feat: add crossover distortion to clipping stage

src/dsp/clipping.rs

@@ -4,10 +4,11 @@ use nih_plug::prelude::AtomicF32;
 use nih_plug::util::db_to_gain_fast;
 use num_traits::{Float, ToPrimitive};
 use std::sync::{atomic::Ordering, Arc};
-use valib::math::smooth_clamp;
+use numeric_literals::replace_float_literals;
+use valib::math::{smooth_clamp, smooth_max, smooth_min};
 use valib::saturators::clippers::DiodeClipper;
-use valib::saturators::Slew;
-use valib::simd::SimdComplexField;
+use valib::saturators::{Saturator, Slew};
+use valib::simd::{SimdComplexField, SimdValue};
 use valib::util::lerp;
 use valib::wdf::dsl::*;
 use valib::{
@@ -16,6 +17,41 @@ use valib::{
     Scalar,
 };
 
+struct CrossoverDistortion<T> {
+    pub t: T,
+    pub drift: T,
+}
+
+impl<T: Scalar> CrossoverDistortion<T> {
+    pub fn new(drift: T) -> Self {
+        Self {
+            t: T::from_f64(1e-1),
+            drift,
+        }
+    }
+
+    pub fn from_age(age: T) -> Self {
+        Self::new(Self::drift_from_age(age))
+    }
+
+    pub fn set_age(&mut self, age: T) {
+        self.drift = Self::drift_from_age(age);
+    }
+
+    #[replace_float_literals(T::from_f64(literal))]
+    fn drift_from_age(age: T) -> T {
+        2.96e-3 * age.simd_powf(1./3.)
+    }
+}
+
+impl<T: Scalar> Saturator<T> for CrossoverDistortion<T> {
+    fn saturate(&self, x: T) -> T {
+        let l0 = x + self.drift;
+        let l1 = x - self.drift;
+        l0.simd_min(T::zero().simd_max(l1))
+    }
+}
+
 type Stage1Module<T> = WdfModule<
     ResistiveVoltageSource<T>,
     Series<Capacitor<T>, Series<Resistor<T>, ResistiveVoltageSource<T>>>,
@@ -181,6 +217,7 @@ pub struct ClippingStage<T: Scalar<Element: Float>> {
     pub(crate) led_rms: Rms<f32>,
     pub(crate) led_display: Arc<AtomicF32>,
     samplerate: T,
+    crossover: CrossoverDistortion<T>,
 }
 
 impl<T: Scalar<Element: Float>> ClippingStage<T> {
@@ -196,11 +233,13 @@ impl<T: Scalar<Element: Float>> ClippingStage<T> {
             stage3: Stage3::new(samplerate),
             led_rms: Rms::new(rms_samples),
             led_display: Arc::new(AtomicF32::new(0.0)),
+            crossover: CrossoverDistortion::new(T::zero()),
             samplerate,
         }
     }
 
     pub fn set_age(&mut self, age: T) {
+        self.crossover.set_age(age);
         self.out_slew.max_diff =
             component_matching_slew_rate(self.samplerate, age);
     }
@@ -244,11 +283,13 @@ impl<T: Scalar<Element: ToPrimitive + Float>> DSPProcess<1, 1> for ClippingStage
             .to_f32()
             .unwrap_or_default()
             * 0.5;
+        let led_value = led_value.select(led_value.is_finite(), 0.);
         let led_value = self.led_rms.add_element(led_value);
         self.led_display.store(led_value, Ordering::Relaxed);
+        let vout = self.crossover.saturate(vplus + vf - T::from_f64(4.5)) + T::from_f64(4.5);
         self.out_slew.process([smooth_clamp(
             T::from_f64(0.1),
-            vplus + vf,
+            vout,
             T::zero(),
             T::from_f64(9.),
         )])
@@ -267,6 +308,17 @@ mod tests {
     use valib::dsp::buffer::{AudioBufferMut, AudioBufferRef};
     use valib::dsp::{BlockAdapter, DSPProcessBlock};
 
+    #[test]
+    fn crossover_dc_sweep() {
+        const N: usize = 100;
+        let crossover = CrossoverDistortion::new(1.);
+        let output: [f32; N] = std::array::from_fn(|i| {
+            let x = lerp(i as f32 / N as f32, -5., 5.);
+            crossover.saturate(x)
+        });
+        insta::assert_csv_snapshot!(&output as &[_], { "[]" => insta::rounded_redaction(4) });
+    }
+
     #[test]
     fn drive_test() {
         const SAMPLERATE: f64 = 1024.;

src/dsp/snapshots/ts404__dsp__clipping__tests__crossover_dc_sweep.snap

@@ -0,0 +1,104 @@
+---
+source: src/dsp/clipping.rs
+expression: "&output as &[_]"
+---
+-4.0
+-3.9
+-3.8
+-3.7
+-3.6
+-3.5
+-3.4
+-3.3
+-3.2
+-3.1
+-3.0
+-2.9
+-2.8
+-2.7
+-2.6
+-2.5
+-2.4
+-2.3
+-2.2
+-2.1
+-2.0
+-1.9
+-1.8
+-1.7
+-1.6
+-1.5
+-1.4
+-1.3
+-1.2
+-1.1
+-1.0
+-0.9
+-0.8
+-0.7
+-0.6
+-0.5
+-0.4
+-0.3
+-0.2
+-0.1
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.0
+0.1
+0.2
+0.3
+0.4
+0.5
+0.6
+0.7
+0.8
+0.9
+1.0
+1.1
+1.2
+1.3
+1.4
+1.5
+1.6
+1.7
+1.8
+1.9
+2.0
+2.1
+2.2
+2.3
+2.4
+2.5
+2.6
+2.7
+2.8
+2.9
+3.0
+3.1
+3.2
+3.3
+3.4
+3.5
+3.6
+3.7
+3.8
+3.9