Hacker News new | past | comments | ask | show | jobs | submit login
Large-scale multilingual audio visual dubbing (arxiv.org)
23 points by jonbaer on Nov 11, 2020 | hide | past | favorite | 2 comments

Definitely an ambitious project. Is there sample videos to evaluate?

I only skimmed this but I'm curious if this also kills the need for a voice actor or if it requires sample voice data from a native speaker to sound good (i.e. is it training from a corpus of attempts of the original actor trying to speak that language or does it just need some samples of different phonemes to train from?).

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact