Fact · Dictation
76.7percent F1
On the reference (correct) transcript of the IWSLT 2011 English benchmark, the best reported comma F1 was 76.7%.
Checked against its source on October 16, 2026
In the source's own words
“Results on Reference test set for Comma were the best: 76.7% F1”
Evaluating OpenAI's Whisper ASR for Punctuation Prediction and Topic Modeling, Lucas Rafael Stefanel Gris et al., arXiv (book chapter), 2023, Results on the reference test set (Section on English results).
Open the original- What it measures
- The chapter's review of earlier English systems on TED talks, using reference text rather than recognized speech.
- What it does not show
- TED talks with a human transcript, not dictation. Per mark, not an average.