2:48 p.m.
I would like to check to see if some of my accessibility goals mesh with the
philosophy of the xemacs project. I want to fix a bridge (VR-mode) which enables
Emacs to work with NaturallySpeaking, a speech recognition package by nuance.
I had a dustup with Stallman over the issue of accessibility support and Emacs.
His assertion was that free software foundation principles come before
everything including integration of disability aids for disabled people, in this
case software developers.
The conflict came because I was looking for a way to drive, Emacs using speech
recognition (NaturallySpeaking). Because this is not a completely free solution,
it was shot down and disabled folks like myself were pushed in the direction of
completely nonfree software.
If a hybrid solution is acceptable, it brings me to my second query. I need help
bootstrapping. When your hands don't work right, it is really hard to write
code. The only language I've been successful in writing code with unmodified
speech recognition is Python and that's only if I ignore any symbol naming
conventions.
What I would need help with is fixing up the package VR-mode which is an
extension to Emacs (currently broken) to provide "Select-and-Say" functionality
inside of Emacs. The package needs work to simplify and fix its network
capabilities which allows you to use speech recognition with Emacs or remote
machine.
My goal is to get to a point where I can start dealing with programming by
speech efforts again. The continual breakage of PR-mode and the reaction I've
received from the free software foundation community about Emacs driven by
NaturallySpeaking has been discouraging. I'm afraid I'm going to be pushed to a
closed solution from top to bottom like I have with terminal emulators.
Anyway, thank you for reading my thoughts.
--- eric
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
3:09 p.m.
On 01/31/11 08:48 AM, Eric S. Johansson wrote:
I would like to check to see if some of my accessibility goals mesh
with the
philosophy of the xemacs project. I want to fix a bridge (VR-mode) which enables
Emacs to work with NaturallySpeaking, a speech recognition package by nuance.
<SNIP>
Anyway, thank you for reading my thoughts.
--- eric
Hi Eric:
I'm only speaking for myself (and I don't know anything about speech
recognition)... But, a mode that I contribute too, ESS, supports both
free and commercial software and I have never heard a peep about it
from XEmacs. $0.02
Rodney
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
12:57 p.m.
Eric S. Johansson writes:
I had a dustup with Stallman over the issue of accessibility support
and Emacs.
His assertion was that free software foundation principles come before
everything including integration of disability aids for disabled people, in this
case software developers.
XEmacs itself *is* free software, and I think most project members
would fight for that principle. However, Stallman's "sabotage
anything that might get in the way of a world that is 100% free
software" approach (which reminds me of the Leninist principle of
"immiseration of the masses" -- but I digress) is not accepted here.
Most of us are more likely to contribute to your work to the extent
that it's based on free software vs. proprietary, yes, but to the
extent that you do the work we're happy to have it. In fact, in the
past Ben Wing and Jerry James have worked with proprietary speech
recognition software if I'm not mistaken (but they're not active at
this point, although I think Jerry reads this list).
We have no objection in principle, but note that you do need to keep
that proprietary stuff on the other side of the "exec barrier" because
of the GPL. Eg, you can't link XEmacs to a "Naturally Speaking"
library. (Well, as a legal fine point *you* can link to anything you
want, but you can't give it to anyone else unless it's 100% GPL, not
even in pure source form. :-( ) You need to use some form of
interprocess communication such as shared memory or TCP/IP or Berkeley
sockets.
If a hybrid solution is acceptable, it brings me to my second query.
I need help
bootstrapping. When your hands don't work right, it is really hard to write
code.
I have sympathy, but no time at present, and when I do get time I'll
have to apply it to release management for old code, not development
of new code. But I'd like to contribute in the medium term.
I'm afraid I'm going to be pushed to a closed solution from
top to
bottom like I have with terminal emulators.
That would suck. Please, try working with XEmacs.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
1:29 a.m.
On 2/1/2011 7:57 AM, Stephen J. Turnbull wrote:
We have no objection in principle, but note that you do need to keep
that proprietary stuff on the other side of the "exec barrier" because
of the GPL.
Not a problem. We have a bridge called "natlink" which
bridges between
NaturallySpeaking and our application world. I wish I could clearly describe
the relationship between various components but, sometimes it seems like a
mystery to the best of us. :-)
One way to think of it is that there are two parallel streams of input. The
keyboard stream and the NaturallySpeaking keyboard stream. natlink (aka
natpython) adds a third stream. When grammars resolve to an action routine, that
routine can inject text into the keyboard stream or can call functionality that
directly tickles a method in another application.
Messy isn't it?
It's also incredibly useful when you want to separate command input versus
textual input. For a fun example, start gnus and type my name. it's not pretty
is it? Well, that's what happens when you say command name and it misrecognizes
it, generating some similar sounding string of text instead of a command. But if
you can recognize when you are in a recognition state, you can turn off single
character commands in an application thereby saving you some horrible disaster.
I'm probably not making enough sense and I apologize but this is the world I
live in.
My goal is to try and drive applications directly which allows for more powerful
class of speech interface. I guess with Emacs, I would end up injecting lisp
code or something twisted like that :-)
I have sympathy, but no time at present, and when I do get time
I'll
have to apply it to release management for old code, not development
of new code. But I'd like to contribute in the medium term.
I get that a lot.
:-) Hopefully somebody else can help. And I hope that folks
don't mind if I do a "thinking out loud" series of dumps in the mailing
list. If
it's a problem, I'll find a blog somewhere and put my brain dumps in that format.
> I'm afraid I'm going to be pushed to a closed
solution from top to
> bottom like I have with terminal emulators.
That would suck. Please, try working with XEmacs.
I will. I just got the Windows version running. There's a couple things that
should be fixed in the gui. There should be a pixel or two more space between
the frame of the window and the body of the displayed text. For example, I'm
looking at a screen which has beta, info, tutorial up against the left hand
margin in the needs to be just a little space between the letter and the window
frame to make it easier on the eyes. Also, any chance of finding a graphics
artist to redo the icons? Unfortunately, people are rather superficial of our
applications and if the images aren't pleasing and modern, they tend to reject
applications. Why is this important ti xemacs? it's my belief that when people
are looking for a version of Emacs to use, they look for what looks good. Silly
I know but I've seen that behavior. If folks are okay with it, I will try to
explore my network and see if I can find anyone who is available to do a new set
of icons etc.
I think the font could be better as well. It's a relatively nice fixed spacing
Sans Serif but something about it doesn't work for me. Again if I can get a
pointer to how to change that, I'll make a proposal of what I think works out
better..
Looking over the menus, it's pretty good I might want to know how to rearrange
them for my own interest/discovery of commonly used features but, as a general
rule, I like them. Better than the other white editor.
Now I need to figure out what the heck is going on with getting Python-mode
working on beta. Something complaining about cannot open load file
"local-mode".
At once I get the basics working and the fundamental speech recognition
command-and-control working (yes, I will donate the macros to the project) it'll
be time to make myself a PITA to the Python-mode group and start begging for
someone take pity on me and help with cleaning up vr-mode. (damm, I'm going to
have to learn Lisp)
And I think that's enough randomly generated but reasonably associated set of
thoughts for the evening. I appreciate the attention and help.
--- eric
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
5:01 a.m.
On Tue, Feb 1, 2011 at 5:57 AM, Stephen J. Turnbull <stephen(a)xemacs.org> wrote:
Most of us are more likely to contribute to your work to the extent
that it's based on free software vs. proprietary, yes, but to the
extent that you do the work we're happy to have it. In fact, in the
past Ben Wing and Jerry James have worked with proprietary speech
recognition software if I'm not mistaken (but they're not active at
this point, although I think Jerry reads this list).
I'm still hanging around, plotting schemes that I don't have time to
actually carry out. :-(
I did tinker with voice recognition some years ago. I've remained
interested in it, to the point where you might be able to convince me
to donate some of my brain cycles to your cause, Eric.
--
Jerry James
http://www.jamezone.org/
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
5:47 a.m.
On 2/2/2011 12:01 AM, Jerry James wrote:
I did tinker with voice recognition some years ago. I've
remained
interested in it, to the point where you might be able to convince me
to donate some of my brain cycles to your cause, Eric.
cool
what do your brain cycle say about http://emacs-vr-mode.sourceforge.net/
I have some philosophical problems with how they implemented a grammar inside of
VR-mode. What I would like to do is slice and dice the current application down
to just enabling Select-and-Say within the buffer. Then I would like to
replicate the commands externally in vocola (one of our higher level macro
action routine packages.)
However creating new macros does raise the question of invoking Emacs commands
by key injection or by calling internal methods.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
3:12 a.m.
On Tue, Feb 1, 2011 at 10:47 PM, Eric S. Johansson <esj(a)harvee.org> wrote:
cool
what do your brain cycle say about http://emacs-vr-mode.sourceforge.net/
It took me a few minutes to work out that DNS == Dragon Naturally
Speaking. My brain insists that that acronym stands for Domain Name
System.:-)
I can't actually try this out, since I'm on a Linux machine. Are the
sources for vr.exe available somewhere?
I'll look through it. Give me a day or two to digest how the code works.
I have some philosophical problems with how they implemented a
grammar inside of
VR-mode. What I would like to do is slice and dice the current application down
to just enabling Select-and-Say within the buffer. Then I would like to
replicate the commands externally in vocola (one of our higher level macro
action routine packages.)
However creating new macros does raise the question of invoking Emacs commands
by key injection or by calling internal methods.
When I messed around with this eons ago, I took the approach of making
voice be a distinct input stream, just as keyboard and mouse input are
distinct. That way I could hang functionality on the voice stream
that I wouldn't want on the keyboard stream, such as making
corrections when the VR engine didn't understand me. I was working
with ViaVoice at the time, back when it used to be owned by IBM, and
they used to support Linux. My stuff leveraged some earlier work by
Larry Sanders.
Over the past couple of years, I've slowly been submitting bits of the
CMU Sphinx stack to Fedora, with an eye towards doing the voice
recognition work over again with Sphinx as the engine. However, I
don't have the full stack in Fedora yet.
--
Jerry James
http://www.jamezone.org/
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
3:43 a.m.
On 2/2/2011 10:12 PM, Jerry James wrote:
On Tue, Feb 1, 2011 at 10:47 PM, Eric S.
Johansson<esj(a)harvee.org> wrote:
> cool
>
> what do your brain cycle say about http://emacs-vr-mode.sourceforge.net/
It took me a few minutes to work out that DNS == Dragon Naturally
Speaking. My brain insists that that acronym stands for Domain Name
System. :-)
yea, a common confusion :-) I use DNS to drive DNS.
I can't actually try this out, since I'm on a Linux machine.
Are the
sources for vr.exe available somewhere?
yes
http://sourceforge.net/projects/emacs-vr-mode/
I'll look through it. Give me a day or two to digest how the
code works.
When I messed around with this eons ago, I took the approach of making
voice be a distinct input stream, just as keyboard and mouse input are
distinct. That way I could hang functionality on the voice stream
that I wouldn't want on the keyboard stream, such as making
corrections when the VR engine didn't understand me. I was working
with ViaVoice at the time, back when it used to be owned by IBM, and
they used to support Linux. My stuff leveraged some earlier work by
Larry Sanders.
That's where I'm thinking about going. Unfortunately, Dragon has moved down the
road of injecting to the same queue is the keyboard. with the natlink
interface, we should be able to do direct injection.
Over the past couple of years, I've slowly been submitting bits
of the
CMU Sphinx stack to Fedora, with an eye towards doing the voice
recognition work over again with Sphinx as the engine. However, I
don't have the full stack in Fedora yet.
I wouldn't waste any time with
Sphinx. The vocabularies are too small for
real-world desktop use. Command-and-control is a nice add-on, Gen. dictation is
critical. The way I look at it, if you can't write a piece of e-mail, or
document piece of code, the recognition system has failed. I would look at the
Simon project instead. http://simon-listens.blogspot.com/
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
5041
days inactive
5044
days old
7 comments
4 participants
participants (4)
-
Eric S. Johansson -
Jerry James -
Rodney Sparapani -
Stephen J. Turnbull