Automatic Speech Recognition on SUPERB 22.05kHz
2.9WERMSRHuBERT
Evaluation results
| Method | Links | |
|---|---|---|
| MSRHuBERTFine-tuning protocol=Mixed-rate2026.03 | 2.9 | |
| Re. 16kHz HuBERT BaseFine-tuning protocol=Mixed-rate, Pre-training sampling rate=16kHz2026.03 | 2.95 | |
| Re. 24kHz HuBERT BaseFine-tuning protocol=Mixed-rate, Pre-training sampling rate=24kHz2026.03 | 3.02 | |
| Re. 16kHz HuBERT BaseFine-tuning protocol=Single-rate, Pre-training sampling rate=16kHz2026.03 | 3.15 | |
| Re. 24kHz HuBERT BaseFine-tuning protocol=Single-rate, Pre-training sampling rate=24kHz2026.03 | 3.28 | |
| Re. 48kHz HuBERT BaseFine-tuning protocol=Mixed-rate, Pre-training sampling rate=48kHz2026.03 | 3.29 | |
| HuBERT BaseFine-tuning protocol=Single-rate, Pre-training sampling rate=16kHz2026.03 | 3.34 | |
| HuBERT BaseFine-tuning protocol=Mixed-rate, Pre-training sampling rate=16kHz2026.03 | 3.35 | |
| MSRHuBERTFine-tuning protocol=Single-rate2026.03 | 3.35 | |
| Re. 48kHz HuBERT BaseFine-tuning protocol=Single-rate, Pre-training sampling rate=48kHz2026.03 | 3.41 | |
| Re. 16kHz HuBERT Base (w/o resampling)Fine-tuning protocol=Mixed-rate, Resampling in fine-tuning=No2026.03 | 3.72 | |
| Re. 16kHz HuBERT Base (w/o resampling)Fine-tuning protocol=Single-rate, Resampling in fine-tuning=No2026.03 | 4.47 |