Making Music
  Computer Music Software and Hardware

Quick Guide To Vocoders

The vocoder is one of the most enduring audio processing effects of all time. We explain how to make it sing. And talk and dance and mutter and play the drums...

Although the human voice is the most versatile and expressive of instruments, in true creative style, musicians still like to change and modify it. A vocal that sounds in some way different to other vocals will stand out and attract attention to a recording.

One of the first commercial recordings to feature voice processing was Sparky's Magic Piano, released by Capitol in 1947. This landmark classic children's story is still available today. It used a device called a Sonovox which was mechanical in nature and worked by pressing two small disks to the performer's throat. Although it was not a vocoder and it worked in an entirely different way, the results were similar and many people assume the result was achieved with a vocoder.

Hooray for Homer

The vocoder was developed back in the 1930s by Homer Dudley who worked for Bell Labs. Its original purpose was to improve speech transmission over telephone lines, essentially by reducing content to allow more data to be sent over the line's limited bandwidth. It was, in fact, also used during World War II to scramble telephone conversations between Franklin D. Roosevelt and Winston Churchill.

Vocoders have been used in all forms of music. They were particularly prevalent in the pop and disco culture of the 70s and they keep dropping in and out of fashion like flared trousers. Memorable songs featuring a vocoder include Kraftwerk's We Are The Robots, Laurie Anderson's O Superman and ELO's Mr Blue Sky. The more recent release of Cher's Believe seems to have shepherded in a new era of interest in vocoders and rare now is the 'vocal artiste' whose voice has not been processed through one of these devices.

Initially, vocoders were only available as hardware devices. Some models from the early 1980s such as the Korg VC-10, and Roland's VP-330 and the rackmount SCV-350 were much in demand second-hand when the analogue retro scene arrived in the 1990s and some are still in use today. Modern hardware vocoders are still being made (such as the Doepfer vocoder synth modules, for example) but the recent trend has been towards software plug-ins which are often more versatile and, of course, much cheaper.

What's in a name?

Tech terms VCA Voltage Controlled Amplifier, typically a synthesiser module that adjusts the volume of a signal according to its input. In a vocoder the controlling signal is derived by analysing the frequencies in the modulator. Voiced/unvoiced The part of a vocoder that detects voiced (tonal) and unvoiced (noise) sections in a speech signal and makes the vocoder react accordingly. Harmonics Frequencies higher than the fundamental pitch of a sound that give it its tone. Sounds with many harmonics make the best carriers. Glide/Slew A control that smoothes the change between two discreet values. It's typically used in synthesisers to slide or glide between the pitches of two notes.

It's not essential that you know exactly how a vocoder does its stuff in order to use one but it won't do any harm either, and if you know what's going on under the hood you'll better be able to use it, perhaps in novel and unique ways.

The name 'vocoder' is derived either from VOice CODER, VOice enCODER or Voice Operated reCOrDER, depending on which source you read. A vocoder has two inputs called a carrier and a modulator and if you're familiar with carrier and modulator terminology from FM synthesis you'll be way ahead of us.

As the name suggests, the modulator modulates the carrier signal. In a vocoder, the modulator is usually a voice and the carrier is the signal that the modulation is imposed upon, usually a pad-type sound. This produces the typical 'talking synth' or 'robot voice' effect most commonly associated with vocoders.

Here's the techy explanation. A vocoder analyses the voice and splits it up into frequency bands, much like a spectral analyser. It might have six, eight, ten, twelve or more bands. Filters are also used to split the carrier signal into the same number of bands which are controlled by VCAs so essentially they control the volume of the frequencies in the carrier signal.

Let's say you say something at a low pitch. This will be analysed into the lower bands which, in turn, will activate the lower-band VCAs which will then pass the corresponding frequencies in the carrier signal. So, when you talk, the analytical filters output energy levels that correspond to the frequencies in your voice with the result that your voice's energy patterns are superimposed on the carrier.

<< Previous    [1]  2  3    Next >>