Re: Fwd: Iggdrasil, a new amazing screenreader

[Didier Spaier <didier@slint.fr>]

Martin, you made my day with your story about the 6800-base microcomputer.

I used a microcomputer for the first time circa 1978 (not sure an Olivetti
Programma 101 with external magnetic cards to store programs and data that I
used previously would be considered a computer nowadays).

This machine was equipped with a keyboard, a screen, two 8 inches floppy disks
readers an a 10 MB hard disk. It had an Inter 8080 inside with 64 KB of RAM.

People seeing it on my desk thought it was kind of a terminal and often asked:
"but where is the computer". You are not alone, my friend <smile>.

I agree that nobody should have to choose between a text interface and a
graphical one (and between Braille and speech). As an aside mpv is perfectly
able to displays videos in a console, but I digress.

Oh, and about OCR I highly recommend the Lios software, which associated to
Tesseract (or Cuneiform) can manage scanning and recognizing, recognize images
also in a pdf file or taken by the computer's camera, speak the recognized text
with espeak-ng and store an audio recording in wav of mp3 format, relying on
speech-dispatcher and any of the associated speech synthesizers.

Cheers,
Didier

On 31/12/2021 00:02, Martin McCormick wrote:
> I've been following this sometimes-difficult discussion intently
> as it goes to the heart of what computer users who happen to be
> blind deal with on a daily basis especially when things go wrong.
> How much insult can your working environment take before you
> simply can't rescue what's there without either blowing the whole
> thing away for a fresh install, getting somebody sighted to watch
> the screen at 3:00 on a Sunday morning while you poke around,
> trying to get the system to boot again.
> 
> 	We've got the same darn problem today that I first
> encountered the first time I ever laid hands on a microcomputer
> which was in 1977 or so.
> 
> 	I got to go to a presentation by the University of
> Houston, Clear Lake City which was given at Oklahoma State
> University.  I was a graduate student then and one of my
> advisors got me a ticket to attend.
> 
> 	My knowledge then was severely limited to say the least
> and I even remember asking the professor who gave the
> demonstration whether this 6800-base microcomputer was actually
> connected to a big main-frame on campus or if it was the whole
> thing.
> 
> 	He was very nice about it all and said it was the whole
> computer, not connected to anything else.
> 
> 	The box probably had only a few KB of ram in it, 8 big
> toggle switches across the front of it and probably a pair of
> 7-segment LED readouts for the display.
> 
> 	No!  It wasn't accessible but it could have been made
> that way if someone connected an extra device to the
> seven-segment display drivers.
> 
> 	It's the same darn problem we have now only in it's
> nascent form.  If only we could just read what those LED readouts
> said.
> 
> 	Of course, I knew nothing about computers at that moment
> except that you could get them to play music, run machines and
> even talk if they were powerful enough and programmed by people
> who were smart at their jobs.
> 
> 	In 1965 or so, I heard a radio program in which a Bell
> Labs computer had been programmed to sing "A Bicycle Built for Two"
> 
> in a male voice that sounded unearthly but was understandable.
> 
> 	In High school, a teacher played that same Bell Labs
> computer singing "A Bicycle Built for two" with a musical
> accompaniment of electronic notes along with the voice.
> 
> 	I wondered at the time how does one do that?
> 
> 	So basically, electronic computers that can talk and or
> print Braille have been around in industrial systems for over
> sixty years and the problem is still that they have tremendous
> potential but it doesn't come by accident.  Somebody has to
> deliberately decide to solve a specific problem and whether that
> solution fits in such a way as to solve other problems, is almost
> an accident if it does.
> 
> 	We still have this fixation on a visual readout as being
> the sole way to peek inside the operation of the system and
> absolutely 0 alternative methods to do so or to put it another
> way, there is no Plan B and it's an utter shock when you raise
> the issue.  Imagine, as a small child, being told that the stove
> is hot but giving your hand a third-degree burn every day because
> one insists on touching the stove even though the outcome is
> quite predictable.  We would call that mental illness anywhere
> else because most of us don't like being burned so tend to
> remember how we got that way and we figure out to, dare I say it?, 
> learn to use the mitt to pickup the hot pan and not have to go to
> the emergency room every day for the same reason.
> 
> 	One day, we may have a video evaluation device that
> connects where a monitor does, reads the screen via OCR plus
> looks for a cursor blinking away somewhere on the screen and
> gives us the functional equivalent of a screen reader or
> talking/Braille monitor but until that day happens, it's a
> chicken-and-egg situation.
> 
> 	We need the computer running for the screen reader to
> work and we need a running computer to produce speech or Braille, period.
> 
> 	This problem will persist in almost it's original form as
> long as Speech/Braille is an afterthought and the full system
> must be running for this afterthought to actually work.
> 
> 	I have tasted what life could be like with such an
> interface when I bought a device that takes VGA output and turns
> it in to web cam video and feeds it in to a USB port on a
> computer equipped with the appropriate driver.
> 
> 	I used Tesseract which is an open-source OCR application
> developed by several blue-chip companies and was able to read the
> output of a few PC's that have VGA ports on them.  I got
> letter-perfect text and could almost have used this setup as a
> screen reader except for the fact that the video constantly
> changes and the system would have to continuously stream the video
> in order to even begin to function like a screen reader.
> 
> 	This would truly be a game changer but that may be what
> we need to hopefully be talking about something other than
> repeating the same discussion which has been going on way too
> long.
> 
> 	Don't misunderstand me.  I am not criticizing anybody's
> position, here, but am stating what the real problem boils down
> to.
> 
> 	The good news is that it is doable but the bad news is
> that it is complex but when it ever gets done, a lot of the same
> old stuff we've been talking about when I got in to computeing
> back in 1979 and earlier for others of you will hopefully be an
> unpleasant memory like walking barefoot five miles to school
> up-hill in the snow and somehow walking 5 miles up-hill again
> barefoot back home.  I get it.  They moved the hill during the day
> so it was up-hill both ways.
> 
> 	I like a good snappy console screen reader that responds
> quickly and helps me trouble-shoot when things go South but there
> is a good argument for a good graphical interface screen reader
> too.  Nobody should have to choose these days.
> 
> 	Martin McCormick
> 
> Jeffery Mewtamer <mewtamer@gmail.com> writes:
>> Sorry if anyone gets this message twice, I got tripped up by the
>> default reply to being a choice between the last respondant and
>> everyone instead of something sensible like "just the list".
>>
>> I might have missed some details on account of several people quoting
>> huge chunks of the conversation and their short replies getting lost
>> in the quote walls(a string of natural 1s upon whoever thought quoting
>> the entirety of the last message should be the default when replying
>> to e-mail), but some of my thoughts on what has been said:
>>
>> Jordan, I don't like to call people out by name, but while I feel like
>> everyone else in this thread has expressed their opinions in a way
>> that is respectful to others, your comments have come across as you
>> stating opinions as fact and that anyone who disagrees is an idiot. I
>> hope that isn't how you intended your comments to come across and have
>> just fallen into the trap of things sounding harsher in text than when
>> spoken, but please try to be more respectful to those who prefer a
>> different Linux setup from your own. And for the record, I'm a fan of
>> Debian as well, and if not for a few things, I'd happily switch to a
>> vanilla Debian setup instead of running  a highly customized
>> derivative. That Debian "just works" for you is great, but it doesn't
>> "Just Work" for me and there are others in the same boat. And I don't
>> think anyone wants you to go back to Windows, we just don't want to
>> feel like we're being insulted for thinking Debian isn't perfect and
>> not all other distros are garbage. Again, I hope I just grossly
>> misreading your comments, but that's how you've been coming across and
>> I don't think I'm alone iin that.
>>
>> As to the original topic, my only real complaint with Orca is that its
>> written in Python and the overhead that introduces, and Yggdrasil
>> being written in Rust doesn't really address that since it's trading
>> an interpreted language I'm at least somewhat familiar with with one I
>> know absolutely nothing about. Honestly, I'd be more interested in a
>> project to rewrite Orca in C++ with a goal of improving performance on
>> weak hardware, but I understand the reasons that is unlikely to ever
>> happen and that I'm unusual for being so young yet finding C++ less
>> frustrating to write code in than Python.
>>
>> Also, isn't Debian 10.1 a very old version at this point? I don't
>> always keep the correspondence between version numbers and codenames
>> straight, but didn't Debian 11 become stable several months ago at
>> this point? And I thought the point releases for Debian 10 were at
>> least up to 10.6. Is whoever is having trouble with Debian sure they
>> are using the latest stable release? And since Debian 9 is working for
>> them, I'm curious if an upgrade breaks things...
>>
>> As for secure boot... I might be misinformed, but I thought that was
>> just a Microsoft ploy to try and prevent people from nuking
>> preinstalled Windows they don't want but is nearly impossible to avoid
>> without paying extra in favor of installing Linux... Is there any
>> downside to disabling Secure Boot and using Legacy Boot if you aren't
>> planning a Linux/Windows Dual boot?
>>
>> As for Slint, while I haven't tried it due to lack of spare hardware I
>> can easily setup to experiment with and reluctance to try anything
>> that requires me to learn a new package management system, I applaud
>> the efforts that have gone into it and wish there was a similar
>> project based on Debian(The Adriane component of Knoppix comes close,
>> and my system is customized from Adriane Knoppix, but Knoppix has
>> priorities aside from just making what its based on more accessible,
>> and its nature as a Live Distro makes it subpar as an installed system
>> running from hard drive) or if there was an effort to make the scripts
>> Slint uses to steamline installation of accessibility tools and easily
>> switch between them on the fly could be downloaded and run on a
>> variety of distros(imagine if Debian had a "reader-switcher" package
>> that recommended all the major console screen readers and then let you
>> switch between the installed ones on the fly, and because its in
>> Debian, it gets inherited by the umpteen dozen distros derived from
>> Debian... Because, as much as I love SBL and wish it was available on
>> a wider selection of distros, there is at least one case where I find
>> espeakup's "try to read everything as it appears on screen" behavior
>> beneficial instead of annoying, and I'm sure there are others who
>> spend substantial amounts of time in the console who would enjoy that
>> flexibility.
>>
>>
>