Add bin segmenter

src/R3StretcherImpl.h

@@ -27,8 +27,9 @@
 #include <map>
 #include <memory>
 
-#include "dsp/BinClassifier.h"
+#include "dsp/BinSegmenter.h"
 #include "dsp/FFT.h"
+#include "system/Allocators.h"
 
 namespace RubberBand
 {
@@ -88,14 +89,6 @@ protected:
         Range channelLock;
     };
 
-    struct BinSegmentation {
-        float percussiveBelow;
-        float percussiveAbove;
-        float residualAbove;
-        BinSegmentation(float _pb, float _pa, float _ra) :
-            percussiveBelow(_pb), percussiveAbove(_pa), residualAbove(_ra) { }
-    };
-
     struct ChannelScaleData {
         int fftSize;
         int bufSize; // size of every array here: fftSize/2 + 1
@@ -134,10 +127,10 @@ protected:
     
     struct ChannelData {
         std::map<int, std::shared_ptr<ChannelScaleData>> scales;
-        std::unique_ptr<BinClassifier> classifier;
+        std::unique_ptr<BinSegmenter> segmenter;
-        BinSegmentation segmentation;
+        BinSegmenter::Segmentation segmentation;
-        BinSegmentation prevSegmentation;
+        BinSegmenter::Segmentation prevSegmentation;
-        BinSegmentation nextSegmentation;
+        BinSegmenter::Segmentation nextSegmentation;
         Guidance guidance;
     };
     

src/dsp/BinClassifier.h

@@ -33,8 +33,9 @@
 
 namespace RubberBand {
 
-class BinClassifier {
+class BinClassifier
-
+{
+public:
     enum class Classification {
         Harmonic = 0,
         Percussive = 1,
@@ -104,21 +105,9 @@ class BinClassifier {
             m_hFilters[i]->push(mag[i]);
             m_hf[i] = m_hFilters[i]->get();
         }
-        
+
-        m_vFilter->reset();
+        v_copy(m_vf, mag, n);
-        int vFilterLag = m_parameters.verticalFilterLength / 2;
+        MovingMedian<float>::filter(*m_vFilter, m_vf);
-        
-        for (int i = 0; i < vFilterLag; ++i) {
-            m_vFilter->push(mag[i]);
-        }
-        for (int i = vFilterLag; i < n; ++i) {
-            m_vFilter->push(mag[i]);
-            m_vf[i-vFilterLag] = m_vFilter->get();
-        }
-        for (int i = n; i < n + vFilterLag; ++i) {
-            m_vFilter->push(0.f);
-            m_vf[i-vFilterLag] = m_vFilter->get();
-        }
 
         if (m_parameters.horizontalFilterLag > 0) {
             float *lagged = m_vfQueue.readOne();
@@ -152,6 +141,9 @@ protected:
     float *m_hf;
     float *m_vf;
     RingBuffer<float *> m_vfQueue;
+
+    BinClassifier(const BinClassifier &) =delete;
+    BinClassifier &operator=(const BinClassifier &) =delete;
 };
 
 }

src/dsp/BinSegmenter.h

@@ -0,0 +1,129 @@
+/* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */
+
+/*
+    Rubber Band Library
+    An audio time-stretching and pitch-shifting library.
+    Copyright 2007-2022 Particular Programs Ltd.
+
+    This program is free software; you can redistribute it and/or
+    modify it under the terms of the GNU General Public License as
+    published by the Free Software Foundation; either version 2 of the
+    License, or (at your option) any later version.  See the file
+    COPYING included with this distribution for more information.
+
+    Alternatively, if you have a valid commercial licence for the
+    Rubber Band Library obtained by agreement with the copyright
+    holders, you may redistribute and/or modify it under the terms
+    described in that licence.
+
+    If you wish to distribute code using the Rubber Band Library
+    under terms other than those of the GNU General Public License,
+    you must obtain a valid commercial licence before doing so.
+*/
+
+#ifndef RUBBERBAND_BIN_SEGMENTER_H
+#define RUBBERBAND_BIN_SEGMENTER_H
+
+#include "BinClassifier.h"
+#include <vector>
+
+namespace RubberBand {
+
+class BinSegmenter
+{
+public:
+    struct Segmentation {
+        double percussiveBelow;
+        double percussiveAbove;
+        double residualAbove;
+        Segmentation(double _pb, double _pa, double _ra) :
+            percussiveBelow(_pb), percussiveAbove(_pa), residualAbove(_ra) { }
+    };
+
+    struct Parameters {
+        int fftSize;
+        double sampleRate;
+        Parameters(int _fftSize, double _sampleRate) :
+            fftSize(_fftSize), sampleRate(_sampleRate) { }
+    };
+    
+    BinSegmenter(Parameters parameters,
+                 BinClassifier::Parameters classifierParameters) :
+        m_parameters(parameters),
+        m_classifierParameters(classifierParameters),
+        m_classifier(classifierParameters),
+        m_classification(classifierParameters.binCount,
+                         BinClassifier::Classification::Silent),
+        m_numeric(classifierParameters.binCount, 0),
+        m_classFilter(classifierParameters.binCount / 64)
+    {
+    }
+
+    Segmentation segment(const float *const mag) {
+        int n = m_classifierParameters.binCount;
+        m_classifier.classify(mag, m_classification.data());
+        for (int i = 0; i < n; ++i) {
+            switch (m_classification[i]) {
+            case BinClassifier::Classification::Harmonic:
+                m_numeric[i] = 0; break;
+            case BinClassifier::Classification::Percussive:
+                m_numeric[i] = 1; break;
+            default:
+                m_numeric[i] = 2; break;
+            }
+        }
+        MovingMedian<int>::filter(m_classFilter, m_numeric.data());
+        double f0 = 0.0;
+        for (int i = 1; i < n; ++i) {
+            if (m_numeric[i] != 1) {
+                f0 = frequencyForBin(i);
+                break;
+            }
+        }
+        double nyquist = m_parameters.sampleRate / 2.0;
+        int top = binForFrequency(16000.0);
+        if (top >= n) top = n-1;
+        double f1 = nyquist;
+        double f2 = nyquist;
+        bool inPercussive = false;
+        for (int i = top; i > 0; --i) {
+            if (m_numeric[i] == 1) { // percussive
+                if (!inPercussive) {
+                    inPercussive = true;
+                    f2 = frequencyForBin(i);
+                    continue;
+                }
+            } else if (m_numeric[i] == 0) { // harmonic
+                if (inPercussive) {
+                    f1 = frequencyForBin(i);
+                }
+                break; // always when harmonic reached
+            }
+        }
+        return Segmentation(f0, f1, f2);
+    }
+
+protected:
+    Parameters m_parameters;
+    BinClassifier::Parameters m_classifierParameters;
+    BinClassifier m_classifier;
+    std::vector<BinClassifier::Classification> m_classification;
+    std::vector<int> m_numeric;
+    MovingMedian<int> m_classFilter;
+
+    int binForFrequency(double f) {
+        return int(round(f * double(m_parameters.fftSize) /
+                         m_parameters.sampleRate));
+    }
+    double frequencyForBin(int b) {
+        return (double(b) * m_parameters.sampleRate)
+            / double(m_parameters.fftSize);
+    }
+    
+    BinSegmenter(const BinSegmenter &) =delete;
+    BinSegmenter &operator=(const BinSegmenter &) =delete;
+};
+
+}
+
+#endif

src/dsp/MovingMedian.h

@@ -80,6 +80,27 @@ public:
 	v_zero(m_sorted, P::m_size);
     }
 
+    // Convenience function that applies a given filter to an array
+    // in-place. Array must have length equal to getSize(). Modifies
+    // both the filter and the array.
+    //
+    static void filter(MovingMedian<T> &mm, T *v) {
+        int n = mm.getSize();
+        int lag = n / 2;
+        mm.reset();
+        for (int i = 0; i < lag; ++i) {
+            mm.push(v[i]);
+        }
+        for (int i = lag; i < n; ++i) {
+            mm.push(v[i]);
+            v[i-lag] = mm.get();
+        }
+        for (int i = n; i < n + lag; ++i) {
+            mm.push(T());
+            v[i-lag] = mm.get();
+        }
+    }
+    
 private:
     T *const m_frame;
     T *const m_sorted;