Pitch shifting in AI voice changers, explained

By the voicechange.io team · November 6, 2026

The pitch setting is the one control that makes the biggest difference to how natural an AI voice sounds. This guide explains what it does, how voicechange.io picks a value automatically, and when you should override it.

Pitch and timbre are two different things

When you hear a voice, you are hearing two things at once. Pitch is how high or low the voice is: the note it sits on. Timbre is the colour of the voice: the raspiness, nasality or warmth that makes Morgan Freeman sound like Morgan Freeman even when he speaks a little higher or lower than usual.

An RVC voice model only changes timbre. It keeps the pitch of your recording exactly as it is, note for note. So if a man with a speaking voice around 110 Hz converts to a cartoon character whose voice sits around 280 Hz, the model is asked to produce that character's timbre at a pitch the character never speaks at. The result sounds wrong: muffled, growly or "dubbed". Pitch shifting fixes this by moving your recording up or down before the model sees it.

Semitones and octaves

The pitch slider moves in semitones, the smallest step on a piano keyboard. Twelve semitones make one octave, which doubles (or halves) the frequency. A few reference points:

Shift	What it does
+1 to +3	Slightly brighter or younger; good for fine-tuning
+4 to +7	Lower male voice to higher male or androgynous voice
+12	One octave up: typical male voice to typical female or child voice
−12	One octave down: typical female voice to typical male voice

As a rough guide, adult male speaking voices average about 85 to 155 Hz and adult female voices about 165 to 255 Hz. Many cartoon and anime characters sit well above that.

How auto pitch works on voicechange.io

Every voice in the catalog has a measured centre pitch, shown on its voice page. When auto pitch is on, the site does the following for each job:

Measures the median pitch of the voiced parts of your recording (silence and unvoiced sounds like "s" are ignored).
Works out how many semitones separate your median pitch from the voice's centre pitch.
If the gap is less than 4 semitones, it leaves your pitch alone, since you already fit the voice and small shifts tend to make things sound less natural.
Otherwise it shifts your recording by that amount, up to one octave in either direction.
The slider value is then added on top, so you can fine-tune the automatic choice.

For example, if your voice sits around 120 Hz and you pick Elmo (whose model is centred around 320 Hz), auto pitch shifts you up by the full octave. If you pick Darth Vader (around 95 Hz), the gap is about 4 semitones, so you get shifted down by 4.

Song covers are different: octaves only

In Song cover mode, the converted vocals are mixed back over the original instrumental. If the vocals moved by, say, 5 semitones, they would be in a different key from the music and the cover would sound badly out of tune. So for covers, auto pitch only ever moves in whole octaves (−12, 0 or +12). A male singer covered by a female voice is moved up an octave; a song that already suits the voice is left alone.

If you adjust the slider yourself on a cover, stick to 0, +12 or −12 for the same reason.

When to override it

The result is close but sounds slightly strained: move one or two steps toward your natural voice (down if you were shifted up).
The voice cracks or flips between two registers: your recording may cover a wide range, like an excited shout. Try a shift 12 steps away from where you are.
You want a caricature: pushing a voice higher or lower than its natural range is a legitimate creative choice. Just expect artifacts past about 5 semitones beyond its range.
Instrument voices such as drums don't have a meaningful speaking pitch, so auto pitch does nothing for them. Use the slider to make the instrument sound higher or lower.

Text to speech

For text to speech you don't record anything, so pitch is handled for you: the site starts from a male or female base voice, whichever is closer to the target voice, and then applies auto pitch. The slider still works if you want the result higher or lower.

Try it now: the voice changer is free and needs no account. Pick from all voices, or read the FAQ.