| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Automatic Speech Recognition | LibriSpeech clean (test) | WER0.98 | 1,156 | |
| Automatic Speech Recognition | LibriSpeech (test-other) | WER1.6 | 1,151 | |
| Automatic Speech Recognition | LibriSpeech (dev-other) | WER2.1 | 462 | |
| Automatic Speech Recognition | LibriSpeech (dev-clean) | WER (%)1.06 | 340 | |
| Speech Recognition | Librispeech other (test) | WER2.2 | 105 | |
| Automatic Speech Recognition | LibriSpeech (other) | WER2.42 | 96 | |
| Automatic Speech Recognition | LibriSpeech 960h (test-other) | WER2.5 | 88 | |
| Automatic Speech Recognition | LibriSpeech (test-clean) | WER0.6 | 84 | |
| Automatic Speech Recognition | LibriSpeech clean | WER1.16 | 80 | |
| Speech Recognition | LibriSpeech clean (dev) | WER0.0123 | 80 | |
| Speech Recognition | LibriSpeech (test) | WER0.0133 | 76 | |
| Text-to-Speech | LibriSpeech clean (test) | WER1.5 | 66 | |
| Adversarial Example Detection | LibriSpeech C&W attack vs. benign | AUROC99 | 60 | |
| Automatic Speech Recognition | LibriSpeech 960h (test-clean) | WER0.016 | 60 | |
| Speech Reconstruction | LibriSpeech (test-clean) | UT MOS4.23 | 59 | |
| Speech Reconstruction | LibriSpeech English (test-clean) | SIM0.97 | 54 | |
| Speech Recognition | LibriSpeech LS-Ave | WER6.23 | 51 | |
| Automatic Speech Recognition | LibriSpeech 960h (dev-other) | WER2.4 | 50 | |
| Automatic Speech Recognition | LibriSpeech Psychoacoustic attack | WER0 | 48 | |
| Automatic Speech Recognition | LibriSpeech C&W attack | WER0 | 48 | |
| Automatic Speech Recognition | LibriSpeech 100h (test-clean) | WER2.4 | 43 | |
| Automatic Speech Recognition | LibriSpeech | WER2.99 | 35 | |
| Voice recognition | Librispeech | WER2.7 | 34 | |
| Text-to-Speech | LibriSpeech PC clean (test) | WER1.3 | 31 | |
| Automatic Speech Recognition | LibriSpeech other Speech Noise - Reverb (test) | WER30.1 | 28 |