Expand description
Word timing input: TimedWord in OUTPUT time, plus engine adapters.
Structs§
- Timed
Word - One spoken word (
end_msexclusive).
Constants§
- MAP_
DURATION_ TOLERANCE_ MS - A mapped word may differ from its source duration by at most this much; a larger difference means a removed range cut through the word, so it is dropped.
Functions§
- words_
from_ seconds - Engine adapter:
(text, start_seconds, end_seconds)-> milliseconds. Negative or non-finite times are clamped to 0; rounding is to nearest ms. This is the shape ofrightkit_asr::TimedWord/TimedToken. - words_
from_ transcript - Maps a source
Transcriptonto an output timeline (cut/edit).