Vocoder, Talkbox, Auto-Tune: Three Different Machines

They all make a voice sound electronic, and they work in completely different ways.

Three techniques get described interchangeably as robot voice, and confusing them makes it impossible to hear what any of them is doing.

You can listen to our full playlist which contains the artist’s music, and know more about the artist’s work by scrolling down the page.

PLAYLIST

---

Three Machines, Three Mechanisms

Three techniques get described interchangeably as robot voice, and conflating them makes it impossible to hear what any of them is actually doing.

A vocoder analyses one signal and imposes its spectral shape on another. A talkbox routes an instrument’s sound through a tube into the performer’s mouth and lets their mouth shape it acoustically. Pitch correction adjusts the pitch of a sung note toward a target.

They produce superficially similar impressions — a voice that sounds processed and electronic — through completely unrelated processes, and the differences are audible once you know what to listen for.

The Vocoder Analyses and Imposes

A vocoder takes two inputs: a modulator, almost always a voice, and a carrier, usually a synthesiser.

It splits the modulator into a bank of frequency bands and measures the energy in each one continuously. It then applies that same pattern of energy to the corresponding bands of the carrier.

The result is the synthesiser taking on the formant movement of the speech while keeping its own pitch and timbre. The pitch you hear is the synthesiser’s, not the singer’s, which is why a vocoded part can hold a chord while a voice speaks a single line.

Band count is what determines a vocoder’s intelligibility. Few bands give a coarse, robotic result where words are hard to follow; many bands track the speech closely enough to be understood clearly, at the cost of some of the effect’s character.

The Talkbox Is Acoustic

A talkbox involves no signal processing at all, which is the thing most often misunderstood about it.

An instrument’s amplified output is fed to a small driver connected to a plastic tube. The performer puts the end of the tube in their mouth and shapes the sound with their tongue, lips and jaw exactly as they would shape their own voice. A microphone in front of their mouth picks up the result.

So the filtering is done by a real mouth in real time. That gives it a fluidity and an idiosyncrasy no digital process reproduces, and it also means the performer cannot sing and use it simultaneously.

Pitch Correction Moves Notes

Pitch correction is a different category of tool, designed originally to make small transparent adjustments to a sung note that was slightly off.

Its use as an effect comes from setting the retune speed to its fastest setting. Instead of easing a note toward the target pitch, it snaps instantly, which eliminates the natural pitch movement between notes and produces the characteristic stepped quality.

Crucially, the voice remains the sound source throughout. Unlike a vocoder, the singer’s own timbre and pitch are present; they have simply been quantised to a scale with no transition between steps.

Talkbox playing is also physically unpleasant in a way rarely mentioned. The tube transmits vibration directly into the mouth and teeth, sustained use is uncomfortable, and the technique requires the performer to abandon singing entirely while using it.

How to Tell Them Apart

With the mechanisms clear, the identification is straightforward.

If the harmony is rich and polyphonic while the words are clear, and the vocal pitch does not track the melody, it is a vocoder. If it sounds like a specific person’s mouth with all its irregularity, and the part is monophonic and tied to an instrument line, it is a talkbox.

If the singer’s own voice is clearly audible with its natural character but the pitch jumps between notes with no glide, it is fast pitch correction. The presence or absence of the original vocal timbre is the most reliable single cue.

They Are Not Interchangeable

Understanding the distinction matters practically, because the three cannot substitute for one another.

A vocoder needs a carrier signal and therefore a decision about harmony, which makes it an arrangement tool as much as a vocal effect. A talkbox requires a performer with the physical technique and ties the part to whatever instrument feeds it.

Pitch correction is the only one of the three that leaves the vocal performance essentially intact, which is why it became ubiquitous while the other two remain specialist. It is also why it is the one most often used without anyone noticing.

Formant correction is the parameter that separates a natural-sounding pitch shift from an obviously processed one. Moving pitch without moving formants keeps the voice recognisable; moving both produces the chipmunk artefact that early shifters could not avoid.

---

About The Author