Automatic Speech Recognition on OpenASR leaderboard (test) benchmark leaderboard · SOTA2 Research

Automatic Speech Recognition on OpenASR leaderboard (test)

1.28LS-clean WERQwen3-Omni-30B-A3B-Instruct

Evaluation Results

Method	Links
Qwen3-Omni-30B-A3B-Instruct2026.07		1.28	11.44	10.47	8.79	2.52	2.44	2.78	6.07	5.72
Kimi-Audio-Instruct2026.07		1.28	18.19	12.25	10.25	2.42	3.53	2.9	8.06	6.62
Qwen3.5-Omni Flash2026.07		1.3	—	—	—	2.43	—	—	—	—
Phi-4-MM + FT2026.06		1.32	11.16	9.57	9.45	2.95	1.7	2.7	6.03	5.61
Audex 30B-A3B2026.07		1.34	17.24	11.92	9.9	3.06	1.76	3.5	5.83	6.82
Voxtral-Small-24B-25072026.07		1.59	15.27	10.5	9.81	3.26	2.02	3.52	6.96	6.62
Canary Qwen 2.5B2026.07		1.61	10.19	10.45	9.43	3.1	1.9	2.71	5.66	5.63
Phi-4-MM + Random Enc FT2026.06		1.62	12.19	14.31	10.38	4.27	2.04	3.29	7.65	6.97
Audex 2B2026.07		1.63	17.1	11.95	10.86	3.77	2	3.64	6.17	7.14
Step-Audio-R1.1 33B2026.07		1.66	22.36	11.54	9.97	3.28	3.8	3.03	7.6	7.91
Phi-4-MM2026.06		1.68	11.69	10.16	9.78	3.83	3.13	2.9	5.91	6.14
Mel-LLM (Phi-4-MM init)2026.06		1.7	12.91	12.99	10.95	4.83	2.28	3.55	7.76	7.12
Qwen3-Omni-30B-A3B-Thinking2026.07		1.77	25.19	10.51	9.49	3.7	3.47	2.38	7.52	8
Mel-LLM (Random init)2026.06		1.83	13.65	11.98	11.38	5.5	2.47	4.42	8.25	7.44
Parakeet-TDT-0.6B-v32026.07		1.92	11.39	11.19	9.57	3.59	3.98	2.8	6.09	6.32
Whisper-Large-V32026.06		2.01	15.95	11.29	10.02	3.91	2.94	3.86	9.54	7.44
Whisper-large-v32026.07		2.01	15.95	11.29	10.02	3.91	2.94	3.86	9.54	7.44
Canary-1B-v22026.07		2.18	16.01	11.79	10.82	3.56	2.28	4.29	6.25	7.15
MiMo-Audio2026.07		3.5	—	—	—	—	—	—	—	—