Re: [Emacspeak] Any solutions for the queuing issue in Swiftmac?

[Jerry Sievers <gsievers19@xxxxxxxxxxx>]

Daniil, thanks for the code.

I have built your package and tested it briefly.

Not had much time to experiment further and I'm still using SwiftMac for
daily use but your synth driver compiled and ran easily right from the
start.

I note that it is minimal in features and shows some different behaviors
from what I'm accustomed to...

things like when viewing info documents and you use page up/down, or
perhaps in any kind of text buffer (not confirmed), the page you move to
after the navigation does not proceed to speak as it had always done
with earlier drivers I've used such as the Python based version and then
Robert's SwiftMac.

For the initial test, all I did was to reset dtk-program in my already
running Emacs and then restart the speech server.  Speaking rate was
very slow so I set the rate to something like 250 (which I assume
relates approximately to words per minute.)  SwiftMac uses rate settings
IIRC in the range of 0.0 to 1.0 or somesuch and I usually ran that one
at 0.55.

I found that upon restarting the server to test yours, I had to manually
stop the side-channel notification process.  and that yours I believe
does not support the secondary notification channel that is a relatively
new feature of emacspeak that I'm not totally sold on but I can
understand why it was a potentially useful new feature.

Generally though, your program seems well behaved and soon when I have a
longer stretch of time to play around, I'll leave yours enable for much
longer and see what else I can learn.

I find your program to be very snappy and responsive and shows no signs
of failure to flush previously queued utterances when rapidly navigating
around in a file.

I'll be curious to see if yours works with whatever alternate sound
device my Mac switches to such as when I put on the bluetooth headset
and/or switch from external speakers to the USB-C wired earbuds.

I found that as of some recent changes on the Mac and when in Google
meet calls, apparently the Google meet app switches to some kind of
phone device for microphone I/O or whatever, and SwiftMac would not play
the side-channel audio until the call was ended and I restarted the
speech server.  SwiftMac has all sorts of features for output to various
devices and I have thus far not had time to figure out whether frobbing
some of the SwiftMac audio device settings could resolve that problem.

SwiftMac also is super good in performance though if I continue using it
long-term, I may revert back to an older version of it that I ran for
quite a while before migrating to the homebrew package.

I went to the homebrew version of SwiftMac for the better packaging and
lots of new features it offered but did notice the queueing quirks
though they were tolerable in my usage for the most part.

I'll report back again later if/when I have a lot more experience with
your driver to inform others about.

Thanks again!

Cheers

"Daniil Gusev" (via emacspeak Mailing List) <emacspeak@xxxxxxxxxxxxx>
writes:

> Greetings,
>
> If you have the time and inclination, I suggest giving a try to a
> speech server I developed with a focus on maximum performance and
> minimal overhead between Emacspeak commands and speech playback, back
> when the swiftmac version in the Emacspeak tree felt a bit laggy to
> me.
>
> I haven't managed to switch to Emacs + Emacspeak for my full-time work just yet, so any feedback would be very valuable.
>
> The latest version of the code can be found here:
> https://github.com/Daniil-Gusev/nsspeaker
> To build it, simply running `make` in the project folder is enough.
>
> The code itself is around 1,500 to 2,000 lines of fairly
> easy-to-understand C89, so exploring the project and changing
> something, should you wish to, won't be too difficult either.
>
> I'd be glad if you find this useful.
>
> Daniil
>
> Emacspeak discussion list -- emacspeak@xxxxxxxxxxxxx
> To unsubscribe send email to:
> emacspeak-request@xxxxxxxxxxxxx with a subject of: unsubscribe