Re: [gnuspeech-contact] introduction and gnuspeech questions

[D.R. Hill]

Hi Eric,

Many thanks for your email and your strong interest in helping with the 
"gnuspeech" porting work.

I have replied to your questions in context.

I will help you in any way I can.  Feel free to keep bugging me.  It will 
be a big help to have you improving and extending the port.  Thank you.

All good wishes.

david
> > > > >

David Hill, Prof. Emeritus, Computer Science  |  Imagination is more       |
U. Calgary, Calgary, AB, Canada T2N 1N4       |  important than knowledge  |
address@hidden OR address@hidden   |         (Albert Einstein)  |
http://www.cpsc.ucalgary.ca/~hill             |  Kill your television      |
----

On Mon, 14 Nov 2005, Eric Zoerner wrote:

> I am looking at gnuspeech and am considering getting involved in the porting 
> and/or development effort.  I am a postgraduate student at Trinity College, 
> Dublin, doing a PhD in speech science.  I have 17 years of commercial 
> software engineering experience working in Smalltalk, Java, C/C++, et. al. on 
> Unix/Linux and Mac platforms. I am potentially interested in working with 
> gnuspeech to explore voice control in speech synthesis as my dissertation 
> topic. I am being funded to do speech synthesis for the Irish language. While 
> the primary effort at this point is for a unit concatenative voice,  I may 
> also investigate what it would take to add synthesis for Irish to gnuspeech.
>
> I have not yet downloaded the software from the cvs repository, but have been 
> reading some of the documentation. I have just a couple of questions at this 
> point for the development team. The platform I would be interested in running 
> gnuspeech on and doing development on is Mac OS/X Tiger.
>
> Questions:
> - is it possible to generate audio output from gnuspeech on Mac OS/X at this 
> point?

Yes.  The Monet system allows "British/American" English speech to be 
synthesised at the segmental level, and the intonation can be added 
automatically or manually, under both 10.3 and 10.4, as described in the 
Monet manual.  Monet is the most important tool in constructing the rules 
and other databases for new languages and dialects.  Steve Nygard did the 
port and made some improvements to the original.  He has had to move onto 
other things for the time being, but he was an ideal person to do the work 
as he had not only worked on the original NeXTSTEP version, but had wide 
experience of Mac OS/X and porting problems as a member of OmniGroup. 
The Monet system is intended for interactive design of the speech 
databases.  We need to make a stripped down version that can be part of a 
speech server/daemon to provide speech as a service -- so-called "Realtime 
Monet".  There's a diagram on the savannah gnuspeech website that shows 
the relationships.

I say "British/American" English because the basic data is British English 
(vowel and consonant targets, rhythm and intonation) but concessions were 
made to American pronunciation in the dictionary entries -- especially the 
use of somewthing approaching the American "r" sound.  I was against that, 
but the majority, at the time, seemed to be convinced that with the US 
market as the biggest market, we should use a more "American" accent. This 
really ought to be changed.  Many Americans actually like a proper British 
accent.

Also we need to port "Synthesizer", which is a GUI front end for the tube 
model.  This is used to gain understanding of the tube model itself by 
providing direct access to all the parameters, and it allows vocal 
configurations needed for a new language or dialect to be pinned down and 
quantified as a basis for creating posture data entries in the Monet 
system.

[Greg Casamento has modified (added to) the code that Steve Nygard 
developed so that it can also be compiled for GNU/Linux, but there's still 
a little work on that to be done.  I haven't heard from Greg for a while.

Adam Fedor took the original NeXT 3.3 code and started the porting by 
making the changes appropriate to OpenStep, which was a good starting 
point for Steve's OS/X port.]

Finally, the dictionary editing tools should be ported and improved.  This 
is a fairly small job.

> - Are there any sound files available that give samples of the speech output?
Yes, I have attached a comparitive synthesis of "Hello" by male, female 
and child voices from the system.  I can send more stuff, including 
sentences and so on if you like.

> - Is it still possible to get the software/hardware necessary to run the full 
> suite of gnuspeech (i.e. NEXTstep?)?

Yes, it is on the savannah site, in the repository.  The original 
Developer kit and in-house tools software are there.  You will need a 
password to install and run the Developer kit, and you will find a bunch 
of valid passwords in the file "serial-nos.txt" also attached (you need a 
"Developer" serial number for the Developer kit -- 2nd half). [Hm!  I 
checked the repository and, for some obscure reason, the equivalent file 
"gnuspeech/gnuspeech/trillium/priv/SerialNumbers/file" has been truncated. 
I'll have to fix that right away.] I probably should re-organise the 
repository to make it much more clear what is available, and add some of 
the oddments that run on the NeXT, like dictionary editing tools and 
information on the dictionary entry formats.  The dictionary in the User 
and Developer kits is encoded, but there is no longer any reason to do 
this.

You should find the package for the most recent release of the NeXT 3.x 
Developer kit (release is 2.01) in 
"gnuspeech/gnuspeech/trillium/TextToSpeechKit/Release2.01/Dev" in the 
repository.

The Mac OS/X stuff is in the directory "current" rather than "gnuspeech".

As I say, feel free to bug me with any questions, requests or whatever.
Let me know if any other repository stuff appears to be damaged.

david

 >
> Thanks for any info you can provide.
>
> Eric Zoerner
> School of Linguistic,  Speech and Communication Sciences
> Trinity College Dublin
>
>
> _______________________________________________
> gnuspeech-contact mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/gnuspeech-contact

helloComparison.snd
Description: Basic audio

serial-nos.txt
Description: Text document