Speech Recognition with Word-level Timestamps on Industrial Data
51.5AAS (ms)Qwen-Audio
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen-Audiozero-shot=true2023.11 | 51.5 | |
| Force-alignergiven ground-truth transcripts=true2023.11 | 60.3 | |
| Paraformer-large-TPTask-specific fine-tuning=true2023.11 | 65.3 |