Text-To-Speech: Festival/speechd working, but no talkers in KTTSMgr

cautha · 2010-05-07 20:14:50

This might be better off in Sound & Multimedia, but since I'm not actually having problems with sound output, and I use TTS to help me work, I figured I would put it here.

I have Festival set up correctly with the HTS voices, and I can get it to say things from the console, but I haven't been able to add a talker in KTTSMgr for quite some time.

Basically, I discovered one day that the TTS feature in Okular wasn't working (possibly after the "Speech Dispatcher" update to KTTSD), and I noticed that my configured talker was screwed up, so I removed it and tried to make a new one, but the list is now completely empty.

I found the package festival-freebsoft-utils, which is supposed to help Festival work with Speech Dispatcher, but I haven't had any luck.

Any ideas?

EDIT: I also tried it using the voice from the festival-jmk-arctic PKGBUILD, but I have the same problem.

Last edited by cautha (2010-05-10 10:42:13)

cautha · 2010-05-10 07:06:41

Where do I start? .... I understand the reasoning behind speech-dispatcher, but setting it up... *facepalm*

Before the change, it was so easy---but I digress.

Anyway, I figured out that speech-dispatcher wouldn't run without dotconf, which wasn't in the PKGBUILD; festival-freebsoft-utils was listed as an optional dependency for festival compatibility, so I also installed that.

When I started speech-dispatcher, I got an error that told me two things---that speech-dispatcher needs to be run as root, and that it needs to be configured separately, which wasn't immediately obvious, since all I had to do prior to the new KTTSD was set a default voice in festival (and sometimes force it to use ALSA). So, I ran spd-conf and did the tests, and I at least got sound output working, but the message that played was from the dummy output module (how nice!). It told me to look in /var/log/speech-dispatcher/speech.log, but there was no file there, same as with the corresponding directory under ~/.

speech-dispatcher also seems to require that festival be run as a server, which I did. I then went into ~/.speech-dispatcher/conf/modules/festival.conf and uncommented the server/host lines, just in case.

I also edited the relevant files inside /usr/share/speech-dispatcher/{...} but no joy. Am I doing something wrong? Should I just switch to espeak or flite?

cautha · 2010-05-10 10:39:31

Rawr! I got it to work, sort of. I think I tried so hard because I hated the thought of using alsa-oss. And I discovered a couple bugs in spd-conf! Does anyone care? lol

Basically, I redid my TTS-related configs & put them under ~/ instead of /usr/share, and ran "festival --server" & "speech-dispatcher -s" as a normal user. I also added myself to the "audio" group somewhere along the line, but I don't think that was the issue.

I'm almost afraid to reboot, because, although speech-dispatcher can be started at boot-time through rc.conf's DAEMONS array, I don't think festival can---at least, not in the same way as it can on distros that have the script /etc/default/festival. I'm about to see if I can track down that file, and hope I don't break everything somehow.

KTTSMgr still doesn't see any synthesizers, but I got past the "dummy module" message. Now I'm hoping I'll get some help from someone who actually knows what to do (as opposed to me).

Take care, all

H

P.S One small complaint: Although my speechd.conf specified port 6561, speech-dispatcher's tests showed that it was looking for a connection on port 6560 first, before scanning to find the real port. I had to change to port 6560 to stop the behaviour, even though I never customized the port number in the first place.

Last edited by cautha (2010-05-10 10:53:55)

cautha · 2010-05-11 00:16:34

Is it considered bad form that I'm basically talking to myself in this thread?

I switched to espeak for simplicity's sake; I also figured out why kttsmgr is not showing any talkers, although I have no idea how to fix it:

kttsd(2379)/kttsd (daemon): could not get a connection to speech-dispatcher

Is there a way that I can get more verbose logs from kttsd?

Last edited by cautha (2010-05-11 00:19:54)

Arch Linux

#1 2010-05-07 20:14:50

Text-To-Speech: Festival/speechd working, but no talkers in KTTSMgr

#2 2010-05-10 07:06:41

Re: Text-To-Speech: Festival/speechd working, but no talkers in KTTSMgr

#3 2010-05-10 10:39:31

Re: Text-To-Speech: Festival/speechd working, but no talkers in KTTSMgr

#4 2010-05-11 00:16:34

Re: Text-To-Speech: Festival/speechd working, but no talkers in KTTSMgr

Board footer