Hello
This is the first time I've waded in on this thread.
I think the original authors may not be aware of the effort to this end
which has already been happening.
For my part, all was well with SpeakUp etc. until April 2013 when an
attempt to bring DMA to the ALSA driver 'broke' eSpeak text-to-speech on
the Pi and produced a tendency to stutter very, very badly, and to crash
the kernel.
This I fixed with my audio code, using the OMX integration layer and a
fork of espeakup, which I called piespeakup.
I am now writing an Emacspeak speech server which uses the same code,
and as stated recently, intend at some time to write a speech-dispatcher
audio driver to use the same code.
Using the OMX code, SpeakUp does not stutter, nor does it crash the
kernel. But it does occasionally lock-up, a deadlock I suspect as there
is a lot of thread-synchronisation in the code.
Here is my Hacker Public Radio podcast which demonstrates the stuttering
problem and the fix:
http://hackerpublicradio.org/eps.php?id=1649
I'm now working on a very cheap and accessible version of Arch Linux for
the F123 project.
It is the intention of that project to bring a Raspberry Pi to the hands
of students and others in the developing world.
Concentration is on Emacspeak, and possibly Google Chrome, running with
no window manager.
I have developed code which I intend to use in that project, and which
is published in my git repositories, which will provide an easy means of
backing up a running Pi, and bootstrapping a new and equally accessible
Arch installation.
The Foundation have ignored me when I have asked them to look into
fixing the stuttering DMA ALSA driver, and anyway the OMX code works
better, as it circumvents the problem of the console going quiet when
speech-dispatcher starts in pulse mode.
I have also discovered that running speech-dispatcher in ALSA mode is a
failure. There is a bug in the sd_espeak driver which regularly and very
quickly crashes and causes itself to be unloaded.
speech-dispatcher can be run with libao, but as all underlying modes,
whether it is pulse, libao, or any mode other than direct audio
rendering on the GPU, uses ALSA at the lowest level.
I have to interleave my efforts to write this stuff with my efforts to
stay alive by trying to find paid work. But I do still do stuff from
time to time.
Mike
Code is still in development.
There is little possibility of sensible voice-operated input.
On 07/08/2017 15:57, Rich Morin wrote:
Voice input is already quite useful and will continue to get better.
However, I don't expect it to be convenient and reliable any time
soon for editing computer source code, etc. So, although I don't
have a problem with adding voice input, it isn't a total solution.
Also, companies such as Apple and Google have the motivation and
resources to push voice input forward for the general population,
including the blind and visually impaired. Any new cell phone will
thus have better voice input than we have any hope of developing.
In contrast, putting together a blind-friendly version of Linux and/or
a handheld personal computer for the blind is a tiny niche. It's a
happy accident of history that Linux has a lot of text-based tooling,
but the Linux community in general isn't interested in taking on these
challenges. Fortunately, a small group of motivated individuals might
be able to make significant progress on these fronts.
-r
--
http://www.cfcl.com/rdm Rich Morin rdm@xxxxxxxx
http://www.cfcl.com/rdm/resume San Bruno, CA, USA +1 650-873-7841
Software system design, development, and documentation
===========================================================
The raspberry-vi mailing list
Archives: //www.freelists.org/archives/raspberry-vi
Administrative contact: <mike.ray@xxxxxxxxxxxxxx>
-----------------------------------------------------------
Raspberry Pi and the Raspberry Pi logo are trademarks of the Raspberry Pi
Foundation.
This list is not affiliated to the Raspberry Pi Foundation and the views and
attitudes expressed by the subscribers to this list do not reflect those of
the Foundation.
Mike Ray, list creator, January 2013
