Automatic Speech Recognition on CHiME-6 (eval)
30.7WERG-Augment
Evaluation Results
| Method | Links | |
|---|---|---|
| G-AugmentModel Scale=ConformerXXL, Pre-training Data Settings=Any2022.10 | 30.7 | |
| ConformerXXL-RNNT-PSPre-training=YT-U, Upstream=YT-T2021.09 | 31 | |
| Previous SOTAModel Scale=ConformerXXL, Pre-training Data Settings=Any2022.10 | 31 | |
| ConformerXXL-RNNT-PPre-training=YT-U, Upstream=SpeechStew2021.09 | 34.4 | |
| G-AugmentModel Scale=ConformerXXL, Pre-training Data Settings=Public2022.10 | 38.2 | |
| G-AugmentModel Scale=ConformerL, Pre-training Data Settings=Standard2022.10 | 38.5 | |
| HMM (SOTA)2021.09 | 38.6 | |
| ConformerXXLPre-training=Libri-Light, Upstream=SpeechStew2021.09 | 38.9 | |
| Previous SOTAModel Scale=ConformerXXL, Pre-training Data Settings=Public2022.10 | 38.9 | |
| ConformerXXL-RNNT-PPre-training=YT-U2021.09 | 39.5 | |
| Previous SOTAModel Scale=ConformerL, Pre-training Data Settings=Standard2022.10 | 40.6 | |
| WavLM + HuBERT (DCA)SSL Model=WavLM + HuBERT, Fusion Method=DCA, Total Params (M)=676.5, Trainable Params (M)=44.4, Encoder=E-Branchformer2026.04 | 47.5 | |
| WavLM + HuBERT (LP + FRL, epsilon=0.6)SSL Model=WavLM + HuBERT, Fusion Method=LP + FRL, epsilon (ϵ)=0.6, Total Params (M)=668.9, Trainable Params (M)=36.8, Encoder=E-Branchformer2026.04 | 49.3 | |
| WavLM + HuBERT (LP + FRL, epsilon=0.4)SSL Model=WavLM + HuBERT, Fusion Method=LP + FRL, epsilon (ϵ)=0.4, Total Params (M)=668.9, Trainable Params (M)=36.8, Encoder=E-Branchformer2026.04 | 49.3 | |
| WavLM + HuBERT (LP)SSL Model=WavLM + HuBERT, Fusion Method=LP, Total Params (M)=668.9, Trainable Params (M)=36.8, Encoder=E-Branchformer2026.04 | 49.6 | |
| WavLM + HuBERT (LP + FRL, epsilon=0.2)SSL Model=WavLM + HuBERT, Fusion Method=LP + FRL, epsilon (ϵ)=0.2, Total Params (M)=668.9, Trainable Params (M)=36.8, Encoder=E-Branchformer2026.04 | 49.6 | |
| WavLMSSL Model=WavLM, Total Params (M)=352.1, Trainable Params (M)=36.7, Encoder=E-Branchformer2026.04 | 50 | |
| WavLM + HuBERT (LP + FRL, epsilon=0.0)SSL Model=WavLM + HuBERT, Fusion Method=LP + FRL, epsilon (ϵ)=0.0, Total Params (M)=668.9, Trainable Params (M)=36.8, Encoder=E-Branchformer2026.04 | 51 | |
| HMM Baseline2021.09 | 51.3 | |
| WavLM + HuBERT (Co-Attention)SSL Model=WavLM + HuBERT, Fusion Method=Co-Attention [20], Total Params (M)=668.9, Trainable Params (M)=36.9, Encoder=E-Branchformer2026.04 | 57.4 |