Fact · Transcription
The WhisperX paper adds word-level timestamps to Whisper output, using voice activity detection and forced phoneme alignment.
Checked against its source on October 17, 2026
In the source's own words
“word-level timestamps utilising voice activity detection and forced phoneme alignment”
WhisperX: Time-Accurate Speech Transcription of Long-Form Audio, Max Bain et al., arXiv (cs.SD), 2023, Abstract.
Open the original- What it measures
- Forced alignment matches a known transcript to the audio so each word gets a start and end time. It is how a text transcript gains caption timing.
- What it does not show
- A research preprint. Word timing quality depends on the audio, the language and the alignment model.