./pizza
There’s a very specific feeling I get when in a classroom at night. I usually only end up there for some kind of special occasion: the last math team practice of the season, a guest lecturer in college, or a presentation night for finals. Even if there are no windows, the fluorescent lights seem a little brighter, the room a little warmer, focus a little easier. I’d like to imagine that Donald Sherman was feeling somewhat similarly as he sat in front of a green phosphor CRT terminal and a telephone handset perched on top of a wooden box. He was surrounded by a group of at least a dozen people who had filed into the Michigan State University AI Lab on the night of December 4th, 1974 to watch Sherman order a pizza.
To understand what the hell is going on, we need to travel back another 4 four years to 1970. Richard T. Gagnon was an electrical engineer employed by Federal Screw Works1. FSW is a screw manufacturer: one of the many companies which sprung up in the early 20th century to support the developing industrial economy of south-eastern Michigan2. Gagnon experienced problems with his vision, and worried that he may go blind in the future. This inspired him to try and create a speech synthesizer that would allow him to use a computer without being able to see a screen or printout. Speech synthesizers had existed since the famous Bell System Voder in the 1930s. They had progressed to digitally controlled phoneme generation by the 60s3, but they were all one off laboratory experiments. Rather than pushing the state of the art, Gagnon focused on simplifying the technology to make it commercially viable to manufacture, while still performing well. The recent advances in transistors and digital methods surely helped with this. After a successful prototype, he commercialized his invention by convincing his employer (who again, made machine screws!) to create a division to fund the manufacture and sale of the speech synthesizer which was now called the Votrax.
The Votrax VS4 was the first commercially sold phoneme based speech synthesizer, consisting of 3 circuit cards and a power supply in a handsome case which sold around 100 units beginning in 1972. One of those units went to the Michigan State University AI Lab. As this was Gagnon’s alma mater, and close to FSW, I find it likely that there was some kind of collaboration, but I can’t find info on the specifics. Either way, their goal was to reverse Sherman’s original design intent, and help an operator use the computer to speak, rather than have the computer speak to the user. To demonstrate their progress, they likely wanted some kind of task that could either work or not, to exercise the technique past simple communication. Something that requires some nuance, but not too much for the admittantly extremely early tech. Something, like ordering a pizza.
Sidenote: I feel it’s important to note that pizza delivery would’ve been a relatively new phenomenon at this time. Pizza itself was introduced to Americans in the 1930s, and delivery of it was popularized by Domino’s in the 1960s4, putting this demo roughly 10 years into the age of pizza delivery. Detroit actually has its own pizza tradition, with square pies baked in high-sided steel pans originally manufactured to be used as drip trays in car plants5. Unfortunately, my scholarship was not able to determine the exact style of pizza ordered.
Donald Sherman ended up in the hotseat for this demo, as he had a condition known as Mobius Syndrome which restricts facial movement6. My guess is that he advised the development of this tool or was fully the inspiration for it. It went, to be completely honest, not incredibly smoothly. I’ll spare you the play by play (watch the video if you can) but it’s fair to say that it was the still early days for electronic speech. Despite all that, after one hangup, the pizza got ordered, and Sherman enjoyed his pie. This is not only pretty concretely the first use of a speech synthesizer for accessibility, but an extremely early example of computers being used in realtime to assist an individual with a task of any kind. Any piece of digital transformation, from the magnifier on your phone, to Google Maps, to AACs used by my autistic comrades are under the same tree of assistive technology with roots in these kinds of developments.
Unfortunately, I’m not sure what happened to the labs’ technology after this, but Votrax would continue onto a few decades of operation. It was spun off from Federal Screw Works, and evolved into making ICs for use in assistive technology, automated telephone systems, and amusements like video games and pinball machines. As most companies, it would eventually decay through a series of mergers and acquisitions before the brand slowly faded from presence by the turn of the century.
“NMAH Smithsonian Speech Synthesis History Project (ss_votr.htm).”: https://web.archive.org/web/20110604022256/http://americanhistory.si.edu/archives/speechsynthesis/ss_votr.htm https://federalscrewworks.com/about/ ↩︎
J. L. Flanagan, “Synthesis of artificial speech,” US3268660A, Aug. 23, 1966 ↩︎
“Tom Monaghan Domino’s Pizza THE PIONEERING PIZZA-DELIVERY CHAIN I STARTED ALMOST DIDN’T MAKE IT OUT OF THE OVEN. - September 1, 2003.” https://money.cnn.com/magazines/fsb/fsb_archive/2003/09/01/350799/ ↩︎
Adam Ragusea, Industry + immigrants = Detroit-style pizza, (Nov. 07, 2022). https://www.youtube.com/watch?v=iY24pIHjT7s ↩︎
“Tech Time Warp of the Week: Return to 1974, When a Computer Ordered a Pizza for the First Time WIRED.” https://www.wired.com/2015/01/tech-time-warp-pizza/