AI subtitles / implemented
See subtitles sooner.
Improve them in the background.
YOMUVIA shows a short draft around the current position first, then replaces it with a high-accuracy full-video result.
Speed and accuracy are separate passes
Pressing the subtitle button transcribes a short range around the playhead with a fast model. A high-accuracy model continues over the full video and replaces the draft without resetting playback.
Seeking or skipping into an uncovered section requests a new local draft around that position.
Use context to reduce mishearing
Filename hints, opening dialogue, genre, neighboring captions, and prior user corrections can help resolve names and natural vocabulary. If cloud assistance is unavailable, local Whisper output and repetition filtering still complete the subtitle flow.
Correct it where you watch it
- Pause and double-click the visible subtitle
- Store corrections separately from generated cache
- Reuse corrections for the same video or identified work
- Follow the Windows display language or choose manually
Privacy boundary
Audio extraction and Whisper recognition run on the PC. When configured cloud correction or translation is used, recognized text, the filename, and limited context may be sent to the AI provider. The video file itself is not uploaded by this flow.