Fact · Languages and accents
17.2BLEU points
On the Fleurs read-speech test, Whisper large-v2 scored 17.2 BLEU when translating Urdu speech into English.
Checked against its source on October 17, 2026
In the source's own words
“Model Swedish Swahili Tamil Telugu Tajik Thai Turkish Ukrainian Urdu Uzbek Vietnamese Yoruba ... Whisper large-v2 35.3 7.2 9.2 12.5 14.5 16.1 26.6 29.4 17.2 6.0 20.4 1.4”
Robust Speech Recognition via Large-Scale Weak Supervision (full text), Alec Radford et al., OpenAI (arXiv), 2022, Table 14 (D.3.1 Fleurs, BLEU scores).
Open the original (DOI 10.48550/arXiv.2212.04356)- What it measures
- Fleurs is a read-speech benchmark. BLEU compares the machine's English output with reference translations, so higher is better. The score is for the paper's zero-shot translate task on the largest model in the table, and it is one column of the paper's table for the language.
- What it does not show
- Read speech in good audio, scored on one benchmark. The same model had a word error rate of 22.6 percent on Urdu transcription, so a translation can inherit those errors. Conversational speech and noisy recordings will score lower.