Automatic Speech Recognition on Common Voice Arabic (test)
38.21WERWSM
Evaluation results
| Method | Links | |
|---|---|---|
| WSMSSL Model=XLS-R, Version=Ours2026.02 | 38.21 | |
| WSMSSL Model=MMS, Version=Ours2026.02 | 39.08 | |
| XLS-RSSL Model=XLS-R, Version=Baseline2026.02 | 40.34 | |
| MMSSSL Model=MMS, Version=Baseline2026.02 | 42.53 | |
| WSMSSL Model=HuBERT, Version=Ours2026.02 | 48.63 | |
| HuBERTSSL Model=HuBERT, Version=Baseline2026.02 | 48.95 | |
| WSMSSL Model=wav2vec 2.0, Version=Ours2026.02 | 51.46 | |
| WSMSSL Model=data2vec, Version=Ours2026.02 | 51.95 | |
| WSMSSL Model=mHuBERT, Version=Ours2026.02 | 52.31 | |
| wav2vec 2.0SSL Model=wav2vec 2.0, Version=Baseline2026.02 | 53.82 | |
| mHuBERTSSL Model=mHuBERT, Version=Baseline2026.02 | 54.54 | |
| data2vecSSL Model=data2vec, Version=Baseline2026.02 | 54.8 |