In many cases, subtitles on dubbed content does not match. Often the dubbed speech does not match the actor's original speech. And, many times the nuances and differences in languages don't match up. Even the dubbed speech and subtext is out of sync with the image frames. What would really help here are few things:
- Quality subtext to match original spoken language
- Actors original voice matched and morphed to the translated dubbed language
- Sync the translations with the image key frames
- This will require video and speech recognition models that have quality training datasets