Fact · Languages and accents
1.3BLEU points
SeamlessM4T improved into-English speech-to-text translation by 1.3 BLEU points over strong cascaded systems.
Checked against its source on October 17, 2026
In the source's own words
“Compared to strong cascaded models, SeamlessM4T improves the quality of into-English translation by 1.3 BLEU points in speech-to-text”
SeamlessM4T: Massively Multilingual and Multimodal Machine Translation, Seamless Communication, Meta AI (arXiv), 2023, Abstract.
Open the original (DOI 10.48550/arXiv.2308.11596)- What it measures
- A cascaded system runs speech recognition first and then text translation, as separate models.
- What it does not show
- The comparison comes from the model's builders and is an average across languages. A 1.3 point gain is small, and it does not remove the error risk that comes from a wrong transcript.