DETERMINISTIC
The same text always produces the same audio, bit-for-bit, across every verified platform. No clocks. No drift.
▸ SIGNAL ACQUIRED — SEMANTIC AUDIO PROTOCOL
EVERY WORD
LEAVES AN
ECHO.
dootdoot reads text and answers in droid—short bursts of warm, warbly chatter. Deterministic to the last bit on verified platforms. Meanings that sit close together sound close together, so the dialect can be learned by ear.
$ brew install skeswa/tap/dootdoot
🍺 dootdoot v0.4.2 poured
$ dootdoot "hello, little one"
doot · 0.9s of chatter · deterministic, bit-for-bit
$ dootdoot "hello, little one" -o echo.wav
wrote echo.wav · sha256 matches golden ✓
$
TRANSMISSION CONSOLE
RENDERED LOCALLY BY VOICE_V12 WEBASSEMBLY — SAME PHRASE, SAME CHATTER
VOICE_V12 on standby
▸ THE DIALECT
THREE LAWS OF THE VOICE
SEMANTIC
Text with similar meaning sounds similar. Paraphrases rhyme; distant ideas diverge. The dialect can be learned by ear.
DROID BY DESIGN
However you phrase the input, the output is unmistakably the same character. One bounded voice, never recorded.
▸ SIGNAL PATH
MEANING, ROUTED TO SOUND
01 / INPUTTEXTWordPiece units
→02 / VECTORMEANINGFour audible axes
→03 / PLANGESTUREPhrase + affect
→04 / OUTPUTVOICEFormant synthesis
▸ ARCHIVE ACCESS
THE FIELD MANUAL
AUTHORITATIVE DOCS — BUILT DIRECTLY FROM THE REPOSITORY
PROTOCOL / THE SOUND LANGUAGE
Every phrase maps into a small semantic space.
Every point in that space renders as a run of chirps, whistles, and warbles. Two phrases that mean nearly the same thing land near one another—and sound like it.
$ dootdoot "hello, little one" -o echo.wav
doot · rendered locally · VOICE_V12 ✓◆ FIELD NOTE The browser console and the CLI run the same Rust renderer. There is no imitation voice.