hardware or software synths

[Samuel Thibault]

Jason White, le Wed 25 Jul 2012 00:36:57 +0000, a écrit :
> Kyle  <speakup at braille.uwo.ca> wrote:
> >As for new high quality open source voices, there is hope. Google TTS as used in Android 4.0 is available under the Apache license and uses BSD licensed
> >voice data provided by the Nagoya Institute of Technology. I'm not sure about the tools to create new voices, but this could become another option,
> >since I would guess that getting the voices running on Linux shouldn't be difficult. I am guessing that the latest Google voices should be just as free,
> >and from what I have heard, they sound even better, and are likely some of the best voices in terms of quality and responsiveness currently available So
> >yes, new open source speech synthesis tsoftware is being created, and the existing software will also be improved. 
> 
> Do you have details of the source repository for the new Google synthesizer?

git://android.git.kernel.org/platform/external/svox.git

> With luck, it should be relatively easy to check out the sources, and I'm sure
> there are people on the list who will be sufficiently interested to resolve
> any compilation issues.

It's not just compilation. It's reverse-engineering the lingware formats
to be able to convert from speech source into the binary format used by
the application (what is called "compilation"). Not impossible since we
have the source of the application reading the binary, and the speech
source is really source, not blobs, but quite involving, because that
means understanding the whole thing.

To give a bit of history, what happened is that google bought this
synthesizer from a company, but apparently they didn't get the source
for the compiler.

Samuel