Voice Changer

Each voice is a small chain of classic audio effects built with the Web Audio API. The pitch voices, Chipmunk (+4 to +12 semitones), Deep (−3 to −8), Monster and Alien, use a delay-line pitch shifter: two short delay lines whose delay times sweep continuously, cross-faded so there are no clicks. This changes pitch while keeping the original speed and length, so speech stays in sync with any video. At large shifts it adds a slight warble, which suits cartoon and monster voices.

The Robot voice uses ring modulation, multiplying your voice by a 30–90 Hz sine wave. That creates new frequencies unrelated to your voice’s harmonics, the metallic tone of classic science-fiction robots, and a short comb delay adds the tinny ring. Telephone, Old radio and Walkie-talkie are mostly band-pass filtering: a landline passes roughly 300 Hz–3.4 kHz, and radios and walkie-talkies are narrower still, with saturation and a little hiss. Echo / cave adds repeats and a large generated reverb, and Underwater combines a resonant low-pass filter with a slow wobble.

The intensity slider moves each voice’s main settings at once: how far the pitch shifts, how narrow the band is, how much drive or noise there is. Because the same effect chain is used for the live preview and for the export, the file you download sounds exactly like the preview. A natural-sounding change of gender or a copy of a specific person’s voice is a different problem that needs machine-learning models; these effects change the character of a voice, not its identity.