Fact · Dictation
63.4percent F1
On speech recognition output from the IWSLT 2011 English benchmark, the best reported systems reached 63.4% F1 for question marks.
Checked against its source on October 16, 2026
In the source's own words
“Results on ASR test set for Period were the best (82.3%); for Comma and Question it was obtained 66.3% and 63.4%, respectively.”
Evaluating OpenAI's Whisper ASR for Punctuation Prediction and Topic Modeling, Lucas Rafael Stefanel Gris et al., arXiv (book chapter), 2023, Results on the ASR test set (Section on English results).
Open the original- What it measures
- The chapter's review of earlier English systems on TED talks, comparing reference text with ASR output.
- What it does not show
- TED talks, not dictation. The figure is the best of several earlier systems and is per mark.