Two detuned saw glottal pulses → three parallel band-pass formant resonators (F1/F2/F3) re-tuned phone by phone from a hand-written pronunciation dictionary. Fricatives are band-limited noise, plosives a closure + burst. Choruses are three detuned voices into a clipper.
Saw + square per string, strum spread, per-note filter sweep → tanh waveshaper
(distortion) → tone stack → cab low-pass → hard-panned pair. - grid bars
ring, p bars are palm muted.
Sine fundamental + saw grit through a low-pass with an envelope sweep, light saturation; every kick ducks the bass bus 4 dB for 85 ms so the low end pumps.
Kick = pitch-dropping sine (155→46 Hz) + beater click. Snare = band-passed noise + two triangle shells. Hats/cymbals = high-passed noise. Toms for the fills.
instruments → bus filters → distortion → dotted-8th delay + convolution reverb (generated impulse) → glue compressor → limiter → tanh soft clip.
480×270 canvas upscaled nearest-neighbour. Band actions are read from the same event list the synth schedules: strums, stick hits, kick pedal, and the mouth shape is taken from the phoneme on the timeline right now.