Saving Old Voices By Dumping ROMs

Some people collect stamps. Others collect porcelain miniatures. [David Viens] collects voice synthesizers and their ROMs. In this video, he just got his hands on the ultra-rare Electronic Voice Alert (EVA) from early 1980s Chrysler automobiles (video embedded below the break).

Back in the 1980s, speech synthesis was in its golden years following the development of TI’s linear-predictive coding speech chips. These are the bits of silicon that gave voice to the Speak and Spell, numerous video game machines, and the TI 99/4A computer’s speech module. And, apparently, some models of Chrysler cars.

IMG_0695We tracked [David]’s website down. He posted a brief entry describing his emulation and ROM-dumping setup. He says he used it for testing out his (software) TMS5200 speech-synthesizer emulation.

The board appears to have a socket for a TMS-series voice synthesizer chip and another slot for the ROM. It looks like an FTDI 2232 USB-serial converter is being used in bit-bang mode with some custom code driving everything, and presumably sniffing data in the middle. We’d love to see a bunch more detail.

The best part of the video, aside from the ROM-dumping goodness, comes at the end when [David] tosses the ROM’s contents into his own chipspeech emulator and starts playing “your engine oil pressure is critical” up and down the keyboard. Fantastic.

Thanks [William] for the tip!

40 thoughts on “Saving Old Voices By Dumping ROMs

  1. This is a lot of fun, particularly the keyboard manipulation.

    I hope I’m not dating myself too much by saying that the original versions of these astoundingly annoying annunciators died a horrible death in the marketplace for a reason. (“So last century” according to my wife). Driving one of these cars was like having Harry Mudd’s android (lower-case “A”) wife in the back seat.

    https://youtu.be/2FGgHEKko3E?t=56s

    1. I dunno.. I’m sure it was annoying at the time, but I’d gladly take the old-school robot voice over today’s incessant beeping for no apparent (or good) reason. I’ve had numerous occasions where I’ve almost put my fist through the dashboard in a rental car because of the constant beeping when parking or when the temperature drops slightly below freezing. Honestly really, really annoying, and the obstacle avoidance systems are just plain ineffective and distracting.

    2. I don’t know if that was the real reason. The quality was lacking. We use GPS now and don’t mind them talking to us. Yes they can get annoying. Usually when they repeat a direction when you’re caught in certain circumstances. lol I did like tinkering around with some of the voice synthesizer chips.

      1. I find the GPS generally speaks too much. Usually I’m trying to talk on the phone (via the same speaker system), which does a great job of filtering out the GPS voice to the person on the other end of the line… however the GPS is usually “In 3 km, at the interchange, keep right and follow signs to A11 City, City, City” …which first really interrupts the call and takes forever to speak, then she reminds at 2km, 1km 500m and so on. By the turn I want to smack her! And worse, because the car filters out the GPS voice when taking audio for the caller, the caller has no idea they’re being spoken over…GAH!

  2. I vividly recall renting a car on an important sales call back in the 80s. We were treated to an upgrade to a fancy car with a synthesized voice, which had similar messages about open doors, etc.

    Imagine our embarrassment when we picked up our very important (and also very large) client. As she sat down in the rear seat, the self leveling suspension adjusted for the weight difference.

    The synthesizer calmly announced NOW LEVELING THE LOAD as she settled in for the ride.

    Needless to say, we were all speechless at that point.

  3. My brother owned a 1984 Dodge Daytona Turbo, and it had one of those. It was a neat gimmick for a few days, but he got annoyed after a while and I figured out how to turn it off (toggle switch inside the glovebox). There were issues with the car, though (he had bought it with 140k hard-driven miles on it), so he replaced it with an ’81 280ZX. It also had a voice module, but this one was female, and was very clear and human-like. I’ve wondered if it wasn’t a digital recording of an actual voice, with a way to trigger the appropriate messages.

    1. This is funny. The above video reminds me of a tale I once heard about someone who bought a couple year old Chrysler car with the voice norifications for too good to be true low price Here in British Columbia. The punchline was that this car was from originally from Quebec and spoke French, the owner however did not and after a year or so of it constantly saying “fermer la porte” (close the door) or similar he had enough and resold the thing. Passing along the curse of the francophone LeBaron to another un suspecting owner

  4. Sega arcade games around 1981 used the G80 system, which used the General Instrument SP0-250 Orator chip:

    http://bitsavers.trailing-edge.com/pdf/gi/speech/General_Instrument_-_SP0250_Applications_Manual_-_Jun1982.pdf

    I’ve been working on a C development environment for this hardware system.

    I would really like to know if there is software out there that would allow someone to record and replay for these devices. I believe the data is LPC encoded, according to the manual.

    1. Thanks for this link I’m thinking Good as Gold. Last year I wanted to play with talking Arduinos for some reason but refused to pay $60 for a speech shield.
      To that end I expect trolls to post aliexpress links to some $9 rs-232 to speech board.

  5. My Aunt had a 1987 Dodge with the voice alerts. A. DOOR. IS. A. JAR. Yes, like that, with a definite pause between A and JAR. Made me wonder what kind of jar? Mason? Jam?

    The TI speech chips (as used in the Home Computer Speech Synthesizer) had a base voice which could ‘simply’ be fed LPC code. There are also ADSR (Attack, Delay, Sustain, Release) and other parameters that can be used to carefully craft the voice you want. It should be possible to interface one of those speech synthesizers to just about anything. They’re an input only device, output is via an analog audio line which went back to the computer console to be mixed with the computer’s analog audio out.

    The other way was to digitally record a voice or sounds to replicate. The recording had to be a specific bits per sample, mono etc. I don’t know what TI used originally but in the early 90’s they developed Windows 3.1 software called QBox Pro. Download it and a DLL it requires here http://atariage.com/forums/topic/153704-ti-994a-development-resources/

    That analyzes the audio and creates LPC data to feed to the synthesizer to use to replicate the original audio.

    Other 1980’s microcomputers and game consoles used digitized audio directly. The Odyssey^2 voice module contained a set of prerecorded words and some phonemes to construct words not in its vocabulary. The IntelliVoice module was just a DAC, reading digitized sound samples in the cartridge ROMs when commanded by the game code.

    The IBM PCjr had an optional speech module which used the TI speech chip (its built in tone/noise chip was also the same as in TI’s computer) but IBM’s implementation was nowhere as good as TI’s so very little PC Jr software used speech.

  6. It took this video for me to realize how dead, dead, dead talking cars were. Another thing to look for at Pull-A-Part. I laughed so hard at the keyboard sample–three hundred dollars to buy this, and a homemade jig to read it. Just because. Yep, he’s one of us.

  7. ISTR Clarkson describing the Austin Maestro with the talking dashboard; a colleague test-drove one and crashed it, as he was laying in a ditch “upside-down and on fire” there was a moment’s silence and then the car said “…oil pressure low.”

    Personally I believe anyone who deliberately adds speech or any noises, bongs, dings, swooshes or tinkles to any piece of technology needs stabbing in the eye with a fork.

      1. Black box recording. Chinese pilots discuss, what means “pull up” repeatedly as they fly level into terrain. Mountains rear out of the sky. A flashing arrow would have been better, though English is the official-air-com worldwide.

        Audible notices are easily missed unless nagging repeatedly or are heard separate form any other audio source, otherwise they will be intrusive to the sound experience. Simple visual warnings are enough. Thus the tinny voice of a Garmin up on the dash can stand out without intruding on sensible aware listening and driving. Deaf people here only have to have an outside mirror, which is standard anyway.

Leave a Reply to BuffaloCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.