Automatic Speech Recognition on CHiME-4 simu (test)
4.94WERWER-OA
Evaluation Results
| Method | Links | |
|---|---|---|
| WER-OAASR Model=Parakeet, SE System=Demucs2026.02 | 4.94 | |
| WER-OAASR Model=Whisper, SE System=Demucs2026.02 | 5.05 | |
| Noisy yASR Model=Parakeet, SE System=None2026.02 | 5.24 | |
| Noisy yASR Model=Whisper, SE System=None2026.02 | 5.36 | |
| Conf-OAASR Model=Parakeet, SE System=Demucs2026.02 | 5.4 | |
| Classifier-OA_3classASR Model=Parakeet, SE System=Demucs2026.02 | 5.48 | |
| Conf-SwitchASR Model=Parakeet, SE System=Demucs2026.02 | 5.58 | |
| Conf-OAASR Model=Whisper, SE System=Demucs2026.02 | 5.64 | |
| Classifier-OA_3classASR Model=Whisper, SE System=Demucs2026.02 | 5.7 | |
| Conf-SwitchASR Model=Whisper, SE System=Demucs2026.02 | 5.76 | |
| DNSMOS-OAASR Model=Parakeet, SE System=Demucs2026.02 | 6.66 | |
| DNSMOS-OAASR Model=Whisper, SE System=Demucs2026.02 | 6.82 | |
| IRTFpost-filter=Wiener2019.05 | 6.95 | |
| Classifier-OA_2classASR Model=Parakeet, SE System=Demucs2026.02 | 7.35 | |
| IRTFpost-filter=none2019.05 | 7.36 | |
| MVDRpost-filter=Wiener2019.05 | 7.47 | |
| GEVpost-filter=BAN2019.05 | 7.94 | |
| MVDRpost-filter=none2019.05 | 7.97 | |
| Classifier-OA_2classASR Model=Whisper, SE System=Demucs2026.02 | 8.16 | |
| Whisper (real label)Adaptation strategy=Finetuning, Supervision type=Supervised2024.05 | 8.7 | |
| STARAdaptation strategy=STAR, Supervision type=Unsupervised2024.05 | 9.4 | |
| TOKreweight (A_t)Adaptation strategy=Self-training, Token-level re-weighting=A_t, Supervision type=Unsupervised2024.05 | 9.5 | |
| UTTfilterAdaptation strategy=Self-training, Filtering=Utterance-level, Supervision type=Unsupervised2024.05 | 9.7 | |
| TOKreweight (G_t)Adaptation strategy=Self-training, Token-level re-weighting=G_t, Supervision type=Unsupervised2024.05 | 9.8 | |
| Whisper (frozen)Adaptation strategy=None, Supervision type=Zero-shot2024.05 | 9.9 | |
| Whisper (self-train.)Adaptation strategy=Self-training, Supervision type=Unsupervised2024.05 | 10.1 | |
| GEVpost-filter=none2019.05 | 10.27 | |
| Unprocessedpost-filter=none2019.05 | 10.79 | |
| WER-OAASR Model=Wav2Vec2, SE System=Demucs2026.02 | 21.91 | |
| Conf-OAASR Model=Wav2Vec2, SE System=Demucs2026.02 | 22.87 | |
| Classifier-OA_3classASR Model=Wav2Vec2, SE System=Demucs2026.02 | 23.08 | |
| DNSMOS-OAASR Model=Wav2Vec2, SE System=Demucs2026.02 | 25.2 | |
| Conf-SwitchASR Model=Wav2Vec2, SE System=Demucs2026.02 | 25.76 | |
| Noisy yASR Model=Wav2Vec2, SE System=None2026.02 | 25.82 | |
| Enhanced x_hatASR Model=Parakeet, SE System=Demucs2026.02 | 27.42 | |
| Classifier-OA_2classASR Model=Wav2Vec2, SE System=Demucs2026.02 | 28.31 | |
| Enhanced x_hatASR Model=Whisper, SE System=Demucs2026.02 | 28.69 | |
| Enhanced x_hatASR Model=Wav2Vec2, SE System=Demucs2026.02 | 48.07 |