[vox-tech] speech synthesis of a browser's window

[Henry House]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

På tisdag, 02 mars 2004, skrev Peter Jay Salzman:
[...]
> p@satan$ dpkg -l "*vox*" | grep -v none
> festvox-ellpc1 1.4.0-3    Castilian Spanish male speaker for Festival
> festvox-kallpc 1.4.0-4    American English male speaker for festival
> festvox-kdlpc1 1.4.0-4    American English male speaker for festival
> 
> do you know, offhand, how to tell which voice file i'm using?

Try running the festival interpreter directly or with test scripts and
setting the speaker explicitly? Debian has a bunch in
/usr/share/doc/festival. The festival executable is actually a lisp
interpreter --- very cool.


> any quick tips on improving quality?

I have terrible speakers but I still found the quality adequate when using
festival and the non-free us1 voice from MBROLA. I have not used
text-to-speech really extensively, though, so your opinion may differ. Still
worth a try. You can get a demo here: <http://festvox.org/voicedemos.html>.

I have occasionally wanted to start a project to create a new high-quality
speaker (with a true Free license) for US English. Is anyone else interested
in such a project?

- -- 
Henry House
The unintelligible text that may follow is a digital signature. 
See <http://hajhouse.org/pgp> for information.  My OpenPGP key:
<http://hajhouse.org/hajhouse.asc>.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQFARlhFi3lu92AVGcIRAovDAJ96/uH+LwAzeS54BEMGPQMopqMIggCgivvF
Hiv6quXZlKa0clWtlykRqC8=
=FmBg
-----END PGP SIGNATURE-----