DECtalk Archive | Hacker News Reader

DECtalk Archive

dectalk.nu

I've got a DECtalk (DTC-01) I bought years ago on eBay, intending to use it for a speech recognition(!) project. It was state of the art when it was released - Dennis Klatt's speech synthesis research from the lab turned into a product.

What made DECTalk interesting is that it is a formant-based synthesizer, producing speech much like a human by taking a broad spectrum input voice source (cf. function of vocal cords) and modulating it via resonant frequencies (formants) similar to how we do it by changing the resonant frequencies of our vocal tract via articulation. When we recognize speech it's the frequency-tuned hairs in our inner ear acting as a filter bank and recognizing these resonant frequencies, which our brain has learnt to map back to the articulatory movements used to produce them.

Later, cheaper, and better sounding, speech synthesizers were based on stitching together partial recorded words (phonemes), which made them sound more natural but also limited them to speech. The DECTalk's more fundamental format-based generation allowed it to sing as well as talk, and the clean computer-generated formants made it highly intelligible (albeit artificial sounding) when sped up considerably, which was popular with the intended market of blind customers using it as a reading device.

Daisy, daisy, give me your answer, do ...

> Perfect Paul would end up being a model emulating how Klatt talked in his younger days.

Anyone who started life as a drunken Swedish sailor then went on to invent something like DECTalk has done well.

I worked on a shipping product that used DECTalk back in the 1980s and I can still hear that voice. There were certain words we could just never get it to say correctly, even when spelled out using the special DECTalk phonetic codes. The one I remember strongest was a customer name ("Guelph" properly pronounced "gwelf" but the closest DECTalk could get was something like "welp"). Given that it's been almost 40 years and it's still burned into my brain, I may have DECTalk trauma.

Could perhaps(?) do better with phonemes, but "gwehllf" seems kinda ok.

https://webspeak.terminal.ink/

Man, this brings back memories of playing with Festival back in the day. I wrote a bunch of scripts to generate dynamic Festival text2speech audio for a Part15 AM radio station, a virtual radio station DJ over 15 years ago. The intelligibility was not great back then, the funny thing was after spending many hours listening and tweaking my input text I trained myself to understand the default voice a lot better. Other people had a hard time with the default voice but I could understand it perfectly. Thankfully some of the improved voices available back then helped it a lot.

To fix pronunciation problems I had developed a set of domain-specific "dictionaries", sed substitution scripts that would translate certain words and phrases into their phonetic versions that would be pronounced better by Festival. Besides announcing the names of songs being played, I was also feeding news articles and some other web-scraped info through it. Tons of fun to hear your project working on the air in public as you drive around town.

Yeah reminds me about the weird way Apple provides multi language support for iMessage announcements over AirPods. I’m from Germany and my phone is set to English because of job related reasons. I message a lot of people in English and the family and friends in German. Siri used to be set to German and in the past was either not cape-able to read English messages or butchered the message. Sometimes it simply says it can’t read it. For a year or so some messages will be read by a computer voice similar to DECtalk. I have no clue when the system decides to use it because it happens randomly. Now I switched Siri to English for Apple Intelligence and it got a bit better. But it’s still strange though.

For a school assignment, we had to find a dusty old computer, to run dusty old software DECtalk, to better understand speech.

The assignment was to annotate lines from some famous sayings and speeches, to make the DECtalk output much better than the default.

My lab partner and I might've spent an hour alone on the Gettysburg Address fragment, "...and that government of the people, by the people, for the people, shall not perish from the earth" be delivered like an impassioned speech.

(That line was the most memorable, since there was so much to work with, and you could make it dramatic.)

Less quality but the Commodore 64 SAM [0] predates it by two years. It is available now in JavaScript [1].

[0] https://en.wikipedia.org/wiki/Software_Automatic_Mouth

[1] https://discordier.github.io/sam/

John Madden!

Is this what Stephen Hawking used at the time?

It is frequently stated that Hawking used a DECTalk system, but in fact he did not. He used a Speech Plus CallText 5010. They sound similar because DECTalk and CallText were both based on the MITalk work of Dennis Klatt. The full story can be found here:

  https://computerhistory.org/blog/how-dectalk-gave-voice-to-a-genius-engineering-stephen-hawkings-wheelchair/

https://news.ycombinator.com/item?id=31427032

DonHopkins on May 18, 2022 | root | parent | next [–]

Here's a historic DECTalk Duet song from Peter Langston (which is actually quite lovely):

Eedie & Eddie (And The Reggaebots) - Some Velvet Morning (Peter Langston)

https://www.youtube.com/watch?v=1l0Ko1GUiSo

Peter S. Langston - "Some Velvet Morning" (By Lee Hazelwood) - Performed By Eedie & Eddie And The Reggaebots

http://www.wfmu.org/365/2003/169.shtml

Eedie & Eddie On The Wire

http://www.langston.com/SVM.html

Peter Langston's Home Page:

http://www.langston.com/

His 1986 Usenix "2332" paper:

http://www.langston.com/Papers/2332.pdf

How to use Eddie and Eedie to make free third party long distance phone calls (it's OK, Bellcore had as much free long distance phone service as they wanted to give away for free):

https://news.ycombinator.com/item?id=22308781

>My mom refused to get touch-tone service, in the hopes of preventing me from becoming a phone phreak. But I had my touch-tone-enabled friends touch-tone me MCI codes and phone numbers I wanted to call over the phone, and recorded them on a cassette tape recorder, which I could then play back, with the cassette player's mic and speaker cable wired directly into the phone speaker and mic.

>Finally there was one long distance service that used speech recognition to dial numbers! It would repeat groups of 3 or 4 digits you spoke, and ask you to verify they were correct with yes or no. If you said no, it would speak each digit back and ask you to verify it: Was the first number 7? ...

>The most satisfying way I ever made a free phone call was at the expense of Bell Communications Research (who were up to their ears swimming in as much free phone service as they possibly could give away, so it didn't hurt anyone -- and it was actually with their explicitly spoken consent), and was due to in-band signaling of billing authorization:

When you called (201) 644-2332, it would answer, say "Hello," pause long enough to let the operator ask "Will you accept a collect call from Richard Nixon?", then it would say "Yes operator, I will accept the charges." And that worked just fine for third party calls too!

>Peter Langston (working at Bellcore) created and wrote a classic 1985 Usenix paper about "Eedie & Eddie", whose phone number still rings a bell (in my head at least, since I called it so often): [...]

>(201) 644-2332 or Eedie & Eddie on the Wire: An Experiment in Music Generation. Peter S Langston. Bell communications Research, Morristown, New Jersey.

>ABSTRACT: At Bell Communications Research a set of programs running on loosely coupled Unix systems equipped with unusual peripherals forms a setting in which ideas about music may be "aired". This paper describes the hardware and software components of a short automated music concert that is available through the public switched telephone network. Three methods of algorithmic music generation are described.

It's true! -=P=-

Before reading about it on your web page, I'd totally forgotten that you'd written Empire in 1972, on an HP2000 no less, which you were famous for long before your seminal Eddie and Eedie Usenix paper!

https://en.wikipedia.org/wiki/Empire_(1972_video_game)

I heard the HP2000 had a command to rewind the tape drive, but you could also use it to rewind the disk drive! But it crashed the whole system. I always wondered if that was from a sudden release of kinetic energy, sparks, smoke, and flames, or if the device driver just crashed.

Walter Bright, who made the 1977 version of Empire, hang out here on Hacker News, too. Are you guys old friends or mortal enemies? ;)

https://en.wikipedia.org/wiki/Empire_(1977_video_game)

> The most satisfying way I ever made a free phone call

I remember as a kid me and my buddies making a free call from one phone booth to another (across the road - not terribly useful) just by asking to reverse the charges. The phone company evidentially never realized that the target number was a phone booth, so it worked as normal by the recipient just agreeing to pay (without actually having to do so by putting any money in).

If that was a customer owned coin operated telephone (COCOT), the phone company may not have known it was a pay phone, and the customer may not have known they could instruct the phone company to prevent accepting collect calls.

This was just an outdoor UK red phone booth (back in mid 70's), presumably owned by BT. I guess back then the operator may not have even had a computer to refer to, so not much they could do.

Reply on news.ycombinator.com