diff --git a/.gitmodules b/.gitmodules
new file mode 100644
index 0000000..d3f66a3
--- /dev/null
+++ b/.gitmodules
@@ -0,0 +1,3 @@
+[submodule "dsp"]
+	path = dsp
+	url = https://github.com/bvontob/logue-dsp-c.git
diff --git a/dsp b/dsp
new file mode 160000
index 0000000..0860d50
--- /dev/null
+++ b/dsp
@@ -0,0 +1 @@
+Subproject commit 0860d5033c92af79d18f7ad6c52d14a098a280eb
diff --git a/inc/biquad.h b/inc/biquad.h
deleted file mode 100644
index f76e859..0000000
--- a/inc/biquad.h
+++ /dev/null
@@ -1,306 +0,0 @@
-#include "float_math.h"
-
-typedef struct biquad_coeffs_s { /* Default constructor set all to 0.0f */
-  float ff0;
-  float ff1;
-  float ff2;
-  float fb1;
-  float fb2;
-} biquad_coeffs;
-
-typedef struct biquad_s { /* Default constructor can use flush() to 0.0f out both */
-  biquad_coeffs c;
-  float mZ1;
-  float mZ2;
-} biquad;
-
-/**
- * Convert Hz frequency to radians
- *
- * @param   fc Frequency in Hz
- * @param   fsrecip Reciprocal of sampling frequency (1/Fs)
- */
-static inline __attribute__((optimize("Ofast"),always_inline))
-float biquad_wc(const float fc, const float fsrecip) {
-  return fc * fsrecip;
-}
-
-/**
- * Calculate coefficients for single pole low pass filter.
- *
- * @param   pole Pole position in radians
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_pole_lp(biquad *bq, const float pole) {
-  bq->c.ff0 = 1.f - pole;
-  bq->c.fb1 = -pole;
-  bq->c.fb2 = bq->c.ff2 = bq->c.ff1 = 0.f;
-}
-
-/**
- * Calculate coefficients for single pole high pass filter.
- *
- * @param   pole Pole position in radians
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_pole_hp(biquad *bq, const float pole) {
-  bq->c.ff0 = 1.f - pole;
-  bq->c.fb1 = pole;
-  bq->c.fb2 = bq->c.ff2 = bq->c.ff1 = 0.f;
-}
-
-/**
- * Calculate coefficients for single pole DC filter.
- *
- * @param   pole Pole position in radians
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_fo_dc(biquad *bq, const float pole) {
-  bq->c.ff0 = 1.f;
-  bq->c.ff1 = -1.f;
-  bq->c.fb1 = -pole;
-  bq->c.fb2 = bq->c.ff2 = 0.f;
-}
-
-/**
- * Calculate coefficients for first order low pass filter.
- *
- * @param   k Tangent of PI x cutoff frequency in radians: tan(pi*wc)
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_fo_lp(biquad *bq, const float k) {
-  const float kp1 = k + 1.f;
-  const float km1 = k - 1.f;
-  bq->c.ff0 = bq->c.ff1 = k / kp1;
-  bq->c.fb1 = km1 / kp1;
-  bq->c.fb2 = bq->c.ff2 = 0.f;
-}
-
-/**
- * Calculate coefficients for first order high pass filter.
- *
- * @param   k Tangent of PI x cutoff frequency in radians: tan(pi*wc)
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_fo_hp(biquad *bq, const float k) {
-  // k = tan(pi*wc)
-  const float kp1 = k + 1.f;
-  const float km1 = k - 1.f;
-  bq->c.ff0 = 1.f / kp1;
-  bq->c.ff1 = -bq->c.ff0;
-  bq->c.fb1 = km1 / kp1;
-  bq->c.fb2 = bq->c.ff2 = 0.f;
-}
-
-/**
- * Calculate coefficients for first order all pass filter.
- *
- * @param   k Tangent of PI x cutoff frequency in radians: tan(pi*wc)
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_fo_ap(biquad *bq, const float k) {
-  // k = tan(pi*wc)
-  const float kp1 = k + 1.f;
-  const float km1 = k - 1.f;
-  bq->c.ff0 = bq->c.fb1 = km1 / kp1;
-  bq->c.ff1 = 1.f;
-  bq->c.fb2 = bq->c.ff2 = 0.f;
-}
-
-/**
- * Calculate coefficients for first order all pass filter.
- *
- * @param   wc cutoff frequency in radians
- *
- * @note Alternative implementation with no tangeant lookup
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_fo_ap2(biquad *bq, const float wc) {
-  // Note: alternative implementation for use in phasers
-  const float g1 = 1.f - wc;
-  bq->c.ff0 = g1;
-  bq->c.ff1 = -1;
-  bq->c.fb1 = -g1;
-  bq->c.fb2 = bq->c.ff2 = 0.f;
-}
-
-/**
- * Calculate coefficients for second order DC filter.
- *
- * @param   pole Pole position in radians
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_so_dc(biquad *bq, const float pole) {
-  bq->c.ff0 = bq->c.ff2 = 1.f;
-  bq->c.ff1 = 2.f;
-  bq->c.fb1 = -2.f * pole;
-  bq->c.fb2 = pole * pole;
-}
-
-/**
- * Calculate coefficients for second order low pass filter.
- *
- * @param   k Tangent of PI x cutoff frequency in radians: tan(pi*wc)
- * @param   q Resonance with flat response at q = sqrt(2)
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_so_lp(biquad *bq, const float k, const float q) {
-  // k = tan(pi*wc)
-  // flat response at q = sqrt(2)
-  const float qk2 = q * k * k;
-  const float qk2_k_q_r = 1.f / (qk2 + k + q);
-  bq->c.ff0 = bq->c.ff2 = qk2 * qk2_k_q_r;
-  bq->c.ff1 = 2.f * bq->c.ff0;
-  bq->c.fb1 = 2.f * (qk2 - q) * qk2_k_q_r;
-  bq->c.fb2 = (qk2 - k + q) * qk2_k_q_r;
-}
-
-/**
- * Calculate coefficients for second order high pass filter.
- *
- * @param   k Tangent of PI x cutoff frequency in radians: tan(pi*wc)
- * @param   q Resonance with flat response at q = sqrt(2)
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_so_hp(biquad *bq, const float k, const float q) {
-  // k = tan(pi*wc)
-  // flat response at q = sqrt(2)
-  const float qk2 = q * k * k;
-  const float qk2_k_q_r = 1.f / (qk2 + k + q);
-  bq->c.ff0 = bq->c.ff2 = q * qk2_k_q_r;
-  bq->c.ff1 = -2.f * bq->c.ff0;
-  bq->c.fb1 = 2.f * (qk2 - q) * qk2_k_q_r;
-  bq->c.fb2 = (qk2 - k + q) * qk2_k_q_r;
-}
-
-/**
- * Calculate coefficients for second order band pass filter.
- *
- * @param   k Tangent of PI x cutoff frequency in radians: tan(pi*wc)
- * @param   q Resonance with flat response at q = sqrt(2)
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_so_bp(biquad *bq, const float k, const float q) {
-  // k = tan(pi*wc)
-  // q is inverse of relative bandwidth (Fc / Fb)
-  const float qk2 = q * k * k;
-  const float qk2_k_q_r = 1.f / (qk2 + k + q);
-  bq->c.ff0 = k * qk2_k_q_r;
-  bq->c.ff1 = 0.f;
-  bq->c.ff2 = -bq->c.ff0;
-  bq->c.fb1 = 2.f * (qk2 - q) * qk2_k_q_r;
-  bq->c.fb2 = (qk2 - k + q) * qk2_k_q_r;
-}
-
-/**
- * Calculate coefficients for second order band reject filter.
- *
- * @param   k Tangent of PI x cutoff frequency in radians: tan(pi*wc)
- * @param   q Resonance with flat response at q = sqrt(2)
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_so_br(biquad *bq, const float k, const float q) {
-  // k = tan(pi*wc)
-  // q is inverse of relative bandwidth (Fc / Fb)
-  const float qk2 = q * k * k;
-  const float qk2_k_q_r = 1.f / (qk2 + k + q);
-  bq->c.ff0 = bq->c.ff2 = (qk2 + q) * qk2_k_q_r;
-  bq->c.ff1 = bq->c.fb1 = 2.f * (qk2 - q) * qk2_k_q_r;
-  bq->c.fb2 = (qk2 - k + q) * qk2_k_q_r;
-}
-
-/**
- * Calculate coefficients for second order all pass filter.
- *
- * @param   k Tangent of PI x cutoff frequency in radians: tan(pi*wc)
- * @param   q Inverse of relative bandwidth (Fc / Fb)
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_so_ap(biquad *bq, const float k, const float q) {
-  // k = tan(pi*wc)
-  // q is inverse of relative bandwidth (Fc / Fb)
-  const float qk2 = q * k * k;
-  const float qk2_k_q_r = 1.f / (qk2 + k + q);
-  bq->c.ff0 = bq->c.fb2 = (qk2 - k + q) * qk2_k_q_r;
-  bq->c.ff1 = bq->c.fb1 = 2.f * (qk2 - q) * qk2_k_q_r;
-  bq->c.ff2 = 1.f;
-}
-
-/**
- * Calculate coefficients for second order all pass filter.
- *
- * @param   delta cos(2pi*wc)
- * @param   gamma tan(pi * wb)
- *
- * @note q is inverse of relative bandwidth (wc / wb)
- * @note Alternative implementation, so called "tunable" in DAFX second edition.
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_so_ap2(biquad *bq, const float delta, const float gamma) {
-  // Note: Alternative implementation .. so called "tunable" in DAFX.
-  // delta = cos(2pi*wc)
-  const float c = (gamma - 1.f) / (gamma + 1.f);
-  const float d = -delta;
-  bq->c.ff0 = bq->c.fb2 = -c;
-  bq->c.ff1 = bq->c.fb1 = d * (1.f - c);
-  bq->c.ff2 = 1.f;
-}
-
-/**
- * Calculate coefficients for second order all pass filter.
- *
- * @param   delta cos(2pi*wc)
- * @param   radius 
- *
- * @note Another alternative implementation.
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_so_ap3(biquad *bq, const float delta, const float radius) {
-  // Note: alternative implementation for use in phasers
-  // delta = cos(2pi * wc)
-  const float a1 = -2.f * radius * delta;
-  const float a2 = radius * radius;
-  bq->c.ff0 = bq->c.fb2 = a2;
-  bq->c.ff1 = bq->c.fb1 = a1;
-  bq->c.ff2 = 1.f;
-}
-
-/**
- * Flush internal delays
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-void biquad_flush(biquad *bq) {
-  bq->mZ1 = bq->mZ2 = 0;
-}
-
-/**
- * Second order processing of one sample
- *
- * @param xn  Input sample
- *
- * @return Output sample
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-float biquad_process_so(biquad *bq, const float xn) {
-  float acc = bq->c.ff0 * xn + bq->mZ1;
-  bq->mZ1   = bq->c.ff1 * xn + bq->mZ2;
-  bq->mZ2   = bq->c.ff2 * xn;
-  bq->mZ1  -= bq->c.fb1 * acc;
-  bq->mZ2  -= bq->c.fb2 * acc;
-  return acc;
-}
-
-/**
- * First order processing of one sample
- *
- * @param xn  Input sample
- *
- * @return Output sample
- */
-inline __attribute__((optimize("Ofast"),always_inline))
-float biquad_process_fo(biquad *bq, const float xn) {
-  float acc = bq->c.ff0 * xn + bq->mZ1;
-  bq->mZ1   = bq->c.ff1 * xn;
-  bq->mZ1  -= bq->c.fb1 * acc;
-  return acc;
-}
diff --git a/inc/delay_line.h b/inc/delay_line.h
deleted file mode 100644
index fc1f5ec..0000000
--- a/inc/delay_line.h
+++ /dev/null
@@ -1,93 +0,0 @@
-#include "float_math.h"
-#include "int_math.h"
-#include "buffer_ops.h"
-
-#ifndef __sdram
-#define __sdram __attribute__((section(".sdram")))
-#endif
-
-typedef struct delay_line_s {
-  float   *line;
-  float    fracz;
-  size_t   size;
-  size_t   mask;
-  uint32_t wr_idx;
-} delay_line;
-  
-static inline __attribute__((optimize("Ofast"), always_inline))
-void delay_line_clear(delay_line *dl) {
-  buf_clr_f32((float *)dl->line, dl->size);
-}
-
-/**
- * Constructor with explicit memory area to use as backing buffer for
- * delay line.
- *
- * @param ram Pointer to memory buffer
- * @param line_size Size in float of memory buffer (must be power of 2)
- */
-static inline __attribute__((optimize("Ofast"), always_inline))
-void delay_line_init(delay_line *dl, float *ram, size_t line_size) {
-  dl->line   = ram;
-  dl->fracz  = 0;
-  dl->size   = nextpow2_u32(line_size);
-  dl->mask   = dl->size - 1;
-  dl->wr_idx = 0;
-  delay_line_clear(dl);
-}
-
-/**
- * Write a single sample to the head of the delay line
- *
- * @param s Sample to write
- */
-static inline __attribute__((optimize("Ofast"), always_inline))
-void delay_line_write(delay_line *dl, const float s) {
-  dl->line[(dl->wr_idx--) & dl->mask] = s;
-}
-
-/**
- * Read a single sample from the delay line at given position from
- * current write index.
- *
- * @param pos Offset from write index
- * @return Sample at given position from write index
- */
-static inline __attribute__((optimize("Ofast"), always_inline))
-float delay_line_read(const delay_line *dl, const uint32_t pos) {
-  return dl->line[(dl->wr_idx + pos) & dl->mask];
-}
-
-/**
- * Read a sample from the delay line at a fractional position from
- * current write index.
- *
- * @param pos Offset from write index as floating point.
- * @return Interpolated sample at given fractional position from write index
- */
-static inline __attribute__((optimize("Ofast"), always_inline))
-float delay_line_read_frac(const delay_line *dl, const float pos) {
-  const uint32_t base = (uint32_t)pos;
-  const float frac = pos - base;
-  const float s0 = delay_line_read(dl, base);
-  const float s1 = delay_line_read(dl, base + 1);
-  return linintf(frac, s0, s1);
-}
-
-/**
- * Read a sample from the delay line at a position from current write
- * index with interpolation from last read.
- *
- * @param pos Offset from write index
- * @param frac Interpolation from last read pair.
- * @return Interpolation of last read sample and sample at given position from write index.
- */
-static inline __attribute__((optimize("Ofast"), always_inline))
-float delay_line_read_fracz(delay_line *dl,
-			    const uint32_t pos,
-			    const float frac) {
-  const float s0 = delay_line_read(dl, pos);
-  const float y = linintf(frac, s0, dl->fracz);
-  dl->fracz = s0;
-  return y;
-}
diff --git a/project.mk b/project.mk
index 86e2816..2c3a327 100644
--- a/project.mk
+++ b/project.mk
@@ -8,7 +8,7 @@ UCSRC = main.c
 
 UCXXSRC =
 
-UINCDIR = inc
+UINCDIR = dsp
 
 UDEFS =