Complexs and Complexity, Or Why Granny needs a 4096 to do her taxes

<Greg> I had another program designed for more general audience, which I created back in 2010. but the special project grant through my university ended in the fall of 2011. And I had to lay it aside. I have always wanted to rewrite this particular program the one for the general audience, using my vastly improved experience. With both programming and accessibility, but due to my poor financial circumstances I would need about three thousand dollars in grant money to make the project feasible. I know that sounds mercenary for an open source developer. But this is not something I can do with my spare time on ten year old equipment. I want to give this project the time and attention it deserves, and that requires money. But that’s enough of that tangent

VSSS then is a speech orchestration system  for my specific print related disability. But what does it do specifically well to quote from the source.

A brief code snipit

VOX="Callie"

rate=190

PIPE\_COLOR="1;33;45m"

export PIPE\_COLOR

audio\_bckend="padsp"

spkedit="pluma"

PATH=/home/matt/pkg/bin:$PATH

LD\_LIBRARY\_PATH=$LD\_LIBRARY\_PATH:/home/matt/pkg/lib64

speak\_bckend() {

   if \[ -f /tmp/vsss.lock \]

   then

       echo "Speech output is currently in use"

       return 0

   else

       touch /tmp/vsss.lock

   fi

   cat $1 | ./emojifilter.py > $1.new

   cat $1.new > $1

   $audio\_bckend swift -n $VOX -p "speech/rate=$rate" -f $1 -m text -t |  ./pcg

   rm /tmp/vsss.lock

   return 1

}

VSSS Explained

Now this requires some explanation this function takes a file name as a parameter the file name could theoretically be anything on the system. But the file is most often generated by DBUS  service which we will get to in a minute then it takes the file through a complicated shell pipeline beginning with the so called emoji filter. Which is really misnamed because it's a generalized preprocessor which also does regular expression replaces. And ending with the invocation of the swift which is a proprietary speech synthesizer originally meant for telephony applications. Which outputs the audio and simultaneously prints the output of the so called normalization process to standard out which then goes to a program called pcg. Which stands for pretty color graphics.  and is just there to make the normalization output easier to read onscreen. The final output looks something like this

I can't give you an audio file of how it sounds due to copyright restrictions. But here's a YouTube video of Cepstral Callie, saying something. There are a few other bells and whistles that allow me to jump around in the document, edit the document before it is read. And as I said there is a whole preprocessor  emojis with their text descriptions, and does standard regular expression replacements for various things.  This is the first component of my accessibility system.

MasterText

The next component which I wrote myself is called MasterText. This requires a little bit of context before I explain what it does. In 2020-2021  I was accepted into graduate school for Information Technology Management. At Empire State College, our state universities remote learning college.

While the disability services there were far from the worst I'd ever seen. It was basically a torture test for my disability. Not helped by the fact that due to how grants worked. I didn't get the money for my textbooks until six weeks into the semester.  

I eventually had to drop out. Both during and after I contemplated ways to augment my accessibility system to cope with the workload.

Out of this came MasterText, which quite simply replaces the  API which fetches text from the screen into the reader. So that every piece of text that is run through the reader also ends up in  persistent database. Which is both full text searchable and content addressable.

Bestowing Superpowers on myself

Every piece of text has a unique address which can be referenced if I need to read it again. And I can search through anything I've read since 2019. It also has a web interface which has some features of a wiki. Meaning I can annotate texts, cross reference them and generally do any shenanigans which I feel are necessary. When researching a topic or writing  a paper and so forth. If I had thought of this idea before I had entered grad school  I would have my masters by now.

This tool  allows me to use my auditory memory, which is one of my most advanced skills. As a very finely  honed tool in any sort of writing endeavor.

BlueProxy

The final tool that I coded for myself is a more  recent addition, and simply leverages Mozilla's Readability API. To simplify the contents of websites so that the speech system produces more coherent output. I could always do this in browser. But having it on  as a web proxy and API frees me to use simpler browsers such as otter, midori  and others. And also will eventually lead to integrations with other tools. That I am currently working on, so stay tuned as they say.

The Rest

On to parts of the system that I didn't write myself. As I stated in the beginning of the article I use Compiz Fusion with the Mate desktop, to provide missing accessibility features in X11. A terminal based word processor  Wordperfect for Linux about which I've written previously. And I have recently replaced my  dictation and voice control system, which used a windows XP based virtual machine and several horrible kludges. With the much more functional, and elegant Talon Voice Control System. Which has made life so much easier in the past few  weeks.

Conclusion

Now if the disability I described, sounds something like you. and you want to set up a similar system. I would recommend starting with the long term support release of Ubuntu Mate. Though I am a Fedora user and proud of it. The release model of fedora can make it susceptible to sudden and dramatic changes, which the user has to work around. For example recently during the transition to pipewire the fedora devs removed the backward compatibility shim that allowed OSS  binaries to run as pulseaudio clients which completely broke my entire text to speech system. They did this even though there was no reason to remove such a component as pipewire, implements the same API as pulseaudio. And so the shim is also compatible with pipewire. I had to find  a way. To reinstate that missing feature all without the ability to read.

Suffice to say fedora is an adventure and if you're just starting out building your own accessible computer system without the disability industrial complex. I recommend starting with something stable, even though new features are slower to get to you.

If you want to get VSSS or MasterText running shoot me at DM  over on tilde.chat,  or an email I am matt -at- piusbird dot spce.  I would be very interested in  what sort of documentation would be needed for a new user to start using the system.

Finally if you know where I can get grant money to pursue first building a speech system for print disabled users, and then secondarily making a  distribution derivative, that more tightly integrates some of these accessibility components. And makes them easier to install and use. I would welcome your  Feedback as well