Fact · Languages and accents
The SeamlessM4T paper lists error propagation from speech recognition into the translation model as a known weakness of cascaded systems.
Checked against its source on October 17, 2026
In the source's own words
“an increased likelihood of error propagation from the ASR model to the T2TT model and other subsequent models in the cascade”
SeamlessM4T: Massively Multilingual and Multimodal Machine Translation (full text), Seamless Communication, Meta AI (arXiv), 2023, Discussion of cascaded systems in the introduction.
Open the original (DOI 10.48550/arXiv.2308.11596)- What it measures
- The same passage lists poor transcription for non-English languages, particularly low-resource ones, as a separate cause of meaning lost in cascades.
- What it does not show
- This is the authors' explanation of why cascades lose meaning. It describes a risk and does not give a measured error rate for any one language.