espeakup and sound cards?

Steve Holmes steve at holmesgrown.com
Sat Mar 5 16:09:13 EST 2011


I haven't looked at the build script for the non-git speech-dispatcher
package but it could be possible that you have alsa routing everything
to pulse which could explain the loss of sound when killing
pulseaudio.  There is that suggested asound.conf file which is all set
up for alsa to route through pulse.  I used that originally to try and
give me full pulse coverage as well as black-listing the snd-pcm-oss
module.  Even with that configuration, I managed some how to have
speech from espeakup at login time.  I think there are situations
where pulseaudio will start up by itself without explicitly running
pulseaudio first.  But again as soon as I ran speech dispatcher or
allowed gnome to cause speechd to AutoSpawn at orca startup, then is
when I either got a bunch of distortion, volume too loud or whatever.

On Sat, Mar 05, 2011 at 04:01:14PM -0500, Kyle wrote:
> On 03/05/2011 03:32 PM, Jacob Schmude wrote:
> >-----BEGIN PGP SIGNED MESSAGE-----
> >Hash: SHA1
> >
> >Hi Steve
> >No, that's not the issue in your case. The volume bug exists when
> >espeak outputs to Pulse, however speech dispatcher doesn't use
> >Espeak's output routines. Sd interfaces to Pulse directly, uses
> >libespeak to synthesize, then outputs the synthesized data itself. I
> >haven't had any volume instabilities with speech dispatcher and
> >Pulseaudio, but I did recompile speech dispatcher with pulseaudio
> >support (the default Arch packages do not have it unless it was added
> >within the last week). Did you recompile speeech dispatcher to have
> >Pulse support? If not, you might be trying to use it through the ALSA
> >plugin, which can create all sorts of wacky problems.
> >As for OSS4, I wish it had lived up to what it could've been. The API
> >is simpler, it's quick and responsive, and has a much better output
> >routine than ALSA. Unfortunately, most of OSS4's drivers are buggy
> >beyond belief, and USB microphones aren't supported at all. If you've
> >got one of the cards which has a stable OSS driver and don't care
> >about USB input, you can use it. I'm not so lucky, unfortunately, and
> >all of the stable OSS4 cards have long since been discontinued.
> >
> >
> >On 03/05/2011 09:05 AM, Steve Holmes wrote:
> >>If GNOME3 is going exclusively pulse, that doesn't leave much
> >>choice. I wasn't originally aware that espeak was maxing out the
> >>volume in pulse; suppose that is the cause of our volume
> >>instabilities then? Because at one point last week, I was in GNOME
> >>and managed to have most other sounds going along pretty good
> >>without distortion or crackling or any of that but it seemed that
> >>as soon as I rebooted and brought up speech dispatcher then it all
> >>went to hell and of course, speech dispatcher launches espeak; so
> >>is that where the problem lies?
> >>
> >>As for other sound alternatives, whatever happened to OSS4. I had
> >>heard comments that it was even better than pulse. I notice that
> >>none of the speech systems I know of implement OSS4 AFAIK.
> >>
> >>On Sat, Mar 05, 2011 at 02:30:15PM +0000, Christopher Brannon
> >>wrote:
> >>>Jacob Schmude<j.schmude at gmail.com>  writes:
> >>>
> >>>>Hi Arch's espeakup packages, just FYI, are using the git
> >>>>version of espeakup with the direct ALSA option. I'd be all for
> >>>>getting rid of it,
> >>>No, they aren't. Look more closely at the PKGBUILD. How did you
> >>>come to this conclusion? If you're trying to use the "device"
> >>>setting in espeak's configuration files to redirect espeak's
> >>>output to another sound card, it won't work, because of an espeak
> >>>bug. The option wasn't being parsed correctly. AFAIK, it is
> >>>fixed in the testing versions of espeak.
> >>>
> >>>-- Chris _______________________________________________ Speakup
> >>>mailing list Speakup at braille.uwo.ca
> >>>http://speech.braille.uwo.ca/mailman/listinfo/speakup
> >>_______________________________________________ Speakup mailing
> >>list Speakup at braille.uwo.ca
> >>http://speech.braille.uwo.ca/mailman/listinfo/speakup
> >
> >-----BEGIN PGP SIGNATURE-----
> >Version: GnuPG v1.4.11 (GNU/Linux)
> >
> >iQEcBAEBAgAGBQJNcp3GAAoJEDtK6CrF/Uo1F2AH/3VhcpJJj2ajj4b8Lan8K7/R
> >Ahe5X59zSDj8TklnnuBZgLaTArbG/tQSrLlyZKNW0B/Sj5QViQwvdyWHU7/i0wib
> >boOmAIu2KJyzrk4AOhIX7VWSZokS9FanL+b5PnAuAreNIeZnEG6yc3J1xF0xm9FV
> >LDTfX+15KWUdSRBxSTX2xcvoLiak0v2aHJGE8cU71OOKW1df1OY0frCnU10px0l0
> >JSOB5b9SBs+AMNmQ651j33aa5z1hVtH8sTPAdbpUcogcbT0BSDY4f1OHD5c4Fa3V
> >YVeqyJPucm/PwgxIhhMki5ZeGNDjSaEH852RZsscjEPMsonVaWCOCM13bhSRcxw=
> >=PlZv
> >-----END PGP SIGNATURE-----
> >
> >_______________________________________________
> >Speakup mailing list
> >Speakup at braille.uwo.ca
> >http://speech.braille.uwo.ca/mailman/listinfo/speakup
> I have speech-dispatcher configured to use pulse. I'm using the
> default Arch package with version number 0.7.1-5. Apparently pulse
> support is compiled in this release. I get no errors regarding pulse
> or alsa from speech-dispatcher, it speaks correctly and if I kill
> pulseaudio I lose speech, all indicating that it is properly using
> the pulse interface.
> 
> From playing around with pulse configurations, it seems that most of
> the trouble appears to come from resampling. For some reason, if I
> play a sound sampled at 16KHz while eSpeak is playing at 22.05KHz,
> everything distorts until the 16KHz sound stops playing. Seems it
> could be a bug on some cards, since this only seems to happen on my
> onboard ATI which is using the snd_hda_intel driver, and it doesn't
> happen on my soundblaster live value.
> _______________________________________________
> Speakup mailing list
> Speakup at braille.uwo.ca
> http://speech.braille.uwo.ca/mailman/listinfo/speakup



More information about the Speakup mailing list