Streaming Automatic Speech Recognition on LibriSpeech other (test)
7.6WERTransducer
Evaluation results
| Method | Links | |
|---|---|---|
| Transducer2026.05 | 7.6 | |
| Chunkwise AlignerAlignment type=ground-truth, Delay=320ms2026.05 | 7.9 | |
| Chunkwise AlignerAlignment type=ground-truth, Delay=480ms2026.05 | 8.3 | |
| Chunkwise AlignerAlignment type=ground-truth, Delay=160ms2026.05 | 8.5 | |
| Chunkwise AlignerAlignment type=streaming CTC, Delay=0ms2026.05 | 9 | |
| Chunkwise AlignerAlignment type=ground-truth, Delay=0ms2026.05 | 9.5 | |
| CTC2026.05 | 10.8 |