Hello, I am working with Ch and would like to know if anyone have seen the list of sounds that is
supposed to be rocognized?
I think I have heard like 150 different sounds...
actually I am working with, Swedish, English, French and Spannish, so there aere defdenitively different ways to prounounce the letters....
Do I have to explain?
well. in english e is e and a is a and y is y and i is i
but then if I change to swedish....
the i is like the english e
and in spannish the a is like the english u in the word un..
If you want to know how the french prounounce the e? well it depends on if it has an accent grave or not etc.....
so then you say but it doesn't matter... because it is lot letters but yes it is....because when I make the layers names.... I have to write letters....
So back to the problem... I need to have a list of the letters that Ch can recognize in english, and if there are a list for all languages??? well better but if not I can do with the list of 150 sounds... so I can work from there...;
/ Mr. Spranglet
The Viseme layer names in Character Animator are hard coded. That is the set you get and nothing else. Think of them like phonemes rather than English. If that sort of a sound is heard (regardless of language), that mouth position is shown. So there is no concept of language translations.
This would be very different if it was text to speech conversion going on - but its not. It just listens to the sound and guesses the viseme based on that (and it gets it wrong at times). If you find languages where it works badly the Adobe team have asked previously in the forums for examples where it does not work (there were concerns around Chinese for example).
Oh, forgot to remind you also this is just to work out the mouth positions. So its not it needs to know the real phonetics - it just needs to have one image per mouth position (visual appearance - hence the name "viseme" rather than "phoneme"), so it does not have 150 sounds - it just has the given set of visual mouth positions. The code then maps sound to the right visual appearance.
Thanks for 2 answers! I don't know how to answer the both in one so I answer to just this first explanation.
I am OK with that it gets it "wrong" sometimes. (whom doesn't)
Well, so the sounds preprogrammed are the ones the puppets have? no more?
Actually it would be better for me if there where like 12-24 positions and I could chose the sound that goes with every position.
I guess it has been a work around hack. is that the consonant are mouth shape prounounced similarly worldwhide. making good consonants and don't worry to much about the ecatness in vovals....
some cartoons look moore like they are loops or random mouths anyway.
I have tried with making the mouth bigger the louder and it also works, and is easy to do with the scale-tool in Ae.
But as I have explained before the vovals doesn't.
Since I speak Swedish, English, French, Basque, Spanish and as a scandinavian also have knowledge in Norweigian, dannish and german. I can say that Dannish probably causes grave problems since they don't articule much. (sorry I weren't around when the question were asked during the work, I would have loved to help.
I have had that issue with a charachters voice in my work, if they don't articulate well, the Ch doesn't follow or better say it doesn't distinguish well and just "freezes" the mouth, (quite easy to fix by changing the letter in the time line, but if it is a 20 minute dialog as in my case it becomes long.
So maybe there is a possibility to make a sensitivity leveler as in sound sensitivity?
It is really a complexe thing but since i use Ch mainly for mouth synching it is for me the main problem with Ch.
I will try to have time to make a video to show how the problem with the different vovals looks in action.
Still mising the coplete list no matter how few the visemes are.
Thanks again and
have a nice day!
You can add extra mouth positions that you manually use via triggers instead of lipsync, but yes - slow and painful for long sections of talking!
The set of supported visemes can be found in almost every provided demo puppet, in the user guide (https://helpx.adobe.com/in/adobe-character-animator/using/behaviors-reference.html#Lip_Sync), or look in the tags section inside CH itself (it shows all the tag names possible over on the right). Google searches bring up lots of example images with the visemes as well.
I was hoping for more possibilities, but I guess I just have to do with that.
CH rules anyway, painful or not, once I got the hang of the things it is OK, but the King is to combine the CH with AE.
Thank you Adobe!
We all loves you!