Pheasant CouchHomebrew Vocal Tuner
Sing it, pick the key, and the notes land where you meant them. Gentle enough that nobody knows, or hard enough that everybody does. It runs in your browser, and nothing you record leaves your device.
Ready
This browser can’t run the tuner (it needs Web Audio). Try a current Chrome, Edge, Firefox or Safari.
Takes
Or drop WAV, MP3, M4A or OGG files anywhere on the page. One voice at a time, as dry as you can get it.
No takes yet. Record one, or drop a file in.
Which notes are allowed
The tuner only ever pulls toward a note in this list. Pick the song’s key, let it guess from the take, or tap the notes yourself.
Now
How hard to pull
Strength is how far toward the note; retune speed is how fast it gets there. Instant and all the way is the famous robot. Slow and most of the way is invisible.
Where the notes went
No take yet
sung tuned harmonythe bright lines are the notes in the key
Space bar plays and stops.
Take it with you
Each take comes down as a 16-bit mono WAV, exactly as long as what you gave it, at the level you sang it, so it drops straight back onto your track in any editor. Save every take at once and they arrive in one zip.
- WAV, 16-bit, mono, the same length and level as the take
- Named after the take and the key, like
verse_G-major.wav - Tagged inside the file as a processed voice made here, with the date and settings
- Yours to use in anything, paid or not; no credit needed
Was this made here?
Every WAV saved from this page carries a tag. Pick a file and this reads it back. Nothing is uploaded.
Getting a good result
One voice, dry, close
The tuner follows one pitch at a time. It can’t pick a voice out of a band, a chord out of a guitar, or a singer out of a bathroom’s worth of reverb. Record close to the microphone with nothing else playing out loud (headphones for the backing track), and skip the reverb until after.
It keeps your timing
The saved file is sample-for-sample the same length as the take and starts at the same instant, so it lines up on your track without nudging. If you recorded to a click or a backing track, drop the result in at the same spot.
Natural or obvious
With Strength at 70% and Retune around 120 ms, the correction follows your voice and most listeners won’t hear it working. All the way and instant turns every slide into a step: that is the effect everyone knows, and it’s a choice, not a mistake.
The right key matters most
If the key is wrong, the tuner pulls notes to the wrong places and it sounds worse than before. Guess from this take is usually right for a phrase with a few different notes in it. When in doubt, tap only the notes the melody actually uses.
Harmonies that fit the song
The harmony voice follows the key, not a fixed interval: a third above in C major is E over C but F over D, the way a real backing singer would do it. On the chromatic setting it falls back to a fixed interval. Save the harmony on its own and you can pan it, lower it, or stack several.
How it works
The pitch is measured every six milliseconds (the YIN method), each slice gets a target from the key, and the voice is rebuilt one pitch period at a time at the new pitch, which keeps your tone and formants in place instead of chipmunking them. It was built after the voice modulator asked for a sibling for the musical friends.
Your voice, or with permission
Use it on your own singing, or on someone’s who said yes. Don’t use it to pass off a tuned voice as someone else’s performance, and don’t use it on anyone’s voice to deceive. You’re responsible for what you make with it.
Private by design
Recording, analysis and tuning happen on your device. No audio is sent anywhere. This page keeps nothing at all.