Fact · Dictation
79.8macro F1 score
On English TED Talks transcripts, the best BERT models reached a macro-averaged F1 of 79.8 for punctuation restoration.
Checked against its source on October 16, 2026
In the source's own words
“Our best models achieve a macro-averaged ... score of 79.8 in English”
Automatic punctuation restoration with BERT models, Attila Nagy et al., arXiv, 2021, Abstract.
Open the original- What it measures
- Abstract of a 2021 preprint on restoring punctuation in text with BERT.
- What it does not show
- One macro average across marks, on written transcripts rather than live speech. It does not show per-mark results.