www.si-community.com

Are there by any chance different viseme libraries for different languages?
And also how would you deal with Arabic for example where you need both visemes and text in file?
Any experiences or that is a bit too much for Facerobot?
Thanks

face robot can only recognize text in english or japanese, we don't have dictionaries for the other languages. If you give it a text that isn't written in english or japanese, you'll just get something random, of anything. I might also work, I've seen people give it french text. But arabic! heh... I'm curious, I expect a failure unless there is some kind of romanisation possible.

now do you need a viseme library for other languages.. I'm not sure that you do. We didn't do anything for japanese. phonemes are kind of an universal thing. We produced viseme for all the different types of phonemes that the speech recognition library can return, which is much, much smaller than the actual documented list of phonemes defined in, for example, the IPA unicode standard. What I'm getting at is that people get a very approximate result for an audio file that is "good enough" and they fudge it a little as necessary. It's not as smart and scientific as it sounds