Untitled (work in progress)

speak in English — recognized letters trigger a waveform per phonetic texture, on a radian scale, in distorted order

Distorted sound sequence

vowel — sine soft — triangle sharp — sawtooth hard — square 🎤 raw voice grain — reordered/pitch-warped 🔁 stutter — chopped and repeated ⏪ reverse — backwards 🐢 slowmo — half speed

Settings

Distortion 0.80

Buffer window 700 ms

Master volume -6 dB

Noise gate 0.010

mic level (last grain): – — set the gate just above the level you see while silent, below the level while speaking

Voice delay 2500 ms

holds back the distorted-voice layers (not Original) so they land with the letter sounds, which only fire after a phrase is recognized

Letter sound generation off = STT keeps transcribing, but stops triggering waveforms/harmony

Original voice -14 dB

⚠️ raw mic passthrough — use headphones, or it can feed back through open speakers.