How to Change a Karaoke Track's Key

To change a karaoke track's key, drop your file into our free karaoke maker and move the "Key shift" slider until the song sits in your vocal range. If you already have a vocals-free backing track, the pitch changer transposes it directly. Both run entirely in your browser, both keep the original tempo, and your audio never leaves your device — no upload, no account. Finding the right key usually only takes a minute of nudging a slider.

The rest depends on what you're starting with: a full song (vocals plus music) that you need to turn into a key-shifted karaoke version, or a backing track you already have and just want to move up or down.

What you'll need

• An audio file — MP3, WAV, FLAC, or M4A.
• A rough idea of which way to go: songs that feel too high get shifted down, too low get shifted up. You don't need to know the exact number in advance; you'll find it by ear.
• A modern browser. The key-shift feature relies on AudioWorklet support, which every current version of Chrome, Edge, Firefox, and Safari has.

No software to install, no sign-up. Processing happens on your own machine.

Change the key of a full song (Karaoke Maker)

Use this path when your file still has the original singer on it and you want a karaoke version in a new key.

1. Open the karaoke maker at /karaoke-maker and drop your audio file onto the upload area. If you're building the track from scratch, our guide on how to make a karaoke track covers the full workflow first.
2. Let the AI separate the stems. The tool uses an AI model (HTDemucs) to split the vocals from the instrumental. The first time you use it, your browser downloads that model (tens of megabytes) from our CDN, then runs the separation on your own CPU. This takes real time — it is not instant — but it's a one-time download, so later sessions are quicker to start.
3. Find the mixer. Once separation finishes you get three sliders: Guide vocal (0–100%, starting at a quiet 25% so you can hear the melody without it drowning you out), Instrumental (0–100%), and Key shift.
4. Move the Key shift slider. It runs from -12 to +12 semitones, labeled -1 Octave to +1 Octave, and shows the current amount as "+2 st" or "-3 st". As the note on the tool says, "Tempo stays the same — only the key moves."
5. Sing along and adjust. Press play, sing the highest and lowest lines of the song, and nudge the slider until both sit comfortably. Use the back-to-start button to loop the tricky part.
6. Drop the guide vocal (optional). Once you've found your key, slide Guide vocal down toward 0% for a clean instrumental, or leave a little in as a pitch reference.
7. Export. Pick a format in the selector — WAV (default, lossless), MP3, or Opus — and click the "Export karaoke" button. The file saves straight to your device.

Change the key of a backing track you already have (Pitch Changer)

If your file is already an instrumental with no vocals, you don't need stem separation at all. The pitch changer transposes it directly — no AI model, no download for this step, just decode and shift.

1. Open the pitch changer and drop in your backing track. It decodes the file and shows a waveform.
2. Move the Pitch slider (-12 to +12 semitones) to the amount you want.
3. Leave the Speed slider at 1.0x. This is the important part: Pitch changes the key, Speed changes the tempo. To transpose without speeding the song up or down, only touch Pitch. There's a separate Volume control if you need it.
4. Export in WAV, MP3, or Opus.

Because there's no separation step, this route is faster to start — it's the better choice whenever you already have a vocals-free track.

How many semitones do you need?

A semitone is one half-step; twelve semitones make a full octave. That's the whole scale you're working with:

• Up a whole step (one full tone) is +2 semitones; down a half step is -1.
• A singer covering a song written for the opposite voice type often moves it a fair way — men taking on a song in a female key commonly go down 2 to 5 semitones, and the reverse going up.

The honest guidance: use the smallest shift that fits your range. Small moves — roughly ±1 to ±3 semitones — sound clean. Push toward ±6 or a full octave and you start to hear artifacts, a hollow or "chipmunk" quality, because the formants (the resonances that make a voice or instrument sound natural) shift along with the pitch. There's more on that trade-off in our guide to pitch shifting without ruining the sound, and on why key and tempo used to be locked together in how pitch, speed, and duration relate. If a song only sounds right two octaves away from where you can sing it, that song may just not suit your range, and a smaller shift plus a bit of octave-dropping on the highest lines will sound better than forcing a huge transposition.

Don't overthink the number. Start at the middle, move the slider a step at a time, and stop where the top and bottom notes both feel easy.

Common problems

The track sounds warbly or metallic after a big shift. You've moved too far. Pull the shift back toward zero and find a smaller amount that still fits. Time-preserving pitch shifting is very good at small moves and gets rougher the further you push it.

The tempo changed too. In the pitch changer, that means the Speed slider isn't at 1.0x — reset it. The karaoke maker never changes tempo when you move Key shift, so this only happens on the pitch-changer path.

The separation is taking a while. That's the AI model working on your CPU, plus a first-time model download. It's normal, it's local (your audio isn't being sent anywhere), and it's a one-time download. If you only need to transpose an existing instrumental, use the pitch changer instead and skip separation entirely.

Faint vocals remain in the karaoke mix. AI separation is strong on typical pop and rock but weaker on dense, heavily-processed, or mono recordings, so a little of the original singer can survive. Turning the Guide vocal slider to 0% removes the guide track cleanly; residual bleed from the original is a limit of separation, not a setting you missed.

Related tasks

If you're building a karaoke version from scratch — separating the vocals out in the first place, then setting your key and levels — start with the full karaoke maker workflow. Once your track is in the right key, the next step is usually performance: our guide to singing over an instrumental covers monitoring and timing. And if you're matching this track to another for a set, beatmatching and key matching explains how keys fit together.

Frequently asked questions

Does changing the key change the tempo?

No. Both tools use time-preserving pitch shifting, so moving the key leaves the tempo exactly where it was. This is the modern approach — unlike speeding up or slowing down a record, which changes pitch and tempo together. In the pitch changer, just make sure the separate Speed slider stays at 1.0x, since that control is what adjusts tempo.

Can I change the key of a song that still has vocals on it?

Yes. Use the karaoke maker, which first separates the vocals from the instrumental with an AI model, then lets you shift the key and lower or mute the vocal. If your file is already an instrumental with no singing, skip separation and use the pitch changer, which transposes the track directly and is quicker to start.

How many semitones should I shift?

Use the smallest shift that puts the song's highest and lowest notes in comfortable reach. Small moves of about one to three semitones sound clean; larger shifts toward six or more can introduce a hollow or chipmunk-like artifact as the formants move. Sing along and nudge the slider a step at a time rather than guessing a number up front.

Is my audio uploaded to a server?

No. All processing runs in your browser on your own device, and your audio never leaves it — there's no upload and no account. The one thing that does download is the AI separation model in the karaoke maker, fetched once on first use so the tool can run separation locally. The pitch changer downloads nothing extra.

Why does the karaoke maker take a while but the pitch changer is quick?

The karaoke maker runs AI stem separation on your CPU and downloads its model the first time, so it isn't instant. The pitch changer only decodes your file and shifts the pitch — no separation, no model — so it starts almost immediately. If you already have a vocals-free backing track, the pitch changer is the faster route.