Research at David AI

David AI’s DAI-ASR-I18N benchmark measures how well automatic speech recognition systems transcribe and diarize real, unscripted conversations.

Top eight models and 21 languages, ranked by average per-language rank for WER/CER. Results from the published benchmark.
All	AR	BN	DE	EN	ES	FR	HI	ID	IT	JA	KO	MR	PT	RU	TA	TE	TH	TL	TR	VI	ZH	System
9.47	9.61	15.99	6.37	4.20	4.92	7.23	8.72	8.91	7.09	5.60	4.10	11.63	6.07	8.09	25.62	28.84	5.32	12.53	6.84	6.81	4.35	MAI-Transcribe-2
14.22	19.35	19.80	7.42	2.74	5.21	12.58	17.88	14.95	9.85	6.84	6.18	22.13	8.27	15.94	40.93	44.31	7.21	16.35	7.25	9.23	4.26	Scribe v2
9.28	16.04	No result	7.96	3.91	5.65	11.16	11.12	11.30	10.45	6.38	4.43	No result	8.07	10.34	No result	No result	7.02	15.67	14.29	8.42	5.47	Grok Voice Transcribe 2.0
13.68	23.31	18.34	10.82	4.81	7.15	13.15	15.73	14.30	11.16	7.22	5.79	18.20	8.77	14.28	No result	37.84	10.08	17.72	14.67	10.93	9.32	Gemini 3.5 Transcribe
15.37	20.56	17.06	14.01	5.85	10.43	18.08	11.60	15.29	12.94	7.50	5.80	19.61	12.38	No result	32.70	46.72	8.85	17.01	14.47	11.26	5.35	Muse Voice Transcribe 1.0
16.10	22.12	24.23	13.71	5.78	10.05	15.79	13.17	15.76	12.78	7.82	6.99	20.97	11.21	15.65	33.25	39.46	10.37	20.65	17.02	13.78	7.59	GPT-Transcribe
26.08	26.35	66.79	9.88	3.70	7.86	14.34	89.09	17.84	10.94	9.78	7.08	35.55	11.51	14.08	29.97	35.31	10.89	75.93	48.60	14.17	7.93	Voxtral Small Transcribe 3.0
39.63	78.10	292.12	10.80	4.73	39.07	14.31	84.03	16.46	11.44	10.17	6.96	44.77	11.11	14.02	88.14	42.37	9.54	17.73	16.48	13.75	6.10	Voxtral Mini Transcribe 3.0

All Research