Vox Ex Machina

Roman Mars:
This is 99% invisible, I’m Roman Mars.

Roman Mars:
In 1939 an astonishing new machine debuted at the World’s Fair in New York City. It was called the Voder, short for ‘voice operating demonstrator’.

Delaney Hall:
It looked sort of like a futuristic church organ.

Roman Mars:
That’s our own Delaney Hall.

Delaney Hall:
An operator, known as a Voderette, would sit at a curved wooden console. Behind her, there was a huge wall with an art deco image of a man’s face with these spirals of curly hair. His mouth was a giant speaker.

Roman Mars:
The Voderette would place her hands on two keyboards in front of her each with five or six white keys. She’d use her feet to work two pedals down below, but instead of musical notes, the machine produced a voice.

Vocoder: Greetings everybody.
Product Demonstrator: Now will you have him, repeat that in a high voice.
Vocoder: Greetings everybody.
Product Demonstrator: And now in his best voice.
Vocoder: Greetings everybody.

Roman Mars:
This is from a demo recording of the Voder, produced in 1939 around the same time as the World’s Fair.

Delaney Hall:
Each of the keys on the keyboard controlled a particular frequency band. One foot pedal controlled pitch, and the other foot pedal controlled whether the sound would be muffled or crisp. Operating this machine required incredible precision and skill. Voderettes would train for up to a year to make the voter actually talk like a person or sing like a person.

Product Demonstrator: Suppose you sing a song for us. Will you?
Vocoder: Yeah. All right.
Product Demonstrator: Well, how about Auld Lang Syne?
Vocoder: Auld Lang Syne? Okay. Should old acquaintance be forgot and never…

Roman Mars:
And remember this is all happening, totally live. The Voderette is not triggering pre-recorded words. Instead, the keys and pedals of the Voder imitate the effects of the human vocal tract, producing the most basic building blocks of speech. The Voderette is playing them in an intricate sequence and she’s actually synthesizing speech real-time.

Delaney Hall:
The crowds at the World’s Fair went crazy for this talking machine.

Jon Paul:
It created a sensation.

Delaney Hall:
This is John Paul, he’s an engineer, inventor, and historian.

Jon Paul:
It was the first time there had ever been working speech synthesis, anywhere. People had no conception that you could even do it.

Product Demonstrator: Here’s Voder’s imitation of a cow.
Vocoder: Mooooooo.

Roman Mars:
The Voder was invented by an engineer named Homer Dudley.

Jon Paul:
And Homer Dudley’s area of expertise was speech science.

Roman Mars:
Dudley worked at Bell Labs, a research facility that belonged to AT&T, and during the 1920s and 30s, Bell Labs was doing all kinds of research into the human voice. How to synthesize it electronically, how to compress it so it could be sent across enormous distances quickly and cheaply, how to encode and disguise it, and all this research helped AT&T’s basic goal, which was to improve the phone system in the United States.

Delaney Hall:
The Voder was a novelty offshoot, related to this research into speech, but it was closely connected to a number of Dudley’s inventions at Bell Labs that still shape our world today.

Jon Paul:
A lot of that basic research turned out to be things which have impacted our modern technology and world, enormously.

Delaney Hall:
Stuff that became critical to the development of digital media, for example, and on top of all of that, Dudley’s inventions helped us win a war.

News Report:
“December 7th, 1941 a date which will live in infamy United States of America…”

Roman Mars:
Just a couple of years after the Voder’s debut at the World’s Fair, Japan launched an early morning attack on Pearl Harbor in Hawaii. It came as a profound shock to the US and the next day President Franklin Delano Roosevelt announced the country would be entering World War II.

President Roosevelt:
“With confidence in our armed forces, with the unbounding determination of our people, we will gain the inevitable triumph, so help us God.”

Delaney Hall:
World War II would be the most widespread war in history. The US would be fighting in Europe, and the Pacific, and parts of North Africa all at the same time, and it was a huge challenge to coordinate troops across that much territory.

Jon Paul:
It was obvious the only communication possible would be by radio and that radio was of course insecure. The enemy is listening.

Roman Mars:
In other words, there wasn’t really anything besides radio that could stretch across the distances involved, but radio is easily intercepted, and so…

Jon Paul:
The problem of encryption and secret messages of military importance became more and more serious.

Delaney Hall:
At the very start of the war, the US military still relied on an old school piece of technology for its sensitive conversations. It was called the A3 Scrambler. It worked by scrambling voice frequencies, swapping high frequencies for low and low frequencies for high, garbling the sound. Here’s an example of what it might’ve sounded like.

[scrambled audio]

Roman Mars:
But it was easy to decode these scrambled conversations.

Dave Tompkins:
The A3 was being compromised, and it was not a secure system.

Delaney Hall:
That’s Dave Tompkins, who’s researched the history of various speech synthesizers and their connections to music.

Dave Tompkins:
German intelligence was… they’re essentially de-scrambling some of these conversations between FDR and Churchill in real-time.

Roman Mars:
And the US military knew the German code-breakers were listening in, so…

Jon Paul:
The US government gave Bell Labs the mission of creating an unbreakable speech encrypter.

Delaney Hall:
And Bell Labs went to our old friend Homer Dudley, the guy who invented the talking machine of the World’s Fair. They said, “Dudley, we need you to figure this out, really fast.”

Dave Tompkins:
They were under tremendous wartime pressure to produce this thing quickly. I think the whole design took about a year.

Delaney Hall:
Thankfully Dudley had already been working on technology related to this problem for years. It all went back to the Voder, and to another related invention of Dudley’s called the Vocoder.

Roman Mars:
Music nerds might know the term vocoder because it’s the great grandfather of the machine, you hear in all kinds of pop music today.

Delaney Hall:
Vocoder was short for ‘voice encoder’, and the machine could break down a human voice, separate it into its basic components, and then compress and transmit those components via short wave channel.

Dave Tompkins:
It could transmit the minimal amount of signals required to reconstruct that message at the other end.

Delaney Hall:
This process was important, revolutionary actually, because it allowed the human voice to be digitized and compressed and sent across big distances.

Roman Mars:
But the vocoder didn’t fully disguise the voice. The transmission could still technically be pulled out of the air and decoded, so Dudley and his team had to add a layer of unbreakable encryption, and that’s how the vocoder became just one small part of a much larger and more intricate apparatus.

Delaney Hall:
A secure communication system that would allow allied military leaders in strategic locations around the world to talk in total secrecy.

Roman Mars:
It was called Project X, AKA the Green Hornet, AKA SIGSALY.

Delaney Hall:
Which was short for nothing really. It was meant to be confusing. Each SIGSALY machine was enormous, weighing more than 50 tons.

Dave Tompkins:
It was a pretty complicated, behemoth of a device. It wasn’t even a device, it was a room in itself.

Delaney Hall:
It occupied about 2,000 square feet and was made up of 40 racks of equipment.

Jon Paul:
It needed air conditioning because a lot of the electronics was so delicately balanced, that if it got too hot or too cold, it would not work properly.

Roman Mars:
In fact, the device was full of so many finely calibrated tubes and gadgets that the military created a whole division of engineers whose sole job was to maintain the machines. The 805th Signal Service Company.

Lt. Donald Mehl:
We were pretty much in charge of the installation and operation of the system.

Roman Mars:
That’s Don Mehl. Before he enlisted, he was an amateur radio operator from Omaha, and he was one of the engineers assigned to work on SIGSALY. He reported for duty in Washington DC and spent his first two weeks in an intensive class.

Lt. Donald Mehl:
Where we learned all the technical details, and the operation, and then it was pretty much on the job training.

Roman Mars:
The first SIGSALY terminal was installed in the basement of the Pentagon, and it was connected to several conference rooms upstairs.

Dorothy L. Mads:
Elegantly furnished. There was wine-colored, thick, thick carpeting on the floor. They were beautiful rooms.

Delaney Hall:
This is Lieutenant Colonel Dorothy L Madsen.

Dorothy L. Mads:
But my nickname is Meg, M-E-G. I was in charge of the global encrypted conference center.

Delaney Hall:
The conference center is where DC-based military leaders would meet to speak over the SIGSALY’s system. So while Don Mehl was down in the basement running the machine, Meg was upstairs coordinating conference calls with the military brass. She hosted everyone from President Harry Truman to General George Marshall. She sat in on their meetings and transcribed the conversations they had.

Dorothy L. Mads:
Any of the men who had the responsibility for the conduct of the war and had to make their phone calls, came to my conference center to do it. Communications is the most important thing and you have to do it with safety and security that you know, nobody else is tuned in on it.

Roman Mars:
The SIGSALY terminal in the Pentagon was connected to a network of close to a dozen other SIGSALY terminals around the world. These were located in the most strategically important places, where politicians and military leaders would need to be talking with each other on a regular basis.

Lt. Donald Mehl:
Algiers and London and Paris eventually, and Hawaii and Guam and Australia.

Delaney Hall:
There was even a SIGSALY terminal based on a roaving ship in the Pacific

Roman Mars:
And one in beautiful, strategically important Oakland, California.

Delaney Hall:
This network allowed leaders in Washington DC to talk securely with any location that had a terminal. All they needed was a short wave radio connection and one other key component.

Roman Mars:
A pair of vinyl phonograph records of totally random noise.

Delaney Hall:
This is where the encryption part comes in because, for every conference that happened over SIGSALY, both the sending and receiving terminals had to have identical records which played the sound of noise.

Roman Mars:
The noise would combine with the voice components as they were transmitted via short wave radio, making it impossible for eavesdroppers to decrypt. On the receiving end, the random noise would be extracted in the voice restored. In cryptography, these records are what’s known as a one time key.

Delaney Hall:
So here’s how they worked. Random noise would be generated, and then pressed on a gold master.

Dave Tompkins:
Normally a phonograph record, you would reproduce the thousands of them from a single master. In this case, they made exactly two records.

Delaney Hall:
The identical and now totally unique records would be assigned a matching code name, and these code names were awesome, like ‘red strawberry’ or ‘wild dog’ or ‘circus clown’.

Roman Mars:
So let’s say president Harry Truman in Washington, DC wants to talk with Prime Minister Winston Churchill in London. Truman keeps one of the two records in Washington, DC.

Dave Tompkins:
The second one would be sent to the station at the other side of the ocean, and then placed on this special precision turntable.

Roman Mars:
That was a part of the SIGSALY terminal. At the scheduled time for the conference, SIGSALY engineer’s in DC and London would tune in to WWV.

International Time Control Station:
“At the tone: 15 hours, zero minutes coordinated universal time.”

Lt. Donald Mehl:
Which is the international time control station, so that we could synchronize our control clocks exactly.

Delaney Hall:
Then they get their identical records spinning on their respective turntables at exactly the same moment.

Roman Mars:
Once the records were synced, Truman would speak into a handset in Meg’s conference room at the Pentagon.

Harry Truman:
“This is the president, Mister Prime Minister.”

Delaney Hall:
This is a reenactment of Truman and Churchill greeting each other when they talk to via SIGSALY to discuss a German surrender proposal in 1945.

Roman Mars:
The signal would go down into the basement where Don was running the SIGSALY terminal. The terminal would digitize Truman’s voice, mix it with the random noise from the record, and then transmit that signal across the ocean, via short wave radio.

Delaney Hall:
On the other end, the SIGSALY terminal in London would reverse the process. It would remove the noise, reconstruct the voice, and feed it through Churchill’s handset.

Harry Truman:
“This is the President, Mister Prime Minister.”

Delaney Hall:
After so much processing and such a long distance, the voice didn’t sound very good.

Dave Tompkins:
Like jibberish orated from the bottom of a barrel.

Delaney Hall:
But it was intelligible.

Dave Tompkins:
You had to kind of train your ear a bit.

Delaney Hall:
Anyway, then Churchill would reply…

Harry Truman:
“How glad I am to hear your voice.”

Delaney Hall:
And the whole process would be reversed.

Harry Truman:
“How glad I am to hear your voice.”

Delaney Hall:
In this way, Truman and Churchill could have a completely secure real-time conversation. Planning, plotting and strategizing about the war. When the conversation ended, the two records – which remember contained the top-secret random noise key – would be destroyed.

Roman Mars:
That’s because if anyone was recording this communication as it traveled over the ocean, the random noise record would be the key to decrypting it. It would allow them to subtract the noise in the same way the SIGSALY engineers did at each end of the transmission. So the records were the most classified and sensitive component of SIGSALY.

Dave Tompkins:
By the end of World War II, there had been a total of about 3,000 top-secret strategic conferences.

Delaney Hall:
In other words, at least 6,000 vinyl records that were pressed, delivered, used, and then destroyed at the end of the call.

Lt. Donald Mehl:
Compared to today, it was rudimentary type technology, but for that time nobody even thought of voice communications in terms of being digitized.

Roman Mars:
And because of SIGSALY, the US was able to communicate with the allies in real-time and by voice.

Delaney Hall:
So if people were listening in to that signal, if Germans were listening in, what would they hear?

Dave Tompkins:
Nothing but random noise, just total white noise.

Delaney Hall:
Wow. I mean, it might’ve been so unintelligible that they didn’t even realize they were listening to anything at all?

Dave Tompkins:
They never knew of the existence of the system, and they never had any inkling that it was an encrypted speech.

Roman Mars:
In fact, that’s why SIGSALY was nicknamed the ‘Green Hornet’ after the popular 1940s radio show.

Delaney Hall:
Because an intercepted SIGSALY transmission sounded like nothing more than a hornet’s buzz

Roman Mars:
SIGSALY was involved in pretty much every major military operation after 1942. It was even critical in the planning of the Manhattan Project and the dropping of the atomic bombs over Japan. Don Mehl was the engineer decrypting and listening in on those SIGSAY conversations, even though he didn’t know at the time the meaning of the code words they were using.

Lt. Donald Mehl:
And I would see this come over and the title was Manhattan Project. Well, I, of course, didn’t know what the Manhattan Project was.

Delaney Hall:
Don didn’t reflect much on the grave implications of those meetings he overheard. He was an engineer doing his best to keep the equipment running.

Lt. Donald Mehl:
You didn’t pay that much attention to everything that was said, you’re more concerned with the transmission.

Roman Mars:
But he knows it would have been hard for the allies to win the war without SIGSALY.

Lt. Donald Mehl:
It’s hard to say what we would do if we didn’t have it.

Roman Mars:
Even if that victory sometimes came at a brutal price.

Delaney Hall:
At the end of the war, the SIGSALY terminals were dumped in the ocean.

Dave Tompkins:
Almost every encryption device was intentionally destroyed when the war was over for security purposes to the extent of destroying the plans.

Roman Mars:
But not before the military developed successors that were smaller and simpler and easier to set up. These new devices still worked on the same principles of encryption as SIGSALY, and they were used for secure communications during the Cold War, the Korean War, and to negotiate the Cuban missile crisis.

Delaney Hall:
A lot of the information related to SIGSALY wasn’t declassified until the mid-1970s, when AT&T sued the US military under the Freedom of Information Act.

Delaney Hall:
Once it was public, technology that originated with SIGSALY went on to entirely new realms.

Roman Mars:
The vocoder, for instance, went on to have a whole new life in music. In the 1970s electronic musicians, like craftwork, began using a smaller version of the machine to create weird robotic vocal effects.

Delaney Hall:
From Germany, the vocoder spread to the Bronx, in Brooklyn where hip hop and electro-funk groups started playing around with it. Now you hear it all over the place.

Roman Mars:
But the vocoders reach actually goes way beyond the robot voice.

Jon Paul:
You’re also using vocoder every time you use your mobile phone.

Delaney Hall:
But of course, in a much, much reduced size. The giant clunky machine of the past now occupies the space of a tiny chip, which compresses speech and allows hundreds of conversations to pass through a single cell phone tower.

Roman Mars:
And a lot of the technology used to transmit media on the internet can be traced back to Dudley too.

Dave Tompkins:
That includes MP3 music. That includes video compression.

Delaney Hall:
So give thanks to Homer Dudley every time you watch something on the internet. All those millions of viral videos would take up too much space if they couldn’t be compressed.

Dave Tompkins:
Everything in our modern digital world of media can be traced back to Dudley’s vocoder and to SIGSALY.

Roman Mars:
In the end, the vocoder kind of came full circle. It started its public life as a silly novelty gag, making animal noises for crowds at the 1939 World’s Fair.

Product Demonstrator:
And here’s a pig.

Roman Mars:
Then like Don Mehl and Meg Mattson, the vocoder enlisted. It did its service, it helped the allies win the biggest war the world had ever known, and now it’s back to civilian life making silly noises once again.

[Vocoder making robotic noises]

Roman Mars:
99% invisible was produced this week by Delaney Hall with Katie Mingle, Sam Greenspan, Avery Trufelman, Kurt Kohlstedt, Sharif Youssef and me, Roman Mars. Special thanks this week to Dave Tompkins, author of ‘How to Wreck a Nice Beach, The Vocoder from World War II to Hip Hop’. His book inspired this episode and gets into the history of how the vocoder jumped from the military into popular music.