reported17.880
- Test WER (+LM) on Robust Speech Event - Dev Dataself-reported14.010