Vocoder

A vocoder is a specialized category of speech coding. The name is a portmanteau of the words "voice" and "encoder."

This technology analyzes and synthesizes human voice signals. It serves several important functions in modern communication and art. It can be used for audio data compression to save space. It is also used for multiplexing and voice encryption. Encryption allows voice transmissions to be secured against interception. This is particularly useful for secure radio communication.

To understand the mechanism, we must look at how the human voice works. The voice consists of sounds from the glottis opening and closing. This creates an acoustic waveform with many harmonics. The vocal tract, which includes the nose, mouth, and throat, then filters this sound. This filtering creates fluctuations in harmonic content called formants. The vocoder analyzes speech by measuring how spectral energy distribution fluctuates over time. It divides the voice signal into many frequency bands. An envelope follower measures the signal level in each band simultaneously. This set of amplitude signals is called the modulator.

Recreating the speech requires reversing this analysis process. The vocoder uses an initial broadband noise called a carrier. This carrier can also be a sawtooth waveform. The carrier passes through a set of band-pass filters. The modulator controls the individual envelope amplitude levels of these filters in real time. This process maps the frequency components of the modulator onto the carrier.

In analog vocoders, there are usually between 8 and 20 bands. Many systems also include an unvoiced band for sibilance. This handles sounds like "s," "f," or "ch" that fall outside typical speech bands. This extra channel helps make the resulting speech more recognizable.

There are different ways to implement this technology. Since the late 1970s, most non-musical vocoders use linear prediction. In linear predictive coding (LPC), an all-pole IIR filter estimates the spectral envelope. This filter replaces the traditional band-pass filter bank. It can be more precise because the filter peaks are determined by the target signal. Another method is waveform-interpolative (WI) vocoding. This was developed at AT&T Bell Laboratories around 1995. Digital systems also use adaptive differential pulse-code modulation (ADPCM). This technique was developed at Bell Labs in 1973 by P. Cummiskey, Nikil S. Jayant, and James L. Flanagan.

The history of the vocoder began with Homer Dudley at Bell Labs. He started his development work in 1928. Dudley received patents for his inventions in 1939. He introduced the "voder," or voice operating demonstrator, at the 1939–1940 New York World's Fair. The voder was a complex machine used for speech synthesis. It used an electronic oscillator and a noise generator. An operator used pressure-sensitive keys and a foot pedal to control it.

This allowed a skilled person to produce recognizable speech.

Vocoders have played significant roles in history and security. In 1943, Bell Labs engineers built the SIGSALY system. This system provided encrypted voice communications during World War II. Later models focused on reducing size and weight. The KY-9 THESEUS voice coder arrived in 1953. It used solid-state logic to reduce weight from the heavy SIGSALY system. By 1961, the HY-2 voice coder was released as a 16-channel system. Modern systems are much more efficient. A good vocoder can simulate a voice with as little as 2.4 kbit/s of data. This is impressive compared to standard recording rates.

In the world of music, the vocoder is a popular electronic instrument.

Musicians often use a synthesizer as the carrier instead of a voice. This creates a unique, mechanical sound. This process can feel "dehumanizing" because it discards the fundamental frequency. It only sends the parameters of the vocal model rather than the full waveform. This effect became very popular in the 1970s. Bands like Kraftwerk used custom-built vocoders for their electronic music. Today, the technology is still used for artistic effects and in cochlear implants to simulate hearing.