Change the key of any song, in the browser

How pitch shifting works

What is a semitone?

A semitone is the smallest step in Western music — the distance from one key on a piano to the very next one, black or white. Twelve semitones make an octave, the point where a note sounds like “the same note, higher”. Moving a song up by 12 semitones doubles every frequency in it; moving it down by 12 halves them. Most useful adjustments are far smaller than that: one or two semitones is usually all it takes to bring a song into a comfortable range.

Changing a song’s key for a singer’s range

Every voice has a range it sits in comfortably and a point above which it strains. When a song was recorded for a different voice, the fix is not to sing harder — it is to move the song. Find the highest phrase you struggle with, shift the whole track down one semitone, and try again. Repeat until that phrase is comfortable. Because pitch and speed are handled separately here, the song still runs at its original tempo, so you can practise or perform against the shifted version directly.

Why pitch and timbre are separate controls

Pitch is the note. Timbre — sometimes called formant, or the “body” of a sound — is what makes the same note sound like a large instrument or a small one, a deep voice or a light one. On a tape deck the two could never be separated: slowing the tape lowered both at once, which is why slowed-down voices sound like a different, larger person rather than the same person singing lower.

Handling them separately is what makes a modern pitch shifter useful. Move pitch alone and you get the same singer on a different note — the natural choice when you are changing key to suit a vocal range. Move both together and you get the classic deeper-or-higher voice effect. Move timbre alone and the note stays put while the character of the voice changes. If a shift sounds wrong, timbre is almost always the control that fixes it.

Does shifting damage quality?

Small shifts are close to transparent. The further you move from the original, the more the process has to invent, and beyond roughly five or six semitones most material begins to show it — usually as a slight smearing of drums and consonants. That is a property of pitch shifting itself, not of any one tool. Shift by as little as the job needs, and start from the highest-quality source file you have, since detail already lost to heavy MP3 compression cannot be recovered by shifting it.