Vocal Tract Synthesizer

A source–filter voice: a synthesized glottal pulse is re‑shaped by four band‑pass “formants” that track the shape of your vocal tract (pharynx + mouth) and a nasal branch (sinuses), so moving the tongue, lips and soft palate retunes the sound exactly like the real articulators do. It also speaks — every English phoneme is a gesture on a keyframe timeline that drives the audio graph and this drawing from the same curve.

Sentence composer →

Vocal tract — sagittal view

back · mid open · slightly rounded /ə

The drawing is the live control surface: the tongue blob, jaw, lips, velum and the beating vocal folds all follow the sliders. The four colored dots are the resonances they produce.

Articulation

Click/drag inside the pad: horizontal = back↔front, vertical = close↔open.

Voice

Space phonate · ←→ tongue · ↑↓ height · o/a mouth shape · y yell

Phonemes & phrases — click to watch the gesture animate

Intonation
drag the right edge of a block to time that phoneme — click a block to hear it alone
now sounding / — / pick a phoneme, or type a word and press Say

Every block on the lane is a phoneme’s length: drag its right edge to hold that sound longer or cut it short, and everything after it slides to match. Lengths are held as multiples of each phoneme’s natural gesture, so speed still scales the phrase you have just timed. The lane above the curves is the same curve the diagram and the audio play: tongue frontness jaw opening voicing closure f0 contour. Consecutive targets glide into each other, which is what makes co-articulation fall out of the model, and the f0 lane is the phrase‑level prosody riding on top of the pitch slider — auto declines a statement the way a spent breath does and lifts the last syllable of a question. Because that contour is keyed to the fraction of the phrase, stretching one phoneme also slides the prosody along the word. space free-play interrupts playback. With choke on, a word that lands on a plosive pops and one that lands on a fricative hisses out, then goes quiet — instead of trailing off into a phantom uh.

Output spectrum — what the tract filter leaves of the glottal buzz