New England Wanted: someone to help out on a project benefiting blind vintage computer users

[twoheartsinthedark]

Closest Major City

boston

Hi all lovers of classic computing. As a blind lover of yester-years computers I would like to share some personal experiences. Starting in the late ‘70s and continuing well into the early 21st century; us blind folks happily plugged along on any number of Atari, Commodore, Apple, and Microsoft DOS machines. While we considered it magic at the time; there really was no magic to it just some simple science. We used external speech synthesizers with names like Arctic, Double-Talk, DEC Talk, Vortex just to name a few. The most common, and arguably the best was the venerable DEC Talk Express (external) and its internal cousin the DEC TALK PC.

Here is the general theory:


https://www.listening.com/blog/speech-synthesis


But from the end user’s point they didn’t have to concern themselves with that. In the case of the popular DECTalk Express the user merely needed to connect the DecTalk to a DB9 serial port on their computer and direct their screen reader software to it.


For blind lovers of vintage computing now the software screen readers are still around in various Internet archives. Many have been released into the public domain for use free of charge. An example of this is JAWS for DOS:


https://allinaccess.com/happ/


The challenge is that the supply of used external speech synthesizers has or soon will be dried up. Unfortunately, as we get old and pass on many of these units get thrown out. Consequently, the world of vintage computing for us blind folks is coming to a close. What would once again, open up the world of vintage computing to us blind folks is a recreation of an External DECTalk Express. The way I can imagine this is that a Linux software speech synthesizer, like ESpeak, could be running on a Raspberry Pie. The Pie would have a USB to serial connection installed to allow for connection to a DB9 serial port on the computer, which is what the screen readers expect. So far all of that is off-the-shelf software and hardware. What would have to be created is a software program that could be run on the Pie that would listen for the text stream coming in from the serial port and then pass that input along to the speech synthesizer which in turn could speak through the headphone jack on the Pie.


This program would have to listen and respond to DECTalk control character strings. They can be found at:


https://rmdir.de/~michael/DECtalk_Express/rongemma/volume.htm


And now for my plea for help. If there are any Linux/Pie developers that would like to take on this project for the benefit of all blind vintage computing enthusiasts, I would love to hear from you. I honestly don’t know what fair compensation would be but I am sure some compensation could be involved. Even if the fine folks here don’t want to get personally involved I would love to hear opinions on the feasibility of this project as well as suggestions.





Thanks


Frank


twoheartsinthedark@hotmail.com











Estimated parts list:


Raspberry Pie


https://www.amazon.com/CanaKit-Rasp...90&sprefix=raspberry+pi+5,aps,157&sr=8-3&th=1


Speak speech synthesizer for Linux


https://espeak.sourceforge.net/download.html


USB to DB9 serial:


https://www.amazon.com/Adapter-Chipset，DB9-Serial-Converter-Windows/dp/B0759HSLP1/ref=sr_1_1_sspa?crid=236I64S61EQUS&dib=eyJ2IjoiMSJ9.xWQsG3VO7IqOId9P7LqGZq38EEH2QvMypF3SndBiNDHWvPNKkkfgYzYA0pD41q3IjI2_ArJV46Wj27eBeQcyBdbW13HiUK_p3b4D_w3wOl3jRSF0UykqtRxJP_l1OKBvCNnI8Gk2f0DaY-xt74-p4WlZ7lZONme6W9bu9qLr_u7-TCe6V3k47ERRnKr1p4STS93i_tAK0u2UP-TBvMkG5QysadotCwOMGo-U-L19WG0.PCwoBg8nZ8_4xUSozrwGK9opLdpTM5qNhfSRaBgtyJU&dib_tag=se&keywords=USB%2Bto%2Bdb9%2Bserial&qid=1757878593&sprefix=usb%2Bto%2Bdb9%2Bserial%2Caps%2C154&sr=8-1-spons&sp_csd=d2lkZ2V0TmFtZT1zcF9hdGY&th=1

 

[Egzoset]

Salutations,

Although i am not directly concerned by the challenge of computer access for people with reduced sight, or total visual blindness, this is a topic which retained my attention starting with the Commo program of Fred P Brucker, then the Braille compatible PLoP Linux of Elmar Hanlhofer who released version 25.2 just recently.

Sorry, i must admit having no special abilities in this domain, except for some very basic awareness about the fact that a graphic user interface does not work well for everyone. Yet, considering you are also seeking input from less informed members as myself, i would point out that a virtual device for DOSBox-X or 86Box might result in some more universal accessibility and possibly even quicker.

Good day, have fun!

 

[nijssen]

Hi Frank, Home Assistant uses Piper as a text to speech engine, see https://github.com/OHF-Voice/piper1-gpl. This also runs on a Raspberry Pi, maybe something to look into.

 

[Slob]

TextToSpeech Click - Mikroe

Text To Speech click is a mikroBUS™ add-on board that carries an Epson S1V30120 speech synthesis IC. MIKROE-2253

www.mikroe.com

The above is supposed to be using DEC's speech engine. The only odd thing I see about it is that it requires some firmware to be uploaded before it operates, so you would need it and a microprocessor between the chip (and amp/speaker) and the screen reader.

I've been using a Chinese SYN6988 speech module and an inexpensive +5V class D amplifier module in a few projects. Documentation is not readily available for the SYN6988 but I did inadvertently discover that it can directly do text to speech from 4800/9600 baud +5V serial data and a READY line (using the packet structure described in the translated documentation is not necessary!). It has only one Chinese female voice (that speaks English), and the TTS algorithm is certainly not as good as DEC's, but you can buy one of these for $15-$20 on Amazon. I would say that the speech quality is FAR better than phoneme-based synths, but again not quite as good as DEC's. It probably has some capability to "fix" mispronunciations in code but this isn't documented. Still, I suspect that blind people just work around things like that...

 

[smp]

I don’t know if this will help or not.

There is a module called V-Stamp that you send text to via RS-232, and it speaks out the words and sentences as voice.

Available from RC Systems:

Text to Speech Voice Synthesizer Modules

V8600A and V-Stamp voice synthesizer modules. Complete data sheets and product info.

www.rcsys.com

With this module, one can use any computer that has RS-232 output, so should be reasonably inexpensive solution.

smp

 

[Istarian]

While going straight to using a Raspberry Pi or Linux SBC seems to be everyone's favorite go-to, I don't see why you need necessarily go that far.

In principle you could use any vintage speech synthesizer chip you can find as long as at least a basic design to utilize it is available.

E.g.

Vintage Speech Synthesizers – Dr. Scott M. Baker

Or you could just make do with a micro-controller and no Linux involvement.

E.g.

Speech Synthesis On A 10 Cent Microcontroller

Speech synthesis has been around since roughly the middle of the 20th century. Once upon a time, it took remarkably advanced hardware just to even choke out a few words. But as [atomic14] shows wit…

hackaday.com

https://github.com/atomic14/ch32v003-audio

10 Cent Microcontroller Makes Music

Compared to the old 8-bit Arduinos, it’s incredible how cheap modern microcontrollers like the ESP32 have become. But there are even cheaper options out there if you don’t need that kin…

hackaday.com

I'm not saying you have to go this route, but using a microcontroller with a software implementation of Texas Instrument's LPC speech synthesis architecture seems like a reasonable way to achieve decent results at low cost And as "powerful" as that particular chip is, using something with more flash and memory would probably reduce the effort involved.

LPC -> Linear Predictive Coding

Linear predictive coding - Wikipedia

en.wikipedia.org

Speech coding - Wikipedia

en.wikipedia.org

Speech synthesis - Wikipedia

en.wikipedia.org

-----

Tangentially, the Wikipedia page for DECtalk mentions that "The video game Moonbase Alpha uses a software version of DECtalk for chat text. Several viral videos were created showcasing users using the chat in a nonsensical manner and using its ability to sing songs.

And then there is this software -> https://github.com/dectalk/dectalk