Secure voice (alternatively secure speech or ciphony) is a term in cryptography for the encryption of voice communication over a range of communication types such as radio, telephone or IP. The implementation of voice encryption dates back to World War II when secure communication was paramount to the US armed forces. During that time, noise was simply added to a voice signal to prevent enemies from listening to the conversations. Noise was added by playing a record of noise in sync with the voice signal and when the voice signal reached the receiver, the noise signal was subtracted out, leaving the original voice signal. In order to subtract out the noise, the receiver need to have exactly the same noise signal and the noise records were only made in pairs; one for the transmitter and one for the receiver. Having only two copies of records made it impossible for the wrong receiver to decrypt the signal. To implement the system, the army contracted Bell Laboratories and they developed a system called SIGSALY. With SIGSALY, ten channels were used to sample the voice frequency spectrum from 250 Hz to 3 kHz and two channels were allocated to sample voice pitch and background hiss. In the time of SIGSALY, the transistor had not been developed and the digital sampling was done by circuits using the model 2051 Thyratron vacuum tube. Each SIGSALY terminal used 40 racks of equipment weighing 55 tons and filled a large room. This equipment included radio transmitters and receivers and large phonograph turntables. The voice was keyed to two vinyl phonograph records that contained a Frequency Shift Keying (FSK) audio tone. The records were played on large precise turntables in sync with the voice transmission. From the introduction of voice encryption to today, encryption techniques have evolved drastically. Digital technology has effectively replaced old analog methods of voice encryption and by using complex algorithms, voice encryption has become much more secure and efficient. One relatively modern voice encryption method is Sub-band coding.
Jean-Marc Vesin, Olaf Schleusing