1.0.2
Improves caption timing by prioritizing the latest visible caption over stale queued speech.
- Keeps at most one caption in flight and one latest pending caption, discarding pending text after 1.5 seconds to prevent an expanding speech backlog.
- Uses the first visible caption after activation, caption discovery, or seeking as a baseline so older roll-up text is not spoken as new.
- Replaces revised, unspoken live-caption text instead of forcing the superseded recognition result to speak.
- Reduces the maximum unpunctuated caption buffer from 1.5 seconds to 0.9 seconds and removes the additional 450-millisecond speech gate.
- Very rapid caption updates may skip intermediate fragments so speech can stay closer to the current video position.
SHA256:
617ac9815a93901b076edae3eef9b4a6389a8bfc79fca47667c086ff9b13c7d4 youtubeCaptionReader-1.0.2.nvda-addon