The crackle was software (our audio graph), not hardware: - Peaks past 0 dBFS were being hard-clipped by the DAC (resonance up to +27 dB and delay feedback easily exceed full scale). Added a brickwall-style DynamicsCompressor (-3 dB, 20:1, 2 ms attack) before the destination; worst-case output now peaks around -6 dB RMS. - LFO cutoff modulation is clamped below the cutoff value so the filter frequency cannot be driven negative (high-Q biquad near 0 Hz sputters audibly). - Envelope retriggers use cancelAndHoldAtTime (with fallback) instead of cancel+set, removing step-discontinuity clicks on fast replays. - Synth meter clip threshold moved to -1 dBFS: the LED now reads as a limiter-activity indicator since true output clipping is gone. |
||
|---|---|---|
| .. | ||
| docs | ||