Re: Espeak Server Likes PulseAudio?

The real solution is likely to re-compile espeak NOT to use pulseaudio. I seem to remember reading somewhere that the latest version has a more dynamic form of configuration (which should eliminate the need to recompile espeak for pulseaudio btw). I suspect that as you do have some pulseaudio installed, this new dynamic configuration is getting confused and using pulse rather than just alsa.

Note that there should be no problems with alsa and pulseaudio. Pulseaudio is not a replacement for alsa. It is another abstraction layer on top of it and pulseaudio needs alsa

I've never used speakup, ut this seems to be where the problems lie. I use emacspeak, speech-dispatcher and gnome all with pulseaudio on multiple systems with absolutely no problems apart from the emacspeak espeak driver being very very unstable. One one system, I use gnome, speech-dispatcher configured to use espeak and emacspeak configured to use IBM viavoice. On another system, I use only espeak, but use speech-el instead of emacspeak because I find espeak crashes far to frequently with emacspeak. The speech-el and speech-dispatcher interface to espeak is much much more stable than emacspeaks interface. I suspect it is related to the use of SSML in the emacspeak driver, but that is just a guess.

All my systems use pulseaudio out of the box these days. I don't use any .ausoundrc file and with most recent ubuntu based distros, don't recompile espeak anymore either. I find speech-dispatcher with espeak to be rock solid. Works great for the gnome stuff and apart from some very minor issues, works well with speech-el. Unfortunately, you need to do a lot of customization to make speech-el as useful as emacspeak. For me, this was worthwhile because the tts interface is so much more stable, even though I don't get multiple voices. I'd love to have time to write an emacspeak to speech-dispatcher interface, but this just ends up another task along with writing a dbus tts interface for emacspeak and finishing my clojure based interface to espeak. Too many half done projects and not enough time. I really want to get back to my clojure based tts interface for emacspeak. It was showing great promise and provided a really convenient way to have both espeak and viavoice support on a 64 bit system which did not force you down the 32 bit space for espeak i.e. all the espeak stuff can stay native 64 bit.

Tim

On 29 March 2013 20:16, Steve Holmes <steve@xxxxxxxxxxx> wrote:

Hey, this is an excellent idea. That is actually working for me now. The only down side to this is I can't use sounds from gnome and that probably kills media players like totem or gnome-mplayer. Not sure about the players for sure. But at least I can use emacspeak again and still have access to speak up.

I sure don't like PulseAudio's general behavior and never have.

On Mar 28, 2013, at 3:20 PM, Derek Roberts <bigd.vi.guy@xxxxxxxxxxx> wrote:

> Hi,
> Another thing you could do is
> chmod -x /usr/bin/pulseaudio
> Unless you actually need it running for something...
>
> Thanks,
> KJ4UFX
>
> On 3/29/13, Christopher Chaltain <chaltain@xxxxxxxxxxx> wrote:
>> I'm not sure if this will help or not, but following some advice I found
>> on the web (I think from Tim.) I recompiled eSpeak from source to have
>> it use PA if it was present. I'm sure you could do the opposite and
>> force it to use ALSA. Here are the notes I wrote up for myself. I hope
>> they help.
>>
>> **** Recompiling eSpeak to Use PulseAudio
>>
>> Unfortunately, the last syllable spoken is always truncated making it
>> unusable.
>>
>> I fixed this by downloading the source code for eSpeak and compiling it
>> for runtime, which uses PulseAudio if it's running and PortAudio otherwise.
>>
>> I went to http://espeak.sourceforge.net/download.html and downloaded the
>> file espeak-1.46.02-source.zip.
>>
>> I had to install the library development packages for PulseAudio and
>> PortAudio.
>>
>> $ sudo apt-get install libpulse-dev libportaudio-dev
>>
>> Now after unpacking the source code for eSpeak, I went into the source
>> directory, espeak-1.46.02-source/src, and I made the following changes
>> to the Makefile:
>>
>> #DATADIR=/usr/share/espeak-data
>> DATADIR=/home/chaltain/Tools/share/espeak-data
>>
>> #PREFIX=/usr
>> PREFIX=/home/chaltain/Tools
>>
>> # 'runtime' uses pulseaudio if it is running, else uses portaudio
>> AUDIO = runtime
>> #AUDIO = portaudio
>>
>> Now I run the commands:
>>
>> $ make all
>> $ make install
>>
>> This creates new eSpeak libraries and binaries, using the runtime
>> option, and places them in my ~/Tools directory.
>>
>> Now to start Emacspeak with eSpeak and these new libraries, I run the
>> commands:
>>
>> $ export LD_LIBRARY_PATH=/home/chaltain/Tools/lib
>> $ emacspeak -e
>>
>> aNow Emacspeak comes up and I don't have the latency issues.
>>
>> Note that I exported the variable LD_LIBRARY_PATH instead of running the
>> following command:
>>
>> $ sudo ldconfig /home/chaltain/Tools/lib
>>
>> so that I'd only run emacspeak with this new library and not making the
>> change globally, so as not to impact Orca if I did something wrong.
>>
>> When using ldconfig, note you can use
>>
>> $ ldconfig -p | grep espeak
>>
>> to see which eSpeak library is being used.
>>
>> On 03/28/2013 11:48 AM, Steve Holmes wrote:
>>> Somehow, it seems like every time I launch emacspeak now and when I
>>> test the espeak server from the emacspeak servers directory, it
>>> appears to be launching pulse audio on my system and that in turn is
>>> giving me all kinds of problems with speech losses and taking over my
>>> ALSA devices. This all began recently when I thought I would give
>>> Pulse another look since I use gnome and want gnome sound support.
>>> But after running into conflicts with Speakup and ALSA, I decided to
>>> disable Pulse by commenting out the default.pa file but every time I
>>> start up emacspeak now, I get some broken or garbled speech and all
>>> ALSA speech such as from speakup is silent until I quit emacs. Anyone
>>> know what could be going on here?
>>>
>>> I would love to just uninstall pulse entirely but other packages
>>> require the dumb thing so can't remove it. Any way to get emacspeak's
>>> espeak server to use ALSA only and not be calling on pulseaudio?
>>>
>>> -----------------------------------------------------------------------------
>>> To unsubscribe from the emacspeak list or change your address on the
>>> emacspeak list send mail to "emacspeak-request@xxxxxxxxxxx" with a
>>> subject of "unsubscribe" or "help".
>>>
>>
>> --
>> Christopher (CJ)
>> chaltain at Gmail
>>
>> -----------------------------------------------------------------------------
>> To unsubscribe from the emacspeak list or change your address on the
>> emacspeak list send mail to "emacspeak-request@xxxxxxxxxxx" with a
>> subject of "unsubscribe" or "help".
>>
>>
>
> -----------------------------------------------------------------------------
> To unsubscribe from the emacspeak list or change your address on the
> emacspeak list send mail to "emacspeak-request@xxxxxxxxxxx" with a
> subject of "unsubscribe" or "help".
>

-----------------------------------------------------------------------------
To unsubscribe from the emacspeak list or change your address on the
emacspeak list send mail to "emacspeak-request@xxxxxxxxxxx" with a
subject of "unsubscribe" or "help".

--
Tim Cross

If you have questions about this archive or had problems using it, please send mail to: