Fact · Transcription
200milliseconds tolerance (typical in prior work)
Earlier studies of word-level timestamps usually counted a timestamp as correct if it fell within a tolerance of more than 200 milliseconds.
Checked against its source on October 17, 2026
In the source's own words
“The evaluation in prior work is also relatively loose, typically using a tolerance of more than 200 ms.”
Whisper Has an Internal Word Aligner, Sung-Lin Yeh et al., arXiv (ASRU 2025), 2025, Abstract.
Open the original- What it measures
- A word timestamp 200 milliseconds off can still look correct under a loose tolerance, so earlier scores flatter the timestamps.
- What it does not show
- The authors describe prior work as a group. A tolerance is a choice of each study, so the figure applies to the studies they reviewed.