Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
(AudioStream, List<TranscriptSegment>) -> List<WordWithTimestamp>
Align a text transcript with raw audio using a phoneme-level ASR model to resolve precise start and end boundaries for each word.
Problem it solves
Native sequence-to-sequence ASR models produce coarse, utterance-level timestamps that often drift or lack word-level resolution.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.