Table 2 CER [%] results of the ASR system trained by clean data and multi-condition training (MCT) without the enhancement
From: Adversarial joint training with self-attention mechanism for robust end-to-end speech recognition
ASR | Training | CER [%] | ||
|---|---|---|---|---|
Clean | Match | Unmatch | ||
Baseline | Clean | 14.0 | 60.0 | 61.6 |
MCT | 14.6 | 20.8 | 27.3 | |
Transformer | Clean | 8.0 | 35.6 | 37.1 |
MCT | 7.8 | 13.1 | 16.2 | |
Conformer | Clean | 6.5 | 32.6 | 31.7 |
MCT | 6.9 | 12.1 | 13.7 | |