Why this is harder than it sounds
Playing a recording faster raises its pitch, and that is one line of code. Raising the pitch while the recording stays exactly as long is a different problem entirely: the waveform has to be rebuilt so that its frequencies move and its duration does not.
The way through is to do it twice. Stretch the audio in time — longer, same pitch — and then play the stretched version faster, which raises the pitch and brings the length back to where it started. Two operations that each change one thing, arranged so that one change cancels.
What it is good for
Singers who need a song a tone lower. Guitarists whose instrument is tuned down a semitone and who would rather move the recording than retune. Anyone learning a part that sits just outside their range. Transposing a backing track without making it faster or slower is the whole use case, and it is one a simple speed change cannot serve.
Intervals are named as well as numbered, because "+7 semitones" means nothing to most people and "a fifth up" means something to anyone who plays.
How far you can push it
Two or three semitones either way is usually indistinguishable from a recording made in that key. Past about five the artefacts start to show — a slight warble on sustained notes, a smearing of the transients — because the rebuild has to invent more of what it cannot know. A full octave is possible and audibly processed.
Voices show it sooner than instruments, since we know exactly what a voice should sound like. If you are shifting a long way, listen for the consonants first; they are what goes first.
Frequently asked questions
Will the song get longer or shorter?
Neither. That is the point of this tool as opposed to a speed change — the duration is identical to the original, whichever way you shift.
How many semitones can I move it?
Up to an octave either way. Two or three sounds natural; past five the artefacts become audible, especially on voices, because more of the waveform has to be reconstructed.
What is a semitone?
One fret on a guitar, one key on a piano counting the black ones. Twelve make an octave. The interval name is shown alongside the number, so +7 also reads as a fifth.
Why does it sound a bit watery?
That is the reconstruction. Moving pitch without moving time means rebuilding the waveform, and sustained notes are where the seams show. Smaller shifts show it less.
Can I change the speed instead?
Yes — the audio speed changer does that, and it can hold the pitch while it does. Between them they cover both axes independently.
Something wrong with this tool, or an idea for it? Tell us