Multilanguage Text-To-Speech
Gilles Casse
gcasse at oralux.org
Mon Jan 21 17:59:14 CET 2008
Hello,
FYI, mokoTTS 0.05 has just been released.
If you are interested in, you might have speech out of the vox (uh, box)
after installing its packages: espeak-pulse, speech-dispatcher (two
espeak Chinese packages are also available). Works using Navit, the
navigation software; current tests using Emacspeak.
https://projects.openmoko.org/projects/mokotts/
Best regards,
Gilles
From the INSTALL file:
Good news :-) :
* The upstream speech-dispatcher 0.6.5 and eSpeak 1.31 include our
PulseAudio extensions,
* OpenEmbbedded includes espeak, speech-dispatcher, libdotconf.
mokotts 0.05 provides now the customizations for OpenMoko (on gta01 +
pulseaudio 0.9.6) which can not yet be provided upstream, and some
tests on the Neo1973.
Note:
we have the best results at the moment using either:
* speech-dispatcher with PulseAudio and espeak providing
44100Hz/stereo samples.
This is the default configuration provided by mokotts.
We suspect that the 44100Hz/stereo give better results because
otherwise PulseAudio 0.9.6 would do CPU intensive resampling.
* speech-dispatcher with Alsa and espeak providing 22050Hz/mono samples.
This conf is not supplied by default by the mokotts packages because
some conflicts can be met with PulseAudio.
If you wish to test it, it would be better to stop the PulseAudio
server and copy our .asoundrc file under /home/root (available in
the tarball in mokotts/test/alsa/.asoundrc.44), and then restart the
speech-dispatcher server.
More information about the community
mailing list