DECtalk DTC01 - 1984 Speech Synthesizer

[@TechTangents]

My apologies to F- *r* -ail Frank, the Fail was on my part there with that typo in the original script that I didn't catch until it was already uploaded.

[@DavidScheiber]

John Madden

[@tomahzo]

Fail Frank is so much better, though ;D

[@TreesandStuff69]

You could say that was a Fail Frank moment you had!

[@hey_imriver]

"If you can hear this, there's a good chance your DECtalk is working" I love how this is phrased. "It's probably working, but don't get too excited just yet"

[@kwinzman]

Typical engineering. Don't overpromise.

[@vwestlife]

Perfect Paul was the voice used by NOAA Weather Radio when they replaced human-read recordings of weather forecasts with speech synthesis in the late 1990s. In 2002 they began replacing him with more natural-sounding voices, but Paul could still be heard giving station IDs for many years after that.

[@RussSirois]

I thought it sounded super familiar!

[@mllarson]

I think my local station still has the ID spoken by a DECTalk of some sort. I'll have to check tomorrow.

[@vwestlife]

@@mllarson Here's Paul giving the forecast in 2016: kzbin.info/www/bejne/aqjGm594qcmclac#t=32s

[@EASsirenVids01]

@@mllarson if it does let me know what station it is

[@KanawhaCountyWX]

You would also sometimes hear stations switching to Paul as a fallback whenever the other systems would get overloaded. I miss the multi voice setup, because of course we don't have that with the BMH anymore.

[@TheFakeVIP]

As a blind nerd I can say that we still talk about these fantastic things. Many a skit has been made with them by a bored kid (blind or not) with access to an emulater. And a lot of us still prefer klat synthesis over more modern solutions because of how much more response (I.E. lower latency) it can be, and sometimes because it's clearer at higher speeds. Neural voices might sound really good, but when your main interface to a system is by speech, you don't care how beautiful the voice sounds, you care about how fast you can interact with it, and how much information it can tell you per second. What good is a nice sounding voice if you have to wait even 0.2 seconds after you press any single key for it to start telling you what control is now focused?

[@kaitlyn__L]

That’s a great point! I hadn’t really thought about the “nicer” voices having lower intelligibility at high speeds. Even though I’ve got blind folks in my family and were aware of more things than most sighted people, there’s still a lot that goes over my head! (Which I shouldn’t be surprised by - when I started using a wheelchair I realised I’d only previously noticed 10% of the obstacles my sister navigated in hers.)

[@ian_b]

I actually prefer to listen to a distinctively artificial voice rather than a nearly-human voice. The nearly-human voice has an "uncanny valley" effect on me. It's disturbing at some level. I can't bear to listen to those AI-voiced KZbin channels.

[@seamarie3111]

I guess I'm in the blind minority because I really can't stand the older voices and the way they speak and pronounce things. I guess this might come from having a name that was never pronounced right, with a silent vowel. I can understand Vocalizer and Siri voices just fine at high speeds. I remember jumping ship from Eloquence to RealSpeak as soon as those voices were available. However, I'm still quite picky. I really only like Nuance voices, and only if I can't use an Apple Siri voice. However, I'm sure some people are glad to have the old Mac voices on, say, iOS, or the eloquence voices.

[@cmd_f5]

I'm 100% with you. Give me a distinctly synthetic voice over concatenated speech sample voices any day. They're faster, lighter on resources, and generally more flexible. Dectalk and eSpeak (now EspeakNG) are my favorites of the formant and synthetic methods. Plus the flexibility and power of Dectalk make it a lot of fun. Ah let's not forget Doubletalk if we're talking old tts. Another great lightweight solution. Hoping for a modern software interpretation of that someday.

[@dakotahrickard]

I admit I prefer Eloquence over many other, arguably better voices. The voice has familiarity, true, but it is just generally more readable for me. I use mine at a rate, generally, of 80 to 90% of maximum, and while I can get that high with some other voices from time to time or in special circumstances, Eloquence is simply the best solution, for me personally, in terms of latency and intelligibility. I'm pretty sure it's also Klat based. That being said, I think that, given the impetous to do so, a voice using more modern technology could probably be optimized for comprehensibility at higher speeds. Sharper consonants, short but defined pauses, and carefully non-truncated phonemes would definitely be nice to have. The problem is, as I said, in optimization rather than actual capability. As we all know, the modern voices are optimized to sound humanlike, including human speaking heuristics at human speech rates. It's all a question of priority and experience. Let's give a brief example. I was very surprised to learn quite how fast the Windows OneCore voices can be made to go. But the comprehensibility, in my experience, of those voices falls off pretty sharply. The trouble isn't speaking rate. The trouble, instead, is an enhancement of the built-in speech errors that work to make those voices, such as they are, sound more natural. I did run into that a little, even with DECTalk, it has certain errors in pronunciation to make it sound more natural. The point is, with modern voices, we have technology optimized to function best at typic human speech parameters. The older tech is more robust using different parameters specifically because it wasn't optimized for anything in particular. Specialization limits general utility. In other words, it's an old, true saying that the jack of all trades is master of none. I would expand that by saying, the master of one trade is good for little else.

[@Lord_Nightmare]

DECtalk DTC-01 v1.8 units are VERY RARE, I only know of one other unit with that firmware in it. Beware if you try to update the unit to DECtalk DTC-01 v2.0, you not only need to replace the 16 EPROMs, you also need to replace the two bipolar PROMs which hold the firmware for the TMS32010 DSP, since if you run DECtalk v2.0 68000 code with the DECtalk v1.8 DSP code, the speech output will be very quiet, maybe 1/2 or 1/4 of the volume it should be. (BTW, do you have high-res pictures of those EPROMs and PROMs? I want to be sure MAME has the EPROM labels correct for them in their DECtalk emulation driver.)

[@TechTangents]

Unless there is something major like a fix for the screeching when singing I don't anticipate updating it to 2.0 because it works. I'm imagining the one other 1.8 unit is probably the one that the MAME ROMs came from? That's good to know it would be more complicated than I anticipated if I tried though. I did put a picture of the entire mainboard up on my device repair site here: caps.wiki/wiki/DECtalk_DTC01#/media/File:DTC01_Mainboard.jpg If you need more specific detailed shots or something else feel free to ask! If email is better you can also reach me at [e-mail] . I'm happy to help preserve this device through emulation!

[@Lord_Nightmare]

@@TechTangents Yes, that's where the MAME ROMs came from.

[@nysaea]

Can't help but love how your videos aren't just "look at how super cool this rare cool thing is!" (which would be great already) but "look at how super cool this rare cool thing is! Now I'm going to educate you on a metric ton of history and technical lore about it in an easily digestible format because I went out of my way to research the subject more in depth than you could ever ask for!" For real, thank you.

[@agranero6]

19:11 noooooo. You must make it sing "Daisy, Daisy give me your answer true, I'm half crazy with my love for you".

[@TheM750]

No no, it needs to sing John Madden.

[@nova-witchwood]

There is a flower within my heart Daisy, Daisy Planted one day by a glancing dart Planted by Daisy Bell Whether she loves me or loves me not Sometimes it’s hard to tell Yet I am longing to share the lot Of beautiful Daisy Bell Daisy, Daisy Give me your answer do I’m half crazy All for the love of you It won’t be a stylish marriage I can’t afford a carriage But you’ll look sweet upon the seat Of a bicycle built for two…

[@Lord_Nightmare]

10:03 "Like earlier speech synthesizers, the DECtalk uses Linear Predictive Coding" - This is entirely wrong. DECtalk uses a formant-based cascaded vocal tract engine, developed by Dennis Klatt around 1977-1980 at MIT and called "KLSYN" (later 'KLSYN80' to differentiate it from a later version he made around 1988 just before his death). The Fortran source code for an early version of KLSYN was published in the article "Software for a cascade/parallel Formant Synthesizer" in the volume 67 issue 3(March 1980) of the Journal of the Acoustical Society of America. It does not use LPC coding at all! Klatt himself ported this Fortran code to C (you can find versions of this on github) and also ported it initially to hand-made assembly code for the TMS32010 in DECtalk DTC-01 v1.8; DECtalk DTC01 v2.0 has a later version of the DSP code with a number of bugs fixed, which was ported from his C version in 1983-ish.

[@TechTangents]

That is fascinating to know! I was basing this off of the Popular Science article I read which phrased it in a way that wasn't entirely clear throughout the article but made me think it was LPC. books.google.com/books?id=f2_sPyfVG3AC&pg=PA42&dq=dectalk&hl=en&sa=X&ved=2ahUKEwjJlvqtgZz6AhUbJUQIHfD2BGAQ6AF6BAgDEAI#v=onepage&q=dectalk&f=false

[@tfmorris02181]

@@TechTangents I can confirm this -- and the weird artifacts you heard in "Let It Be" were overloads in the vocal tract model.

[@thegeforce6625]

@@tfmorris02181would ROM V2.0 fix the screeching?

[@tfmorris02181]

@@thegeforce6625 The 2.0 ROMs would likely fix many of the issues, but I don't remember if there were any compatibility constraints (e.g. minimum rev PCB, etc).

[@Dorelaxen]

Hell, those youtube channels that use AI speech bots can't even get half the pronunciations right. This is truly impressive.

[@seamarie3111]

As a blind person I even hate those poor-quality voices they use. If you're going to stick a TTS voice on your channel, at LEAST use a Nuance or Siri voice, or something that doesn't sound like total garbage.

[@kargaroc386]

I don't think that's AI, just a python script and a traditional voice engine. But the voice used in those video is iconic itself, so I doubt it'll be changed.

[@knightcrusader]

It took me a moment to figure it out, but yeah Perfect Paul (or Variable Val) is the voice they used for the jacket in Back to the Future II: "Your jacket is now dry!"

[@Screamingtut]

I used to work for DEC back in 1978-1981 as a Sr Field Service Rep in NYC & 1984-1987 As then a Field Engineer in Woburn, MA. I even met Ken Olson at the Mill At Maynard, MA

[@kaitlyn__L]

Wow, that street/saint is VERY impressive - even with today’s more realistic speech synthesis, that abbreviation differentiation still trips-up Google’s webpage reader.

[@SyphistPrime]

I first learned about DECTalk from Moonbase Alpha. It's the in game text to speech engine and it is fully capable of all the commands. People would make it sing about John Madden and whatnot years ago. I also didn't realize the phone dialing commands had a use until you showed how it could be built into a phone system, that makes sense now. Thanks for covering this subject, I love the piece of history that surrounds DECTalk and not just because of the Moonbase Alpha memes.

[@kaitlyn__L]

That’s fascinating, for me it was just a neat easter egg, but I’m really really happy that it’s introduced more people to the culture of speech synthesis history :)

[@zaremol2779]

John Madden

[@s8wc3]

@@zaremol2779 AIOUAOIUAOIUAOIUAOIU

[@zaremol2779]

@@s8wc3 holla holla get dolla

[@RetroGamerr1991]

Mamma Mia. Papa pia. baby got the diarrheeeeeeeeeaaaah

[@Okurka.]

John Madden.

[@SyphistPrime]

aeiou

[@SudosFTW]

Holla Holla Get Dollar!

[@F_I_J_I_W_A_T_E_R]

Your jacket is now dry!

[@AlistairBrugsch]

My dad was a blind computer user back in the 80s (and from then on til he passed) and DEC talk was always the gold standard that his speech systems would be trying to meet. He couldn't afford an actual DEC but he did a really good immitation of perfect paul!

[@SudosFTW]

John Madden John Madden John Madden.

[@FairPlay137]

As someone who has messed around with the Variable Val voice, I can definitely say that those distortions/feedback in some scenarios are definitely frustrating, even in the native PC ports of the DECtalk software. (Which is much newer than the DECtalk units shown here - rather it’s the same version number that the first iteration of the NOAA Weather Radio automated speech synthesis system) I can tell this problem's even worse in DECtalk 1.8 as compared to the PC port's version, since I customized _way_ more of Val's parameters before it glitched - 1.8 experienced issues just by changing the nasal resonator gain, which one of the songs did. Not sure if this glitch is triggered that easily in the 2.0 upgrade though, but I imagine it doesn't.

[@nebular-nerd]

We just need it to start singing covers of the GlaDOS songs from Portal 🤓

[@jimbotron70]

Very clear and intelligible speech, especially for something from 1984!

[@caseyrevoir]

Your productions are quite remarkable, I particularly enjoy the way you speak. Great work!

[@Ghilliedude3]

Hearing this say sharp instead of pound instead of hashtag made me experience a brief, yet intense, temporal dislocation.

[@official-obama]

octothorpe?

[@nanopone]

Command error in phoneme.

[@djdjukic]

Great video, one of your best ones so far. I really appreciate the little skits with the DECtalk voice. On another note, the 68k chip really went into everything. If you had a Mac, a DECtalk and a LaserJet back then, you had three of them. Imagine having a Threadripper in your printer nowadays...

[@mikebledig7208]

I still think the Dectalk Speech Synthesizer is one of the very best real Computer Speech Synthesizers. I'm so glad to have 2 of them! I only hope that the source code to the software version of the Speech Synthesizer will one day be made open source

[@rommix0]

at least it's on github

[@mikebledig7208]

@@rommix0 Do you have the github link so I can check it out?

[@dakotahrickard]

There's one setting that's very culturally relevant to a much wider swath of people than it would, at first glance, appear. I've seen DECTalk in a variety of movies, mostly or entirely from the 80s. It's in Back To The Future II and also the first Short Circuit movie, just to name a couple. It's also featured in some music, both from the time and from later. But the most lingering cultural heritage DECTalk gave us is its singing. Listen to some of the early Vocaloids. Notice anything familiar? Even the vibrato patterns they use are similar to those found in singing DECTalk. I don't know if they give credit or not; that kind of thing doesn't appeal to me as a topic of research, but there's some pretty elaborate stuff to be found which was sung by DECTalk. I happen to know some of the pioneers in that field, so to speak. It's little wonder that someone would want to capture the amazing novelty of that and package it in a much more user friendly form. I have never personally made Hatsune Miku or her relatives sing, but I can't imagine it'd be so popular if it took the effort DECTalk's interface does. I've definitely used the DECTalk singing manual a time or two, and it's mentally rather taxing, as the video creator can easily confirm. And there's one more piece of heritage. Again, I can't confirm who borrowed from whom or to whom credit was given, but several of the Macintalk or Plaintalk voices owe their inspiration, if not their existence, to the DECTAlk and the work of Mr. Klat. The easiest example to locate for most of the populace is going to be on iOS or iPadOS. The recent update to iOS 16 gives us more examples, but even older iOS versions have access to Fred, a voice that's been around for a long time. Giving Fred a listen certainly elicits strong DECTalk vibes, if nothing else. I wish we had a voice more like Perfect Paul, whereas Fred is a bit more like Frail Frank, but honestly, even having Fred around is a nice bit of nostalgia. DECTalk is a serious icon of computer speech synthesis technology, and I'm glad people are starting to remember how amazing it really is and what a truly inspiring legacy and lineage it has. Thank you for the wonderful video.,

[@stasprze1685]

I my self still to this day use a Klatt based synthesizer last updated in around 2003 called ETI Eloquence. That particular synth has-been through lots of owners, to the point where untill june this year I wasn't even sure which company owned it. If you've ever worked with IBM ViaVoice in the 90s and early 2000s, you've probably heard it, as it was briefly forked by IBM and went by ViaVoice TTS. Many blind people, including me, adore this synthesizer to this day simply because of how responsive it is at fast speeds. Synthesizers like DECtalk tend to slurr at fast speeds which makes them hard to understand. Eloquence is so widely used by the blind community to the point where Apple has introduced this 20 year-old synthesizer to iOS 16 and mac OS 13. That choice was widely applauded by people like me as I just can't stand using natural synthesizers for longer than a few days without missing my trusty Eloquence at this point.

[@davidkessler6677]

What a fantastically thorough and entertaining look at DECTalk. In the summer of 1984 I worked on porting the precursor software of DECTalk; MITalk or KlattTalk, from a VAX computer system to a PDP-11 (or vice versa. It has been a very long time). I was able to hear DECTalk sing that summer in a speech lab at Indiana University. The song I remember most was Tom Lehrer's Poisoning Pigeons in the Park. Speech synthesis was very exciting to me then and still is interesting to me. Prior to that I had an Echo II speech synthesizer in my Apple II+. Thanks for posting this!

[@Lord_Nightmare]

I'd love to know if any source code to either MITalk or Klattalk has actually survived, I've been unable to find anything myself.

[@Lord_Nightmare]

I haven't even been able to find a binary copy of either MITalk nor Klattalk, so if you have any leads...

[@kektuss]

AEIOU

[@ThomasGrillo]

I still have one of the 90s vintage Dectalk devices. It's the size of a paperback book. Was one of a box of a thousand units that were going to be destroyed by the state, if nobody asked for them. I was lucky to get one, for free!. Most of my other blind friends jumped on that deal. By the way, you really need to be working in commercial broadcasting. Your voice is the kind PDs look for, in both radio, and television. I know, because I've worked in commercial radio. Thanks for the video. Really enjoyed the history of this device.

[@UpcomingJedi]

What a waste to destroy them.

[@gigglesseven]

moonbase alpha still uses the DTC 01 command sets. check out the singing astronauts naenaenaenaenaenaenaebaettmaenn

[@anon_y_mousse]

I love those old speech synthesizers. They really are the best. Also, I've still got your old channel name as the bookmark for your video page, I should probably update it, maybe make an addendum that it's now Tech Tangents. I keep thinking who's Tech Tangents every time I see it until I realize, oh yeah, he used to be AkBKukU.

[@davidp7414]

Growing up near the DEC campus, I would love a history of DEC. As a kid, many of my friends parents worked there and also Wang. So hard to believe these giants are gone.

[@Jason_Quinn]

RIP Alta Vista. They were googed to death.

[@KRAFTWERK2K6]

RIP Wang Computers.

[@paulstubbs7678]

Years ago I built a speech synthesiser from a chip-set sold by Tandy (Radio Shack) For an idiotic project, I 'made' a telephone ringer out of it, as in the old land line phones, not the modern cell junk. I connected a hayes compatible modem to the phone line (which solved any legal woes about connecting to the line). These modems send out the text 'Ring' when there is incoming ring/call. this was fed into the speech synth, then from there into a PA amplifier. So when the phone rang, a large horn speaker started yelling 'Ring, Ring Ring' It worked, it was really bad, after the second call I pulled it down, the synth has not seen any use since....

[@mikolasstrajt3874]

Inputing individual phonemes also unlocks other languages. I experimented with similar Commodore based program and learning it how to pronounce things in Czech language. It worked as a filter - I had Czech text written as normal, my magic filter which translated it into list of phonemes and original commodore program to turn it into sound. Impractical but very funny.

[@martinbcooper]

Strange urge to go and watch the 1986 classic movie, Short Circuit...

[@Pest789]

One of those used to rat people out for skipping classes at my highschool by calling parents at home. I think the only time it called my house, I was the one who answered the phone, luckily. Just before I dropped out, I had a summer job at the school district datacenter writing a program used to validate the attendance records from a printout after the cards were scanned. Digital made some cool stuff.

[@CommodoreFan64]

Truly great stuff with all hard work you put into this video, but the moment I heard DECtalk speak It took me right back to watching Star Trek: TNG as a kid, and to the scene in the Holodeck where Steven Hawking guest stared, and was playing cards with Data, and Mr. Hawking smiles!! 👍👍

[@ncot_tech]

Cool, you managed to get Stephen Hawking to do a voiceover. Wait… he… hmm… I guess if anyone was going to figure out time travel it’d be him 😉

[@MrHack4never]

4:38 I love how YT's automatic subtitles thinks that it's saying "d/ck talk version 1.8"

[@KRAFTWERK2K6]

Paul, Rita, Dennis, Frank, Ursula & Kit all still sound cooler than all the "natural" text to speech voices we have today. Dunno why but i just always preferred the robotic synthesized text to speech voices so much more but i do because i just love to hear a synthetic voice that does not hide its synthetic nature. For the same reason i love vintage CGI animations so much. I just wish there would be a modern multiplatform based software version for modern operating systems or even a standalone handheld Text to Speech computer that uses the DecTalk Speech Synthesis. Even the VOTRAX SC-01A would be great to have as a modern standalone Speech Computer module. However i still prefer DecTalk. There have even been SAPI3 Text to Speech voices from IBM that sounded almost like that. Same voice but in different languages. The german version of DecTalk sadly wasn't perfect and had a bit of an accent and some misspronounciations and wrong vowls but it was still kinda usable. But just not as good as the english one.

[@pandeomonia]

Dang. All the robots in the movie Short Circuit musta been voiced by this fella. As you were going through the list I was like "yep, that one's in the movie too". Very cool!

[@KRAFTWERK2K6]

Yup, that's what they used for them. At least in the original. In the german dub they got voiced by actual humans with robotic sound effect.

[@RandomInsano2]

Great video! I’m reminded of Short Circuit’s serving robots from the intro. One thing though, it seems there was quite a bit of low frequency noise that could be removed with something like Audacity.

[@KRAFTWERK2K6]

It was also massively used for all the speaking 2015 gadgets in "Back to the Future Part II".

[@nochan99]

Fun fact: the "Dennis Klatt" speech synthesizer that was used in DECTalk is available as open source. It is a tiny but very elegantly written C program that takes a stream of 19 parameters to produce an audio wave output. All you have to do is feed it with the 19 parameters and it will vocalize it.

[@KRAFTWERK2K6]

Wish someone would give that a proper GUI and release a lightweight standalone version for Linux and Windows.

[@tfmorris02181]

It's not actually open source, but pirated. The source code was released without authorization by a former developer. Lack of a license hasn't prevented people from using it though.

[@jeromethiel4323]

This brought back memories of my high school days. I was given a TRS-80 speech synthesizer by a high school teacher. I thought it was a loan and gave it back. He later told me he intended for me to keep it, but i was too young and stupid to do so. That device was way simpler than the DEC talk, but it worked similarly, and probably had some of the same tech in it. But you HAD to send phonemes to it for it to work, and it could not sing. But still, for the time it was amazing tech.

[@winstonsmith478]

Fascinating. Some kind of hardware emulating it, please. Also, no HAL 9000 signing "Daisy"?

[@NPrescott]

I have an original DECtalk Express from 90s. It was actually purchased new for a 286 DOS computer then used later with a 386 and Windows 3.11. It is still working today and after all these years, the original flat cell battery just recently failed. Can't say that for new products! I still hear it in my head "DECtalk Express is running version 4.2c....external power on" or "internal power on" and gives you the percentage of life left. The DECtalk really is an outstanding piece of hardware. Have you seen the "Type n Speak"? Those are really interesting computers without screens that have this same voice or simular chip built in. The deep voice on the type n speak is named Rocco "rock-oh" 😅

[@WX4CB]

you missed out the complete obvious one......... (yes i know it wasnt done on this but would be worth the giggle factor)....... "Greetings professor Falken... How about a nice game of chess?"

[@bad.sector]

That is easily one of your best videos! Thanks for that insight into that chunk of a speech synth!

[@identitymatrix]

No way understandable TTS was already possible in the early 80s. 😂 What a crazy world we live in.

[@Sorana914]

have ya'll heard the moonbase alpha synthesizer music covers? they are terrifyingly good.

[@paulwratt]

As a side to this, I had a loner Bondwell portable CP/M unit for a while in 1986 and spent many hours with _that_ speech chip :)

[@ExperimentIV]

i LOVE dectalk! wish i could get myself a working unit somehow 😭

[@cbmeeks]

Amazing video. I love 80's speech synth. Hopping over to the emulator, I enjoyed having it speak some highly intelligent phrases like "farts" and "butts". LOL

[@AKATenn]

The computerized voices in the movie Short Circuit.

[@timothystevenhoward]

Ok I think this is the right time to ask. Does anyone remember a "Dr Sbaitso" text to speech demo in the 80s? I remember it so clearly. Sounded just like this device.

[@josephkanowitz6875]

ב''ה, not sure the origins of the synthesis engine, but fairly certain that was packaged by Creative Labs to demo the SoundBlaster (possibly the now classic and venerable SB16) or at least named after it.

[@datajake1999]

Dr Sbaitso was from the early 90s, and used a different speech synthesizer (Smooth Talker from First Byte).

[@rommix0]

@@datajake1999 That is correct. SmoothTalker version 3 to be exact.

[@pablopicaro7649]

Commodore 64 had a cartridge that sounded like this in about 1984 or 1985

[@asdfasdfasdf12]

why is old digital technology so damn fascinating, I love it. I need a big house or storage facility, to buy ebay empty from this stuff

[@intelbreak]

What about using OpenAI's api to send the output of ChatGPT to the DECtalk? Old meets new in a logical useful way. ♥

[@Adam_Boots]

Nice Weird Al reference. 'Don't go makin' phony calls Please stick to the seven-digit numbers you're used to'

[@UpLateGeek]

That was pretty cool! Although you should've had the DECtalk do the sign off at the end to bring it back around to the setup at the beginning of the DECtalk presenting the video.

[@m4rgin4l]

Hello DECTalk, this is dog.

[@BreakingPintMedia]

Rough Rita is to blame for the excessive yellowing on 80s and 90s computer hardware

[@fischX]

Sounds exactly like the say command in Amiga OS - 68000 processor checks out I bet it's the same software

[@nticompass]

Admit it, the first thing you did with this was make it say dirty words 😛

[@Wekulu]

this was super cool and its taught me a lot about how impressive old tech is even compared to something fairly recent. i work on utau a lot (which i have videos of, if you dont know what it is), and even after using paul's voice for a song, its made me realize something interesting-- his data is actually really small! a regular utau voice bank is almost double or triple the size, and is capable of far less on its own compared to this which can talk and even sing with relative ease. of course, the medium its used though is different, but i think thats what makes it more unique

[@dotz0cat]

I was thinking of vocaloid and utau the entire time watching this

[@DominatorHDX]

LOL what a coincidence! Yesterday I found this old windows text to speech program in some lost directory on my PC called Microsoft Talk It for kids from 1997 (still works in Windows 10) and many of the types of voices it can do sound exactly the same as the DECtalk 😁 PS: Rita sounds like she smoked way too many sigarettes 😅

[@datajake1999]

Talk It! actually used the SVTTS system.

[@DominatorHDX]

@@datajake1999 Ah I didn't know that.

[@ehudgavron1]

Ah, you watch one video and it recommends another one... and this one was great as well. I did kernel hacking on VAX/VMS and then Alpha/OpenVMS. DECTalk (DTC01) preceded DECVoice (DTC04) but the concepts were similar. When you say it has relevancy today you are right. Stephen Hawking. It was revolutionary in its measly requirements for pre-processing TTS, and its intonation worked well for most English sentences. That is even contrary to today's smartphone TTS systems that need "a little help." For example, my car bluetooth speaker thingie announces the name of the device being connected. My phone being called "Ehud2" was being mal-pronounced so I had to rename it to "Ay-Hood Too". DECTalk would have handled that much better. Cheers, Ehud (the original, not 2)

[@TastyBusiness]

Man, this is a great break down of what the DECtalk can do. Makes my old Votrax look like a toy by comparison.

[@creepingnet]

I have a similar device with all of the same voices. Its a Words+ Commpac System 2000 I got with a touch screen equipped NEC Versa M75 486 Laptop. Words+ made the speech synths for Stephen Hawking. Words+ was founded in 1981 by Walter and Virginia Woltocz and worked with Hawking around 1985. The system 2000 unit is from the mid-late 1990s. Cool finally knowing who created all those voices. IIRC the internal synth uses a lot of DEC branded silicon. Got to wonder if DEC and Words+ had a close working relationship.

[@BollingHolt]

Spectacular video. I LOVE old Digital-branded equipment.

[@Qardo]

Could have let the DECTalk do the whole video, lol. Though that would be a bit meta.

[@mckensieparker6414]

Wow! Fun fact, dectalk is still used in computers today! Also, another synthesizer you should check out called eloquence.

[@steveg5122]

Nice Bell Atlantic Phone, i have a little nostalgia for that company, it was the phone company from when i was a kid until i turned 11.

[@PiggyPorkchop]

I grew up with Perfect Paul from NOAA Weather Radio to the TTS on one of my first computers its associated with quite a few childhood memories. I occasionally let it read emails and articles to me just for nostalgia kicks.

[@infango]

Black Box v8 for C64 was able to synthesize polish language .. hours of clean fun for 14 yo me ;p

[@Mrshoujo]

Missed chance for a DECtalk Rickroll.

[@ViewpointProd]

Love the voices of this thing, used in 1986's Short Circuit, specifically for the voice of the S.A.I.N.T.S (Possibly Paul or Harry with pitch modifications) & the Catering/Coffee Robot (Tweaked Betty voice).

[@Fuzy2K]

Coffee service coming through. Please watch your step.

[@adimifus]

Yep, they mostly used Huge Harry (monotone, no inflection) but Perfect Paul was also used in introduction/demonstration scene.

[@ViewpointProd]

@@adimifus interesting that during that entire demonstration, its only number 1 speaking, yet his voice changes lol. at least, the subtitles SAY its number 1, as he's the first to come on stage and speak. then the voice changes during the tonic scene.

[@adimifus]

@@ViewpointProd I guess its hard to say what robot is "speaking" when. Since the cocktail scene involves #2 and #3, I would assume its one of them talking during that part. Later on in the movie, all three of them use the Huge Harry voice, so I guess they do change. Maybe they switch Perfect Paul when they're trying to be more "personable" like during the demonstrations at the beginning. Or maybe I'm just reading into it too much... 🤷

[@paulwratt]

I've mentioned this on some other peoples videos about "speech synthesizers" - there is an article in Byte Magazine (is it May 83?) with circuit board layout for a Vortrax speech chip, connects to the printer port, and code to use it from GW-BASIC - I built one in around 1996, but it appeared the Vortrax chip I had was dead, and although I could buy one from Germany at the time, it was cost prohibitive for me (on the other side of the world, over 200 dollars in local currency, not including delivery).

[@tfmorris02181]

The other expensive component, which looks like it might be covered/obscured in your unit, is a TMS32010 digital signal processor and associated RAM. The DTC01 had not one, but *two*, expensive processors. I had the first internal beta unit and used it to write in email reading application for DECmail (naturally) in the early 1980s.

[@SteveChisnall]

3:55 I saw 4 empty chip-sockets on the motherboard. Is there anything one could install there to upgrade the DTC-01's specs any? Also, would the machine still work if one were to swap out the Motorola 68000 with a 68010?

[@ROOKCHUCK]

You need to look for Snoopi Botten in Ohio. A disabled person who has developed DECtalk to sing. He has made so many tracks in DECtalk. He is very knowledgable reegarding DECtalk.

[@rustyfloorboards]

Probably not very common knowledge but this speech synthesizer was used for some of the voices of the Robots in Short Circuit during the beginning of the movie. And though I can’t confirm it maybe the scene in Bttf 2 during the “dry mode” scene.

[@ViewpointProd]

The coffee robot uses the betty voice, no idea on the Others though, it seems they may have had the pitches changed

[@RiosTubeChannel]

aeiou John Madden auauauauauauauau

[@Vamptonius]

Dennis had to rush to finish recording for his mate Stephen Hawking because he was dying. Many times over the years Hawking was offered improved voices with his native British accent but he always refused with "I already have a voice, that of my friend Dennis Klatt."

[@josephtotter7484]

It was great but super expensive and i am sure you know there was a pc bus card version too. . For my college EE Projects I used the digitalker chip set. It had a limited vocabulary but sounded way better. "This is digitalker" :)

[@breakbumper]

You missed a trick there. The DECTalk could have easily Rickrolled us!

[@CrkdLtrN]

I was hoping you would make it say "Your jacket is now dry" from Back to the Future 2. Curious if they used the DECtalk to do that part in the movie?

[@jaut-76]

I really want one now. More than when I saw it on stream

[@ronny332]

I'm thrilled, what an enjoyable summary of DECtalk. Thank you!

[@tcpnetworks]

We had 16 of these units hosted on a pair of VAX VMS machines, that ran our automated payment and information systems at our council. The whole thing was in a pair of racks - and worked 24/7. It allowed you to pay parking fines, rates, everything. They needed constantly needed maintenance. I got very good at writing-out ROMS and swapping things over. We decomissioned the whole thing in 1998 and sold it all off - replacing it with a couple of Windows PCs and simpler modems.

[@benjaminpmartin]

This piece of kit really speaks for itself 😆

[@petevenuti7355]

Waa waa waaaaa ba doomp tisss (I hope you're not listening to this on one of those old synthesizers that third wa might have sounded like the computer was gargling underwater, as it crashed)

[@sxftie.len-P]

you don't need to buy the whole thing, we have a DECtalk UTAU voicebank now for both eng and jp ^^

[@LarryRobinsonintothefog]

68K processor, awesome. I like getting a voice synthesizer to sing.

[@paulmoscatt6529]

I used to have modem access to work. We had a Dectalk. So it was pretty easy to type in who and see if anyone was there late at night. If there was, the fun began. As you could send commands to make it speak from home and know who to address.

[@SmeddyTooBestChannel]

aeiou

[@LaskyLabs]

Dennis H. Klatt died far too young, I believe he deserved to see two things. One, speech synthesis get better. Two, his Klatt Talk system still being in use to this day, and Mr.Hawking continuing to use it as well before his passing.

[@detaart]

DEC was on another level.

[@andreassjoberg3145]

I remember playing with it 1993. I preferred the Amiga speachsynth, but it was better than the Soundblaster speachsynth.

[@RealZomBiE8192]

You reminded me all i almost forgot about this lovely thing 😀

[@RealZomBiE8192]

Our rare C64 thing sounded lovely with SID 😁