DECtalk Archive

9 条评论

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 ...

bregma15 天前

> 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.

评论 #43879287 未加载

dosman3315 天前

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.

larusso15 天前

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.

neilv15 天前

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.)

wslh15 天前

Less quality but the Commodore 64 SAM [0] predates it by two years. It is available now in JavaScript [1].[0] <a href="https://en.wikipedia.org/wiki/Software_Automatic_Mouth" rel="nofollow">https://en.wikipedia.org/wiki/Software_Automatic_Mouth</a>[1] <a href="https://discordier.github.io/sam/" rel="nofollow">https://discordier.github.io/sam/</a>

ValveFan696915 天前

John Madden!

hcarvalhoalves15 天前

Is this what Stephen Hawking used at the time?

评论 #43865266 未加载

DonHopkins15 天前

<a href="https://news.ycombinator.com/item?id=31427032">https://news.ycombinator.com/item?id=31427032</a>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)<a href="https://www.youtube.com/watch?v=1l0Ko1GUiSo" rel="nofollow">https://www.youtube.com/watch?v=1l0Ko1GUiSo</a>Peter S. Langston - "Some Velvet Morning" (By Lee Hazelwood) - Performed By Eedie & Eddie And The Reggaebots<a href="http://www.wfmu.org/365/2003/169.shtml" rel="nofollow">http://www.wfmu.org/365/2003/169.shtml</a>Eedie & Eddie On The Wire<a href="http://www.langston.com/SVM.html" rel="nofollow">http://www.langston.com/SVM.html</a>Peter Langston's Home Page:<a href="http://www.langston.com/" rel="nofollow">http://www.langston.com/</a>His 1986 Usenix "2332" paper:<a href="http://www.langston.com/Papers/2332.pdf" rel="nofollow">http://www.langston.com/Papers/2332.pdf</a>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):<a href="https://news.ycombinator.com/item?id=22308781">https://news.ycombinator.com/item?id=22308781</a>>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.