The sound this app already makes

Small, and worth knowing about, because it is the only part of this app that is not made of light.

It pronounces. The twenty-two letters are spoken aloud by the browser's own voice; the Sanskrit letters are spoken in the annex, with an honest warning when no Hindi voice is installed and it is falling back to an English approximation.

It hums each letter at the pitch of its own mouth-position. The letters are grouped by where they are articulated — throat, palate, tongue, teeth, lips — and each group is given its own frequency, low at the throat and rising to the lips. So the tone you hear is not decorative: it is a map of the mouth, and moving up the scale is moving forward through the place where speech is made.

It builds the five vowels out of their formants rather than playing five arbitrary notes — a buzzing source filtered at the frequency that actually distinguishes an ah from an ee. Which is why they sound like a throat and not like a synthesiser.

And it keeps time for a breath, for Abulafia's tempo.

⚠ What it cannot do is hear you, and what it has not yet done is sound a connection rather than a single letter — see app-aspaklaria, and chorez, where the whole argument is that the charge lives in the join.