The 2-Pole SVF Filter Machine is one of three filter options available per voice in the DSP engine (FilterSection in private/src/SquiggleBoy.hpp), alongside the Naive and Conservative Reupping Ladders.
This filter machine consists of two 2-pole State Variable Filters (SVF) in series, interspersed with non-linear saturation stages:
TPTSVFLowPassFilterGeneric).TPTSVFHighPassFilterGeneric).The audio signal from the source machine passes through the low-pass SVF first, then the sample rate reducer, and the reduced signal is then processed by the high-pass SVF. The high-pass follows the reducer so it also removes low-frequency content the reducer creates.
m_vcoBaseFreq) multiplied by m_hpCutoffFactor. In Hz, the LP cutoff is 261.6256 * (vcoHz / 261.6256)^0.707 * hpCutoffFactor * lpCutoffFactor. LP tracks 0.707 octaves per VCO octave around middle C (261.6256 Hz), while the HP cutoff factor multiplies both cutoffs linearly. At a middle-C VCO frequency, the original LP/HP relationship is preserved for every HP cutoff factor. The lpCutoffFactor remains an ExpParam(0.25, 1024); both cutoffs are independently clamped at the base-rate Nyquist frequency.Q = 0.5 * 11^resonance. Cutoff is the prewarped natural frequency (g = tan(pi * cutoff)), so each LP/HP section has magnitude Q at cutoff: -6.02 dB at zero resonance, and -3.01 dB only at Q = 1/sqrt(2). The wrappers preserve the engine-rate 0.499 cutoff ceiling. Coefficients use Math4096 trig and generated audio uses explicit FMA.TanhSaturator placed before the low-pass filter, between the low-pass filter and the sample rate reducer, and after the high-pass filter. This drives the filters into non-linear distortion, controlled by m_saturationGain.SampleRateReducer) adds digital aliasing and grit. Aliases folded below the high-pass cutoff are attenuated by the high-pass.