Learn — By WATMM Desk —
Vocoders and talk boxes explained
A vocoder imposes the spectral shape of a voice on a synth or noise carrier electronically, band by band. A talk box pipes an instrument into the player's mouth and lets a microphone pick up the mouth-shaped result. Both make instruments talk; only one needs your mouth.
Key facts
- Homer Dudley at Bell Labs invented the vocoder; he received patents in 1937 and 1939.
- A vocoder splits the voice into frequency bands (typically 8 to 20), follows the level in each, and uses those levels to shape a carrier such as a sawtooth or noise.
- A talk box sends an instrument through a speaker and tube into the performer's mouth; the mouth shape filters it and a microphone picks it up.
A vocoder and a talk box both make an instrument sound like it is speaking. The difference is where the "talking" happens: inside a circuit or plug-in for the vocoder, inside your mouth for the talk box.
How a vocoder works
The vocoder was invented by Homer Dudley at Bell Labs, who received patents in 1937 and 1939. It has two halves:
- Analysis. The voice (the modulator) is split into frequency bands, typically 8 to 20. An envelope follower tracks how loud each band is over time. The pitch of the voice is thrown away.
- Synthesis. A carrier, usually a sawtooth wave or noise, runs through a matching bank of band-pass filters. Each band's level is set by the voice envelope, so the carrier takes on the shape of the speech.
Because "s" and "f" sounds sit outside the main speech bands, many vocoders add a separate sibilance or noise band so consonants stay legible. The same idea had a wartime job: Bell Labs built the encrypted SIGSALY voice system on Dudley's vocoder in 1943.
Today the vocoder is a stock synth feature: Korg's microKORG lists a vocoder and ships with a mini microphone for the modulator.
How a talk box works
A talk box is usually a floor unit with a speaker driver feeding a plastic tube. The tube is taped to a microphone and runs into the player's mouth. The instrument's sound fills the mouth, the player shapes vowels and consonants silently, and the mic picks up the filtered result. Peter Frampton made it famous on Frampton Comes Alive; Roger Troutman of Zapp played it from a keyboard.
Side by side
| Vocoder | Talk box | |
|---|---|---|
| Where shaping happens | Filter bank (hardware or software) | Your mouth |
| Inputs | Voice plus carrier | Instrument only |
| Pitch comes from | Carrier (notes you play) | Instrument |
| Typical home | Electronic music | Rock, funk |
What we think
Pick by how you want to perform, not by sound alone. A vocoder is the practical choice for most producers now: it runs as a plug-in, works on a recorded vocal after the fact, and the band count is a creative control (fewer bands sound more robotic, more bands more intelligible). A talk box can't be fixed in the mix, since the mouth is the filter, but that is its strength: articulation is physical, so it reads as more expressive and human on stage. If a plug-in "talk box" sounds a little flat, that is why. One practical tip for vocoders: feed them a bright, harmonically rich carrier; a pure sine has too little content in the upper bands for the voice to shape.