Re: Sometimes bad responsiveness with outloud and alsa 1.0.16

["Lukas Loehrer" <listaddr1@xxxxxxxxxxx>]

Hi Tim, 

Thanks for testing. I posted an updated version of the patch (dated
2008-08-09). It will now only allow 1 callback per call to
eciSpeaking(). I believe this is the correct behavior given the
overall design of the speech server. It seems that with this
modification, a single dummy call to eciSpeaking() before calling
eciStop() is sufficient to ensure stability so long text chunks should
not be a problem anymore.

When tweaking parameters in .asoundrc, you should also include the
timeout of the select() system call in the outloud tcl script in the
function service. The default value is set with a period size of 256
samples in mind. The delay should probably not be too short, but
certainly not longer than the period time.

About the delays with utf-8 characters, I guess this is related to the
unicode patch. I am still working on this for emacs 23. I would expect
the following delays: The very first time a non-ascii character is
spoken in an emacs session, emacs loads the file with the character
names. This might lead to a noticeable delay. Secondly, if a line is
read that contains many different special characters that have not
been encountered in the current emacs session yet, their might also be
a very small delay the first time the line is read. The character
names are cached, so if the same special character is encountered
later, their should not be any delay. You can see this in the "hello"
file (\C-h h).

Could you give me an example of a webpage where you are experiencing the delay you describe?

Best regards, Lukas

Tim Cross writes ("Re: Sometimes bad responsiveness with outloud and alsa 1.0.16"):
> Hi Lukas,
> 
> just letting you know that your patch has significantly improved the
> situation on my system with the Audigy 4 sound card. Not only was I getting
> significant delays in interrupting speech, I was also experiencing very
> sluggish responsiveness in character echo, frequent segfaults and often,
> complete duplicate echos of words. 
> 
> I had to tweak my .asoundrc file slightly, but now it all seems very
> stable. Speech is easy to interrupt and the poor responsiveness in
> character echo appears to be gone. There does seem to be some small minor
> problems with word echo, but they are infrequent and not much of an issue. 
> 
> The only very slight thing I've noticed is that when movinig around in a
> web page under w3m, if the text is inside a table that is using UTF
> characters to generate the borders etc, there is often a frequent delay
> when you move to a line before the text begins speaking. Its not a big
> issue, but if your unaware of it, it can be easy to skip a line of text
> because you think it is empty. I've not tried re-configuring w3m so that it
> only uses ascii characters yet. I suspect this is not directly related to
> your patch, but rather more likely something that has now been revealed as
> a result of the patch. 
> 
> Note that I've only noticed very slight problems on my work machine, which
> is using a SB Live clone sound card and have not tried applying the patch
> to that system yet. I will try to find time to do so this week and will
> report back if it has any unexpected problems. 
> 
> Thanks a lot for chasing this issue down. I must admit I was tempted to try
> and debug it, but its been so many years since I've done any real debugging
> of C/C++ code, the prospect wasn't very inviting. 
> 
> Thanks very much
> 
> Tim
> -- 
> Tim Cross
> tcross@xxxxxxxxxxx
> 
> There are two types of people in IT - those who do not manage what they 
> understand and those who do not understand what they manage.
> 

-----------------------------------------------------------------------------
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"