Speech-to-text Translation

Benchmarks

Dataset Name	SOTA Method	Metric
CoVoST low-resource X-to-En 2 (test)	Whisper	BLEU (Avg)25.2	24	1mo ago
MuST-C En-X (tst-COM)	SpeechUT	BLEU (German)30.1	16	1mo ago
MuST-C (En-Ru) 1.0 (test)	End-to-end (w2v2-mBART)	BLEU20	12	1mo ago
IWSLT25Instruct en-de	Hikari	BLEU37.67	10	1mo ago
FLEURS 9×27 directions (test)	MCAT-Large-27B	ara78.7	10	1mo ago
FLEURS supervised training languages (test)	POTSA (Our Model)	BLEU (en→zh)40.87	9	18d ago
CoVoST2 zh-en	APin	BLEU23.06	9	9d ago
FLEURS zero-shot evaluation languages (test)	POTSA (Our Model)	Average BLEU20.84	8	18d ago
FLEURS 9×69 directions (test)	MCAT-Large-27B	ARA Score79	8	1mo ago
CoVoST2 fr-en	Qwen-Audio	BLEU38.5	8	1mo ago
MuST-C En-Ro 1.0 (test)	Multi. T-Md	BLEU23.8	8	1mo ago
MuST-C En-Pt 1.0 (test)	Multi. T-Md	BLEU31.1	8	1mo ago
MuST-C En-It 1.0 (test)	Multi. T-Md	BLEU24.6	8	1mo ago
MuST-C En-Fr 1.0 (test)	Multi. T-Md	BLEU Score34.9	8	1mo ago
MuST-C En-Es 1.0 (test)	Multi. T-Md	BLEU28.2	8	1mo ago
MuST-C (En-Nl) 1.0 (test)	Multi. T-Md	BLEU28.6	8	1mo ago
MuST-C En-De 1.0 (test)	Multi. T-Md	BLEU24.5	8	1mo ago
CoVoST-2 high-resource X-to-En (test)	XLS-R (2B) + mBART-ML50N1	Quality Score (Fr)37.6	8	1mo ago
Tatoeba 4392 utterances		SacreBLEU (Fr)30.7	6	1mo ago
IWSLT25Instruct en-ja	Hikari	BLEU42.17	6	1mo ago
FLEURS (test)	Qwen3-Omni-30B-A3B-Instruct	ara86.6	5	1mo ago
MuST-C English-Spanish (test)	NAST-S2T	BLEU28.21	5	1mo ago
CoVoST2 en-zh	Qwen-Audio	BLEU41.5	5	8d ago
MuST-C En-Es (test)	End-to-end (w2v2-mBART)	BLEU0.324	4	1mo ago
IWSLT Gle-en 2025 (eval)	GDPS	BLEU40.2	3	19d ago

Showing 25 of 35 rows