I'm a developer for Chevron Phillips during the day hours and they get me a MSDN subscription every year for work. so Last year I decided to play around with Microsoft's Voice technology SDK. I was able to get the demos running and I could get the machine to say text from a text box. Not very impressive I know. I am however interested in doing what I can to get this type of lip sync up and going. I'm a programmer too and I know the feeling you get when a user request something completely out of the applications scope and you just have to agree with them on how good an idea it is and blah blah blah. So I'll admit that I'm not familiar enough with the voice technology to know whether or not I'm completely out of reach. I would love to help out if you could just point me in a direction. Here goes. I'm assuming that the better sounding voices are just collections of what you might call a sound font. These common sound parts are recorded and re-recorder with different inflections and then somehow mapped to basic rules of vocabulary so that the computer can essentially sound out a specific word.

1. Is there a tool that will allow me to record my own samples?
2. If so I could record sets off these "sound fonts", only I would record them multiple times based on mood.

3. This is where your software fits in... The text can be written like a marked up language. Sample:

<speech actor="M_Father" mood="calm"> How was school today son? </speech>
<speech actor="M_child" mood="sad"> Shucks, I don't know dad. </speech>
<speech actor="M_Father" mood="curious">Well what's wrong? </speech>
<speech actor="M_child" mood="sad">Well, I kind of got sent to the
principals office. </speech>
<speech actor="M_Father" mood="angry">what? That's the third time this
week</speech>

The intensity and variations in facial distortion as well as the sound samples would be determined by the mood attribute. The deformations etc are directed to the selected model in the scene using the actor attribute. Edit timing by adding <wait frames="120"/>

I could go on and on about group conversations everything Right now I'm just curious if you know of a voice sample tool. The other stuff is probably off the wall. You have a wonderful package I guess I'm just excited so that why I'm coming up with all of this stuff.

That's again.
db

I found this page, please take a look.
www.zabaware.com/forum/topic.asp?TOPIC_ID=1950

Since you have the MSDN library, gosh you lucky maybe you can check out the
"Microsoft speech application SDK"
This is a different bundle from the Microsoft Speech SDK. The Speech SDK you and i have both messed with, the Speech Application SDK is a whole other bird. I read that there is an application which helps you make voices. However, i went searching and I couldn't find it on microsoft's site anymore. The tool used to be called
"Microsoft speech application SDK Beta". Seems that with all the new .NET stuff, we may not be able to download it anymore. This Speech App SDK is not free, I found a page talking about a 60 day trial on Microsoft but the download page was gone

As far as a conversation in PoserSpeak, check out the manual for some of the XML tags which you can use in your script. You can pause with a statement like: <silence msec='500'/>
You can also change voices mid-stream too with a voice tag. PoserSpeak should be run per actor though with version 1. Currently, PoserSpeak doesn't marry two wav files together. It could do this and you've got me thinking on it however.

Thanks again db for your comments!
jtrue

"What's new from Loquendo:

Loquendo TTS now supports MRCPv1 and v2, as well as the new innovative Lexicon Editor tool, which uses phonetic mapping to automatically suggest the transcription of foreign words. We will be introducing a long-awaited feature: customisable Loquendo TTS voices - timbre, rate and pitch can be edited and saved to create a new, tailored persona – to better fit our customers’ needs."

I think this solution is kind of expensive, but itīs good to know that this exists.

You can read the full story at:
http://www.loquendo.com/en/news/news_voxnauta_7_0.htm

Regards,
Guile3D