Short for vocal encoder. An electronic audio device which allows synthesizers to 'sing'.


A vocoder takes two input signals; a modulator and a carrier. The output consists of the waveform of the carrier, colored with the frequency spectrum of the modulator. Typically the carrier input will be a synthesizer, and the modulator input will be a microphone.

The modulator is split to an array of bandpass filters, each one examining a unique frequency band. The amplitude of the output of each frequency band is passed forward to the carrier module. The carrier signal is split into a second array of bandpass filters, of the same frequency bands as the first. The output is a mix of the outputs of the second array of bandpass filters, with the level controlled by the amplitude of the signal in the first array.

More simply stated, you can imagine a graphic equalizer on the carrier, and the height of each slider is being constantly updated according to a spectrum analyser showing the frequency spectrum of the modulator.

A vocoder cannot add frequencies which are not present in the carrier. It can only emphasise frequencies which are in the carrier and modulator, and subtract frequencies which are present in the carrier but missing from the modulator. Therefore for best results, the carrier signal should have a wide bandwidth, such as a sawtooth wave or white noise. The worst case would be a sine wave carrier, which would simply change in amplitude.

A vocoder with a large number of very narrow bands will pass the modulator signal more accurately, although the computing power (if digital) increases dramatically.


The first vocoder was designed in 1939 by Bell Labs. This was a 10-filter analyser, working in the 250 to 3000Hz range. Speech could be reconstructed with a noise or buzzer source.

1939 Dudley, H., "The Vocoder", Bell Labs. Rec., 17, 122-126 (1939). (I,K)

Famous applications

There are two main applications of a vocoder, they can make a 'singing synthesizer' effect, as in the following:

Another application is to electronically tune a singing voice. This can be more (S Club 7) or less (Cher) robotic. The model used on Believe was a Digitech Talker, and was chosen over the Korg vocoder for its clarity:

Not a vocoder

Peter Frampton did not use a vocoder, although the effect is similar. He used a TalkBox.

A simple vocoder effect may be acheived extremely cheaply by talking through a harmonica - the carrier is generated by the vibrating reeds in the instrument, and the tone coloring is performed by the oral cavity.

Log in or register to write something here or to contact authors.