How to change your voice
- Press Record your voice and allow the microphone, or choose an audio file. Recordings and files can be up to 10 minutes long.
- Tap an effect. The tool applies it and plays the result straight away, so you can try one after another and compare with the original.
- Open Fine-tune to set the exact pitch in semitones or mix your own sound from robot, wobble, distortion, echo, reverb and 8-bit crunch.
- Choose MP3 or WAV and press Download. The file is named after your clip and the effect.
The effects
- Chipmunk, High voice, Deep voice: pitch up 9 or 5 semitones, or down 5, at the same speed.
- Monster and Ghost: a much lower voice with distortion or a long, wobbly reverb.
- Robot and Alien: ring modulation, which multiplies your voice by a steady tone. A low 50 Hz tone gives the metallic robot sound; a higher tone with wobble sounds bell-like and strange.
- Echo, Big hall and Cave: repeats every 0.28 seconds, a reverb of 2 to 4 seconds, or both.
- Old radio, Telephone and Megaphone: the voice squeezed into a narrow band of frequencies with some distortion. Telephone uses the classic 300 to 3,400 Hz voice band of phone networks.
- Underwater, 8-bit game and Backwards: muffled and wobbling, crunchy like an old game console, or reversed.
How it works
Pitch is measured in semitones, the steps between neighbouring piano keys. Moving up 12 semitones doubles every frequency in your voice (one octave), and each semitone multiplies it by 21/12, about 1.0595. To change the pitch without changing the length, the tool first stretches the recording in time without touching its pitch (WSOLA: it copies short overlapping slices of the sound, lined up where the waveforms match), then plays it back faster or slower by the same ratio. The other effects are built from your browser's Web Audio filters, delays and reverb. The finished sound is levelled so it is as loud as your original without clipping.
Example
Say your voice has a fundamental frequency of 120 Hz in a 30-second clip. Chipmunk raises it 9 semitones: 120 × 29/12 = 120 × 1.6818 = 201.8 Hz, and the clip is still 30 seconds long. Deep voice lowers it 5 semitones: 120 × 2−5/12 = 120 × 0.7492 = 89.9 Hz. Saved as an MP3 at 192 kbps, 30 seconds is 192,000 ÷ 8 × 30 = 720,000 bytes, which the tool shows as 703.1 KB. With Echo, Big hall or Cave the file runs a second or two longer, so the effect can fade out instead of stopping dead.
Tips
- Record close to the microphone in a quiet room: the effects make background noise louder too.
- For music rather than voice, try Audio effects (slowed + reverb, 8D, bass boost); to record without effects, use the Audio recorder.
- Use it for fun, videos, games and stories. Don't use a changed voice to mislead or impersonate anyone.
MP3 files are encoded with LAME (lamejs 1.2.1, licensed under the LGPL-3.0), which is loaded unmodified as a separate file. Licence and source code.