In an exclusive extract from his new book The Radio Phonics Laboratory,
  Justin Patrick Moore takes us through the secret history of the links
  between wartime cryptography and electro-funk

  By the 1940s, radio and telephone had revolutionized communications in
  the industrialized world. Its strategic use in warfare, as a way to
  transmit information and intelligence, was not lost on military
  personnel. Yet these signals needed to be protected. Especially with
  radio, it was easy to become an undetected eavesdropper wherever
  signals could be received. After World War I, the first devices to
  attempt secure transmission of voice were developed. These were
  substitution devices that inverted frequencies in a way similar to how
  alphabetic substitution codes swapped one letter for another. High
  frequencies were substituted for low frequencies and vice versa. It was
  easy to make a device that could do this, but the system was also easy
  to break.

  In 1931, Bell Telephone Laboratories developed the A-3 scrambler that
  was used by Roosevelt and Churchill when they talked on the phone.
  Although the United States was still not at war in 1941, they were
  aiding the Allies with materiel, and secure calls had to be made. The
  A-3 worked by dividing the voice frequency into five subbands. Each of
  these was inverted. On top of the measure of substitution, it also used
  transposition, with the voice shifted from one subband to another every
  twenty seconds.

  The security of the A-3 was eventually compromised by Germans based at
  a radio post in South Holland who had been intercepting the British
  prime minister’s telephone calls in 1941. Aware that their system was
  insecure, the situation was becoming intolerable for the Allies.
  Tensions had been mounting between the US and Japan, and a “warning of
  war” had come in late November, but no one knew when or where. On
  December 7, 1941, American cryptanalysts were hard at work breaking a
  long ciphertext intercepted from Tokyo to the Japanese embassy in
  Washington D.C. The team of codebreaker William Friedman, of the
  Signals Intelligence Service, had broken Japanese diplomatic ciphers
  before, but this naval cipher, scrambled in fourteen parts, was
  impenetrable.

  One part they did tease out: the Japanese embassy had been instructed
  to destroy their cryptography equipment and meet with the US Secretary
  of War, suggesting that armed conflict seemed imminent. Admiral Stark,
  Chief of Naval Operations, needed to send Admiral Kimmel at Pearl
  Harbor another message to reinforce the warning of war. Stark knew the
  A-3 scrambler was no longer secure, so he chose to send his message by
  radio. However, atmospheric propagation conditions over the Pacific
  stalled his message in its tracks. The next option was to send a
  message by undersea cable via Western Union. Unfortunately, by the time
  it arrived the battleships of the Pacific Fleet had already been bombed
  and the Japanese planes were flying back to their bases.

  The need for secure transmissions became even more paramount as the
  United States entered the war. In 1942, the Army contracted Bell Labs
  to assist with the communication problem and create “indestructible
  speech” that could withstand attempts at code breaking. From this
  effort, the revolutionary twelve-channel SIGSALY system was born.
  SIGSALY was not an acronym, just a codename for the project. To create
  SIGSALY, workers sifted through over eighty patents in the general area
  of voice security, settling on Homer Dudley’s vocoder to form the basis
  of the system. For SIGSALY, a twelve-channel vocoder was used. Ten of
  the channels measured the power of the voice signal in the audio
  portion of the frequency spectrum where most talking occurs, while two
  channels were devoted to “pitch” information and whether or not
  unvoiced (hiss) energy was present. The vocoder enciphered the speech
  as it went out over phone or radio. In order to be deciphered at each
  end of the conversation, an audio crypto-key was needed. This came in
  the form of vinyl records.

  From the standpoint of music history, it is interesting to note, as
  Dave Tompkins did in his book How to Wreck a Nice Beach: The Vocoder
  from WWII to Hip-Hop, that the SIGSALY system employed two-turntables
  and a microphone. The classified name for this vinyl part of the
  operation was SIGGRUV, while the vinyl records were produced by the
  Muzak Corporation, a company famous for the creation of elevator music.
  The sounds on these records weren’t aimed at soothing weekend shoppers
  or people sitting in waiting rooms, but rather contained random white
  noise, like channel 3 on an old television set. The noise was created
  by the output of very large mercury-rectifier tubes that were four
  inches in diameter and over a foot high. These generated wide band
  thermal noise that was sampled every twenty milliseconds. The samples
  were then quantized into six levels of equal probability. The level
  information was converted into channels of a frequency-shift keyed
  audio tone signal recorded onto a vinyl master. From the master, only
  three copies of a key segment were ever made. If any SIGGRUV vinyl
  still exists, and for security reasons they shouldn’t, those grooves
  are critically rare.

  It had to be insured that no pattern could be detected, so the records
  had to be random noise. Even if the equipment had somehow been
  duplicated by the Axis powers, the communications would still be
  uncompromised, as they still required the matching vinyl crypto key at
  each terminal. This made the transportation of these records, via
  armored truck, the most secure since Edison invented the phonograph.
  Just as the masters were destroyed after making three keys, each vinyl
  key was only ever to be played once, as operators were instructed to
  burn them after playing. The official instruction read, “The used
  project record should be cut-up and placed in an oven and reduced to a
  plastic biscuit of ‘Vinylite.’” As another precaution against the
  grooves falling into enemy hands, the turntables themselves had a
  self-destruct mechanism built into them that could be activated in case
  one of the terminals was compromised. Thinking of all this adds a
  fascinating new dimension to the idea of a DJ battle.

  Synchronizing turntables at two different terminals across the globe
  was another technical hurdle Bell Labs overcame, which was no small
  feat given that if a needle jumped or the system went out of sync, only
  garbled speech was heard. At the agreed upon time, say 1200 GMT,
  operators listened for the click of the phonograph being cued to the
  first groove. The turntables were started by releasing a clutch for the
  synchronous motor that kept the turntable running at a precise speed.
  Fine adjustments were made using fifty hertz phase shifters (Helmholtz
  coils) to account for delays in transmission time. The operators would
  listen for full quieting of the audio as synchronization was
  established. Oscilloscopes and shortwave receivers were also used to
  keep systems locked to international time.

  A complete SIGSALY system contained about forty racks of heavy
  equipment comprising vacuum tubes, relays, synchronous motors,
  turntables, and custom made electromechanical equipment. In the
  pre-transistor era, all of this gear required a heavy load of power, so
  cooling systems were also required to keep it all from getting fried
  from the heat. The average weight of a set up was about fifty-five
  tons.

  The system passed the inspection of Alan Turing, “the father of modern
  computer science,” if not his test. He had been briefly involved with
  the project on the British side, just as Claude Shannon had been in
  America. On July 15, 1943, the inaugural connection was established
  between the Pentagon and a room in the basement below Selfridges
  Department Store in London. Eventually a total of twelve SIGSALY
  encipherment terminals were established, including locations in Paris,
  Algiers, Manila, Guam, Australia, and one on a barge that ended up in
  the Tokyo Bay. In the year 1945 alone, the system trafficked millions
  of words between the Allies.

  To keep all of this operational, a special division of the Army Signal
  Corp was set up: the 805th Signal Service Company. Training commenced
  in a school set up by Bell Labs and soldiers were sent to various
  locations, requiring security clearances and a firm grasp on the
  cutting edge technology they were tasked to operate and maintain. For
  every eight hours of operation, the SIGSALY systems required sixteen
  hours of maintenance and calibration.

  In putting the system together, eight remarkable engineering “firsts”
  were achieved. A review conducted by the Institute of Electronic and
  Electrical Engineers in 1983 lists them as follows:
   1. The first realization of enciphered telephony
   2. The first quantized speech transmission
   3. The first transmission of speech by Pulse Code Modulation (PCM)
   4. The first use of companded PCM
   5. The first examples of multilevel Frequency Shift Keying (FSK)
   6. The first useful realization of speech bandwidth compression
   7. The first use of FSK – FDM (Frequency Shift Keying-Frequency
      Division Multiplex) as a viable transmission method over a fading
      medium
   8. The first use of a multilevel “eye pattern” to adjust the sampling
      intervals (a new, and important, instrumentation technique)

  SIGSALY left the world with a rich inheritance that spans developments
  in cryptology and digital communications, a legacy that ticks away in
  the background of history every time someone hits the decks with two
  turntables and a microphone.

  The Radio Phonics Laboratory by Justin Patrick Moore is published by
  [1]Velocity Press

References

  1. https://velocitypress.uk/product/radio-phonics-laboratory-book/