Voice Changer

How the Voice Changer works

Every preset here is a classic studio trick that has been around for decades, rebuilt with the audio processing that every modern browser already contains. None of them uses a neural network or a cloud service, so the result is predictable and repeatable: the same file and the same setting always give the same output.

Chipmunk and Deep

The original chipmunk sound was made in 1958 by recording a voice at half speed and playing the tape back at normal speed. That doubles the pitch, an octave up, but it also doubles the speed and halves the length. This page does it the modern way: the voice is first time-stretched with a waveform-similarity overlap-add algorithm so it lasts longer, then played back faster so it returns to its original length at a higher pitch. Chipmunk raises the pitch by 7 semitones, a musical fifth, which is high and cartoonish but still easy to understand; Deep lowers it by 5 semitones, a fourth, which gives a large, slow sounding voice without turning to mud.

Both presets shift the formants along with the pitch. Formants are the resonances of the throat and mouth that make a voice sound like a particular person, and shifting them is exactly what makes the chipmunk sound small and the deep voice sound huge. For a more natural shift with finer control, use the Pitch Shifter, which goes up or down by any number of semitones and cents.

Robot

The Daleks of Doctor Who (1963) got their voices from a ring modulator: the voice is multiplied by a steady sine wave, which replaces every frequency in the voice with a pair of new ones above and below it. The result has no natural harmonic relationship, so it sounds metallic and mechanical while keeping the rhythm and the consonants of speech. The Robot buzz slider sets the frequency of that sine wave. Around 30 Hz gives a deep, warbling Dalek; 50 to 80 Hz is the classic film robot; above 100 Hz the voice becomes harsh and bell-like.

Radio

AM radio, telephones and walkie-talkies pass only a narrow band of frequencies. The traditional telephone band runs from 300 to 3,400 Hz, and the Radio preset filters the voice to the same range with steep high-pass and low-pass filters, then adds gentle saturation to imitate a small overdriven loudspeaker. It is the sound used in films for phone calls, cockpit chatter and announcements, and in music for intros that open up into the full mix.

Echo

The Echo preset feeds the voice into a delay line whose output is fed back in at half level, so each repeat is 6 dB quieter than the last. The delay slider sets the gap between repeats: 60 to 120 ms is a short slapback like a 1950s rock and roll vocal; 250 ms is a clear canyon echo; 500 ms and above leaves room for whole words between the repeats. The output is lengthened by eight repeats so the tail is not cut off. For control over mix, feedback and stereo ping-pong, use the Echo and Delay page.

Getting a good result from the voice changer

• Start with a clean, dry recording. Background noise is changed along with the voice, so a hum or a fan becomes a robot hum or a chipmunk fan. Record close to the microphone in a quiet room, or clean the file first.
• Keep levels below clipping. Ring modulation and saturation can add peaks. If the export sounds crunchy, lower the level of the source with the Audio Volume Changer and try again.
• Speak more slowly for Chipmunk. Raising the pitch makes speech feel faster even when the timing is unchanged. A deliberate delivery stays intelligible.
• Stack effects with export and re-import. Export the Deep voice as WAV, drop it back in and apply Echo for a booming announcer. WAV keeps quality across passes.

You can record straight in your browser with the Voice Recorder and drop the file here.

Live use

This page processes recordings, not a live microphone. Real-time voice changing for calls and streams needs software that sits between the microphone and the app, because a web page cannot replace the microphone that other programs see.

Related tools