The voice becomes a control signal
A vocoder splits incoming speech into several frequency ranges, from low to high. It tracks how strongly each range is present as the speaker talks. These changing levels preserve much of speech’s rhythm and articulation: the swell of a syllable, the pause between words, and the crisp bursts of consonants.
The original voice is not simply distorted. Instead, its changing pattern controls a second sound, called the carrier. That carrier might be a sustained synthesizer tone or another steady, harmonically rich sound.
Filters shape the synthesizer
The carrier passes through a bank of filters, each tuned to one of the frequency ranges measured in the voice. When the speaker’s voice grows stronger in a range, the matching filter lets more of the carrier through; when the voice fades, that band quiets. The synthesizer therefore takes on the voice’s changing contours without inheriting the vocal cords’ original tone.
That separation explains the characteristic effect. The listener can still recognize speech because its timing and many of its sound patterns remain, but the familiar human timbre has been replaced by the carrier. A smooth synth produces a clean, electronic voice; a rougher carrier can make the result harsher or more textured.
Why words remain recognizable
Speech depends on quick shifts in frequency emphasis, not just on a person’s underlying voice quality. The vocoder preserves enough of those shifts for vowels and consonants to stay intelligible, even as it changes what seems to be speaking. In effect, the voice supplies the instructions, while the synthesizer supplies the sound. That is why a vocoder can make a machine seem to pronounce a sentence rather than merely play a note.
