| WSJ (92-eval) | SOTA | WER1.3 | | 131 | 3d ago |
| Hub5'00 SWB (test) | ResNet+LSTM1+LSTM2 | WER6.7 | | 91 | 3d ago |
| LibriSpeech clean (dev) | Stateformer 25L | WER0.0176 | | 59 | 3d ago |
| LibriSpeech (test) | Adam | WER0.0133 | | 59 | 3d ago |
| WSJ nov93 (dev) | | WER3 | | 52 | 3d ago |
| LRS2 (test) | Whisper-Flamingo | WER1.3 | | 49 | 3d ago |
| CommonVoice | FAMCONN | WER25.49 | | 40 | 3d ago |
| VoxPopuli (test) | FC-XXL | WER5.39 | | 37 | 3d ago |
| LJ-Speech (test) | | WER3.01 | | 35 | 3d ago |
| WSJ nov92 (test) | Fast Conformer | WER1.42 | | 34 | 3d ago |
| MuST-C (test) | Specialized | WER (Avg)6.54 | | 30 | 3d ago |
| AISHELL-1 (dev) | Qwen-Audio | WER1.2 | | 28 | 3d ago |
| Wall Street Journal open vocabulary (dev93) | Deep Speech 2 | WER4.4 | | 28 | 3d ago |
| Hub5'00 CH (test) | n-gram + model M + NNLM | WER12.2 | | 28 | 3d ago |
| LibriSpeech 960hr (test) | w2v-Conformer XXL | WER1.4 | | 26 | 3d ago |
| LibriSpeech 960hr (dev) | wav2vec 2.0 | WER1.1 | | 25 | 3d ago |
| LRS3-TED | EG-seq2seq | WER7.2 | | 25 | 3d ago |
| Hub5'00 | System Combination (6+7+8+9) + TrafoXL LM | SWB Score4.3 | | 25 | 3d ago |
| LibriSpeech (dev) | FAdam | WER1.29 | | 21 | 3d ago |
| LibriSpeech (test-clean) | Phi-4-Multimodal + RLBR | BWER0.59 | | 20 | 3d ago |
| Librispeech other (test) | Phi-4-Multimodal + RLBR | BWER2.11 | | 20 | 3d ago |
| Fleurs En | | WER2.4 | | 20 | 3d ago |
| In-House dataset | G1 + KOO | CER0.0305 | | 19 | 3d ago |
| LRS3 high-resource | USR 2.0 | WER (V)17.6 | | 18 | 3d ago |
| LRS3 low-resource | USR 2.0 | WER (V)23.7 | | 18 | 3d ago |