Animalese Generator

Each syllable becomes one short sound. English words are split at their vowel groups, so "hungry" is two sounds. The bunny picks from nine sounds (u, ha, ya, hu, nya, ra, hya, wa, i), adds a little "u" at the start of a sentence and a "ha" at the end. The same word always gets the same sounds, so a line sounds identical every time.

No. It borrows the idea that fans call Animalese (fast, pitched-up babble, one sound per letter) but uses none of the game's audio. Synth sounds are generated from scratch with vowel formants in your browser, and Real voice uses the open recordings described below. This site isn't affiliated with Nintendo.

It follows the open-source PyAnimalese project (github.com/hwi-middle/PyAnimalese, MIT license, by Ju Hwijung) and uses its recordings: 19 Korean consonants spoken as short neutral syllables. Each syllable plays the recording for its first consonant, pitched up (about 5 times for the bunny, about 3 for the bear), with a little random pitch per sound. The last letter carries the punctuation: ? bends up, ~ stretches out. For English, the first letter of each syllable picks a similar consonant, so it sounds like quick Korean-flavored babble. Full license text: wimouniv.com/animalese/rec/LICENSE.txt

An exclamation mark makes the ending pop. A question mark bends the last sound upward. A tilde (~) stretches the ending. With a period or nothing, the end drops a little. Commas and spaces add short pauses. Synth adds its own little sounds at the start and end of a sentence; Real voice lets the last letter do that job.

Yes. Save gives you a WAV file (24 kHz, mono, 16-bit). A sentence is usually 1 to 2 seconds, so files are tens of kilobytes, small enough for video editors, stream alerts or a chat message. The file is created on your device and never uploaded.

Yes. Korean, Japanese and other scripts get one sound per character, numbers too. Emoji and symbols are skipped silently.

Turn the volume up first. If it's still quiet, press Speak again: browsers only allow sound after you tap the page, so a shared link never autoplays. Playback uses regular media audio, so it works even with the silent switch on.

Only your text, the sound type, the voice and the speed, stored after the # in the URL. That part never leaves the browser, so it isn't sent to our server or analytics. Anyone you send the link to can read it, though, so don't put secrets in it.