The DSP engine (implemented in private/src/SquiggleBoy.hpp and related files) is built around a polyphonic voice architecture feeding into a quadraphonic effects and mixdown section.
The host runs at SampleTimer::x_hostSampleRate (48 kHz). The engine runs SampleTimer::x_oversample (4) engine samples per host frame, at SampleTimer::x_sampleRate (192 kHz). For each host frame, NonagonWrapper::ProcessHostFrame runs x_oversample engine samples. Each one advances SampleTimer, runs ProcessFrame on process-frame boundaries, and processes one engine sample. The frame’s output is the host-rate sample produced by the output decimator. Engine samples are numbered from 1, so a host frame ends on its sub-sample 0 (SampleTimer::IsHostFrame()), where the encoder matrix computes and recording frames are captured.
Slow one-pole slews and low-frequency phase accumulators keep double-precision state where the per-sample coefficient or increment at 192 kHz falls below float resolution. These are the LFO and random-LFO phases and slews, the Theory of Time parameter filters, the reverb and delay modulation filters, portamento, the PolyXFader mix slew, the sub envelope slew, and the pan oscillator. Their outputs stay float.
The synthesizer consists of 9 voices. Each voice (SquiggleBoyVoice) is a complete synthesizer chain containing:
m_ahd). It also houses a second phase-driven envelope (m_modulationAHD), exposed as a modulation source in the encoder system.The outputs of the 9 voices are panned quadraphonically using a Lissajous LFO. The mixed quadraphonic signal is then sent to the send effect buses and final output, where the send effects consist of:
Finally, the system generates parallel quadraphonic and stereo mixdowns, along with a dedicated sub channel, all glued together by a final multiband saturator.
Source machine details now live in a dedicated page: