CHIMERA

GAME BRIEF

What's the game?

Press START AND HEAR A SONG and a cutaway of a man's head sings 'Breathe in, breathe out, I am made of air' while the tongue, jaw and lips move. After that you play it: each note sings the next syllable of the lyrics, and you can type your own. SPEAK reads out what you type ('She sells sea shells by the sea shore' is one of the samples) and draws the gestures and a consonant chart as it goes. BEATBOX has drum pads and a step sequencer, and BESTIARY swaps the human for a songbird, a Weddell seal, an American alligator or an African elephant, each with its own calls. Male, female and child voices are buttons at the top. There are no recordings in it: Jameson Hodge says it models the vocal tract as a network of branching tubes, unlike Vocaloid, which stitches together and pitch-shifts recorded sounds.

BEFORE YOU PLAY

What to expect

Free, with no download and no sign-up, and open source under the MIT license. It came to about 1.5 MB in our check, more than half of it a pronunciation dictionary, and it needs your sound on: press START AUDIO or the song button first. There's no score or goal; it's an instrument. Jameson's November 7, 2026 post of it had about 2.6 million views three days later, and people have already built songs, an AI agent that sings back and a yodel synth on its open-source code (links below). We haven't tried it on a phone; his README says phones are the likely next step.

Controls: Mouse: click the tabs at the top (PLAY, SPEAK, SING, BESTIARY, BEATBOX) and the on-screen keys. In PLAY you drag the tongue, lips and jaw in the head. Your computer keyboard and a MIDI keyboard also play notes (the browser asks before using MIDI). In SPEAK, type a sentence and press SPEAK.