Speech Understanding on X-ARES benchmark leaderboard · SOTA2 Research

Speech Understanding on X-ARES

80.56CREMA-D ScoreDashengTokenizer

Evaluation Results

Method	Links
DashengTokenizer2026.02		80.56	83.39	79.98	68.71	98.06	83.06	97.35	23.92	93.15	51.16	93.6
Dasheng 0.6B2026.02		79.39	97.79	69.67	69.64	98.58	78.13	97.22	29.67	92.21	84	86.33
SPEAR-XLarge2026.02		78.41	98.2	69	88.43	98.73	76.59	97.3	31.28	92.62	79.42	90.49
Dasheng-Base2026.02		77.2	94.38	68.8	72.4	98.47	72.5	96.7	25.05	91	77.98	81.29
SPEAR-Base2026.02		72.9	96.9	62.4	90	98.6	68.8	97.1	29.1	90.2	65.5	86.1
Whisper-Large-V32026.02		71.32	97.78	64.4	90	94.93	68.47	97.75	29.3	91.48	24.76	97.38
MSM-MAE2026.02		68.64	0.47	71.22	0	97.58	66.39	92.77	0.34	87.56	71.64	66.63
MingTok-Audio (Unified)2026.02		66.85	98.58	62.92	93.45	97.1	57.08	96.14	20.64	88.59	35.28	77.44
OpenBeats2026.02		65.9	58.5	69.9	0	97.3	63	94.4	21.4	87.7	0	0
SemantiCodec2026.02		60.3	46.4	66.05	0	98.14	55	88.02	23.82	84.28	60.99	26.66
HuBERT2026.02		58.06	98.73	49.72	82.45	85.94	47.71	95.67	18.82	81.71	3.39	45.06
FlexiCodec2026.02		58	99	43	0	0	50	0	0	79	0	72
WavLM2026.02		45.16	96.07	50.37	67.06	75.95	32.57	89.77	10.71	71.19	4.51	40.89
DAC2026.02		43.9	2.55	43.93	0	91.58	36.45	15.5	2.23	42.17	13.24	7.96
MingTok-Audio (Acoustic)2026.02		41.22	2.82	39.42	0	86.91	30.83	20.34	2.45	38.63	6.97	9.88
UniFlow-Audio2026.02		39.78	1.32	37.15	0	74.57	23.26	11.7	1.37	34.73	6.18	6.34
Encodec2026.02		39.16	3.48	33.78	0	88.44	27.71	26.7	2.16	43.06	10.92	7.89
EzAudio2026.02		38.8	1.63	37.19	0	84.9	25.56	12.47	1.21	35.26	5.99	6.96
WavTokenizer2026.02		33	2	38	0	0	27	0	0	27	0	9