Why does the tool not ask for a key?

Because it corrects to the nearest semitone, which is valid in any key. Choosing a scale would mean committing to one, and a note that belongs to a key other than the one chosen would be pulled to the wrong degree — a risk that chromatic correction removes entirely.

What are voiced frames?

The short analysis windows in which the material has a measurable pitch, as opposed to breaths, consonants and noise, which do not. The count reported is how many of those voiced frames the correction actually moved, which is a clearer measure of the change than an impression of whether it sounded different.

Why does a high amount sound robotic?

Because at a hundred percent every frame lands exactly on its target semitone, so natural pitch movement disappears. A singer's vibrato and expressive slides are part of the performance, and removing them entirely makes the result sound like a machine rather than a person.

Will it affect breaths and consonants?

No, unvoiced frames pass through untouched. That is what keeps the result natural, since processing noise-like material through a pitch shifter gives it a glassy, unstable quality that is often more noticeable than the pitch error being corrected.

How much correction is too much?

The guidance in the panel is a good one: keep it between forty and seventy percent for a natural result. If a note is so far out that it needs more, correcting the performance again from the source, or moving on, is usually better than processing the note harder.

Pitch Correction

You Might Also Need

Auto-Tune Effect

Classic pitch correction effect

Harmonizer

Generate vocal harmonies

Audio Pitch Shifter

Shift pitch without changing speed

Pitch Shift Calculator

Calculate semitone shifts

Pitch Correction pulls sung or played notes toward the nearest semitone without removing their character. It is the gentle end of the same engine whose hard setting is the obvious auto-tune effect.

What is Pitch Correction?

Drop a file, set the correction amount, and process; the report states how many voiced frames were touched out of how many were analysed.

The controls and the readout:

A correction amount from nothing to everything, defaulting to sixty percent, where nothing is a fully transparent pass.
A retune strength from nothing to everything, defaulting to sixty percent.
A report giving the correction amount and the count of touched frames against the total voiced frames, which is the honest measure of how much the file changed.
Correction to the nearest semitone, with no scale required, which is the mode that suits cleanup because it cannot pull a note to a wrong degree.

The engine is a whole-file phase vocoder with a smoothed per-frame rate map: the audio is analysed in short frames, each frame's pitch is measured, a rate is computed to move it toward its target, and the rates are smoothed across frames so the correction arrives gradually rather than in steps. Two details are what make the result usable as cleanup rather than as an effect. The first is that unvoiced frames pass through untouched, so breaths, consonants and noise are not processed at all and cannot acquire the glassy quality that pitch shifting gives them. The second is that the amount is a proportion of the distance to the target rather than a switch. At sixty percent a note that is thirty cents sharp moves eighteen cents of the way, which removes the audible error while leaving the rest of the pitch intact, including a singer's vibrato. That is the important difference between this tool and its sibling: at a hundred percent every frame lands exactly on the nearest semitone and any natural pitch movement is flattened, which is heard as a mechanical quality, whereas at forty to seventy percent the file still sounds like a performance and simply sounds better tuned. Chromatic targeting is what makes the tool safe on material in any key, since it aims at the nearest semitone rather than at a scale that would have to be known in advance.

What it does not do:

No scale or key targeting, so it corrects to the nearest semitone rather than to a key.
No note-level editing, so a particular phrase cannot be corrected harder than the rest.
Unvoiced frames are passed through, which is deliberate.
No formant control, so very large corrections can still sound unnatural.
Polyphonic material is not analysed as separate voices.
Output is sixteen bit WAV.

Use it for the job it is named after: tightening a good take that drifts slightly, preparing a vocal that was recorded in one pass, or correcting an instrument that cannot be retuned. When the hard, obvious effect is what is wanted, the auto-tune effect snaps every frame to a scale with an adjustable key, and when the pitch should move by a fixed musical interval rather than being corrected, the harmonizer shifts a copy of the signal.

How to use Pitch Correction

Drop an audio file onto the panel and let it decode.
Leave the correction amount between forty and seventy percent for a natural result.
Process the file and read the touched-frame count to see how much was affected.
Compare against the original and lower the amount if the performance has lost its movement.

When to use Pitch Correction vs related tools

Pitch Correction is the tool for making a performance better tuned without making it sound processed. Auto-Tune Effect is its hard sibling, with key and scale targeting and the obvious snap. Harmonizer adds a shifted copy rather than correcting the original, and Pitch Shift Calculator works out the arithmetic of a shift before applying it.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.