Wednesday, November 20, 2013

"Text To Speech for Video", new open source program

I am pleased to announce that Text to Speech for Video is now available. Having been rather disappointed in the quality, price, and licensing conditions of many of the text to speech products in existence, I decided to create my own free and open source alternative specifically intended for creators of animated videos who want an easy to use way of generating speech for their characters.  Below is a short video demonstrating what the output sounds like:

Right now there are two voices available - one male with a southern U.S. accent (demonstrated in the video above) and one female with an asian accent (I'd describe it as Indian subcontinent).  Each voice has about 1500 words now (including some duplicates with different emotiveness - questioning, emphatic, etc.).  I hope to add more words to the existing voices and more voices as well in the coming months/years.

It's also quite easy (if you have the patience) to record your own "voice". Since each word is stored as a separate wav file (22 khz, 2 channel stereo), the voices take up quite a lot of room (program plus the two voices currently about 80mb which will grow over time as more words/voices are added).  The upside is that, in contrast to many synthetically generated voices, the output can sound a lot like real speech (since it is real speech, at least on a word for word basis). *


Bill Meisel said...

Ted, I'll mention this in my newsletter, Speech Strategy News. I'd like to give you credit by name (more than Ted the Blogger). If OK, send to
- Bill Meisel

TedTheBlogger said...

Thanks Bill!

Just sent you an email.