Make your narration stop sounding like you. Pitch and vocal-tract shape are adjusted independently, so the result sounds like a different person — not a chipmunk.
Two things carry your identity, and they are not equally important. Pitch is the note your vocal folds vibrate at — it is the obvious one, and it is the one every cheap voice changer moves. Vocal tract length is the shape of the tube above them, and it is the stronger cue: it barely changes while you speak, whereas pitch wanders up and down within a single sentence.
That is why shifting pitch alone sounds like helium rather than like another person. A real speaker with a higher voice usually has a shorter vocal tract as well, so the resonances move with the pitch. Move one without the other and the ear immediately hears a processed version of you rather than someone new.
This tool moves both together, and gives you a separate slider so you can push the vocal tract further than the pitch went. If you only remember one thing: the vocal tract slider is the one that hides you.
A tool for staying anonymous that sends your voice to someone else's server is a contradiction. The entire engine here — the pitch shifter, the noise reduction, the export — runs inside your browser using WebAssembly. Your recording never leaves your device, there is no account, and there is nothing for us to store or leak.
The practical consequence is that the first run loads a few hundred kilobytes of code, and processing takes roughly as long as your machine needs rather than as long as a queue takes. Long files work fine; they just take proportionally longer.
Start with a preset and use Play original and Play changed back to back. The presets are spread out deliberately, so pick whichever sounds least like you rather than whichever sounds best.
Then open Fine-tune. Nudging the vocal tract slider by one or two semitones from the preset is usually enough to make a voice that no preset produces, which matters if you publish regularly — a distinctive preset used every week becomes its own fingerprint. The speed slider is worth a small nudge too, because speech rhythm is a recognisable habit in its own right.
Keep the total shift moderate. Past roughly five semitones in either direction the vocoder has less harmonic detail to work with and the result starts sounding synthetic rather than merely different.
Yes — completely free, no sign-up, no limits and no watermark on the output.
No. Everything runs locally in your browser. Your recording never leaves your device.
For a casual listener, yes — provided you move the vocal tract slider and not just pitch. This is disguise for narration, not protection against forensic voice analysis.
Because a real person's resonances move with their body size, not with the note they are hitting. Shifting pitch alone leaves the resonances behind, which is exactly the chipmunk artefact. This tool moves both together by default.
Not yet. This tool works on a recording you load or capture here and hands back a WAV — it is built for narration and voiceover rather than live calls.
MP3, WAV, M4A, OGG, WebM and the audio track of MP4 files. You can also record straight from your microphone. The download is always a WAV.
Only if the recording is actually noisy. The denoiser is good at room tone and traffic, but it can soften an s or an f, so leave it off when your audio is already clean.
