diff --git a/sdrbase/audio/audiofilter.cpp b/sdrbase/audio/audiofilter.cpp
index 696a94299..5cf963c96 100644
--- a/sdrbase/audio/audiofilter.cpp
+++ b/sdrbase/audio/audiofilter.cpp
@@ -14,11 +14,16 @@
 // along with this program. If not, see <http://www.gnu.org/licenses/>.          //
 ///////////////////////////////////////////////////////////////////////////////////
 
+#include <math.h>
+#include <algorithm>
+#include <QDebug>
+
 #include "audiofilter.h"
 
+// f(-3dB) = 3.6 kHz @ 48000 Hz SR (w = 0.0375):
 const float AudioFilter::m_lpa[3] = {1.0,           1.392667E+00, -5.474446E-01};
 const float AudioFilter::m_lpb[3] = {3.869430E-02,  7.738860E-02,  3.869430E-02};
-// f(-3dB) = 300 Hz @ 8000 Hz SR (w = 0.075):
+// f(-3dB) = 300 Hz @ 8000 Hz SR (w = 0.0375):
 const float AudioFilter::m_hpa[3] = {1.000000e+00,  1.667871e+00, -7.156964e-01};
 const float AudioFilter::m_hpb[3] = {8.459039e-01, -1.691760e+00,  8.459039e-01};
 
@@ -26,12 +31,166 @@ AudioFilter::AudioFilter() :
         m_filterLP(m_lpa, m_lpb),
         m_filterHP(m_hpa, m_hpb),
         m_useHP(false)
-{
-}
+{}
 
 AudioFilter::~AudioFilter()
 {}
 
+
+void AudioFilter::setDecimFilters(int sr, uint32_t decim)
+{
+    int downSR = sr / (decim == 0 ? 1 : decim);
+    double fcH = (0.45 * downSR) / (sr <= 0 ? 1 : sr); // high cut frequency normalized to SR
+    double fcL = 300.0 / downSR; // low cut frequency normalized to downsampled SR
+
+    calculate2(false, fcH, m_lpva, m_lpvb);
+    calculate2(true, fcL, m_hpva, m_hpvb);
+
+    m_filterLP.setCoeffs(m_lpva, m_lpvb);
+    m_filterHP.setCoeffs(m_hpva, m_hpvb);
+}
+
+void AudioFilter::calculate2(bool highPass, double fc, float *va, float *vb)
+{
+    double a[22], b[22];
+
+    cheby(highPass, fc, 0.5, 2, a, b); // low-pass, 0.5% ripple, 2 pole filter
+
+    // Copy to the 2-pole filter coefficients
+    for (int i=0; i<3; i++) {
+        vb[i] = a[i];
+        va[i] = b[i];
+    }
+
+    va[0] = 1.0;
+
+    qDebug() << "AudioFilter::calculate2:"
+        << " highPass: " << highPass
+        << " fc: " << fc
+        << " a0: " << va[0]
+        << " a1: " << va[1]
+        << " a2: " << va[2]
+        << " b0: " << vb[0]
+        << " b1: " << vb[1]
+        << " b2: " << vb[2];
+}
+
+/*
+ * Adapted from BASIC program in table 20-4 of
+ * https://www.analog.com/media/en/technical-documentation/dsp-book/dsp_book_Ch20.pdf
+ */
+void AudioFilter::cheby(bool highPass, double fc, float pr, int np, double *a, double *b)
+{
+    double a0, a1, a2, b1, b2;
+    double ta[22], tb[22];
+
+    std::fill(a, a+22, 0.0);
+    std::fill(b, b+22, 0.0);
+    a[2] = 1.0;
+    b[2] = 1.0;
+
+    for (int p = 1; p <= np/2; p++)
+    {
+        cheby_sub(highPass, fc, pr, np, p, a0, a1, a2, b1, b2);
+
+        // Add coefficients to the cascade
+        for (int i=0; i<22; i++)
+        {
+            ta[i] = a[i];
+            tb[i] = b[i];
+        }
+
+        for (int i=2; i<22; i++)
+        {
+            a[i] = a0*ta[i] + a1*ta[i-1] + a2*ta[i-2];
+            b[i] = tb[i] - b1*tb[i-1] - b2*tb[i-2];
+        }
+    }
+
+    // Finish combining coefficients
+    b[2] = 0;
+
+    for (int i=0; i<20; i++)
+    {
+        a[i] = a[i+2];
+        b[i] = -b[i+2];
+    }
+
+    // Normalize the gain
+    double sa = 0.0;
+    double sb = 0.0;
+
+    for (int i=0; i<20; i++)
+    {
+        if (highPass)
+        {
+            sa += i%2 == 0 ? a[i] : -a[i];
+            sb += i%2 == 0 ? b[i] : -b[i];
+        }
+        else
+        {
+            sa += a[i];
+            sb += b[i];
+        }
+    }
+
+    double gain = sa/(1.0 -sb);
+
+    for (int i=0; i<20; i++) {
+        a[i] /= gain;
+    }
+}
+
+/*
+ * Adapted from BASIC subroutine in table 20-5 of
+ * https://www.analog.com/media/en/technical-documentation/dsp-book/dsp_book_Ch20.pdf
+ */
+void AudioFilter::cheby_sub(bool highPass, double fc, float pr, int np, int stage, double& a0, double& a1, double& a2, double& b1, double& b2)
+{
+    double rp = -cos((M_PI/(np*2)) + (stage-1)*(M_PI/np));
+    double ip = sin((M_PI/(np*2)) + (stage-1)*(M_PI/np));
+
+    // Warp from a circle to an ellipse
+    double esx = 100.0 / (100.0 - pr);
+    double es = sqrt(esx*esx -1.0);
+    double vx = (1.0/np) * log((1.0/es) + sqrt((1.0/(es*es)) + 1.0));
+    double kx = (1.0/np) * log((1.0/es) + sqrt((1.0/(es*es)) - 1.0));
+    kx = (exp(kx) + exp(-kx))/2.0;
+    rp = rp * ((exp(vx) - exp(-vx))/2.0) / kx;
+    ip = ip * ((exp(vx) + exp(-vx))/2.0) / kx;
+
+    double t = 2.0 * tan(0.5);
+    double w = 2.0 * M_PI * fc;
+    double m = rp*rp + ip*ip;
+    double d = 4.0 - 4.0*rp*t + m*t*t;
+    double x0 = (t*t)/d;
+    double x1 = (2.0*t*t)/d;
+    double x2 = (t*t)/d;
+    double y1 = (8.0 - 2.0*m*t*t)/d;
+    double y2 = (-4.0 - 4.0*rp*t - m*t*t)/d;
+    double k;
+
+    if (highPass) {
+        k = -cos(w/2.0 + 0.5) / cos(w/2.0 - 0.5);
+    } else {
+        k = sin(0.5 - w/2.0) / sin(0.5 + w/2.0);
+    }
+
+    d = 1.0 + y1*k - y2*k*k;
+
+    a0 = (x0 - x1*k + x2*k*k)/d;
+    a1 = (-2.0*x0*k + x1 + x1*k*k - 2.0*x2*k)/d;
+    a2 = (x0*k*k - x1*k + x2)/d;
+    b1 = (2.0*k + y1 + y1*k*k - 2.0*y2*k)/d;
+    b2 = (-(k*k) - y1*k + y2)/d;
+
+    if (highPass)
+    {
+        a1 = -a1;
+        b1 = -b1;
+    }
+}
+
 float AudioFilter::run(const float& sample)
 {
     return m_useHP ? m_filterLP.run(m_filterHP.run(sample)) : m_filterLP.run(sample);
@@ -45,4 +204,4 @@ float AudioFilter::runHP(const float& sample)
 float AudioFilter::runLP(const float& sample)
 {
     return m_filterLP.run(sample);
-}
\ No newline at end of file
+}
diff --git a/sdrbase/audio/audiofilter.h b/sdrbase/audio/audiofilter.h
index 4a003fc55..de2dbddf6 100644
--- a/sdrbase/audio/audiofilter.h
+++ b/sdrbase/audio/audiofilter.h
@@ -21,7 +21,7 @@
 #include "dsp/iirfilter.h"
 
 /**
- * This is a 2 pole lowpass Chebyshev (recursive) filter at fc=0.075 using coefficients found in table 20-1 of
+ * By default this is a 2 pole lowpass Chebyshev (recursive) filter at fc=0.075 using coefficients found in table 20-1 of
  * http://www.analog.com/media/en/technical-documentation/dsp-book/dsp_book_Ch20.pdf
  *
  * At the interpolated sampling frequency of 48 kHz the -3 dB corner is at 48 * .075 = 3.6 kHz which is perfect for voice
@@ -36,6 +36,8 @@
  *
  * This one works directly with floats
  *
+ * It can be generalized using the program found in tables 20-4 and 20-5 of the same document. This form is used as a
+ * decimation filter and can be set with the setDecimFilters method
  */
 
 class SDRBASE_API AudioFilter {
@@ -45,18 +47,29 @@ public:
 
     void useHP(bool useHP) { m_useHP = useHP; }
     bool usesHP() const { return m_useHP; }
+    void setDecimFilters(int sr, uint32_t decim);
     float run(const float& sample);
     float runHP(const float& sample);
     float runLP(const float& sample);
 
 private:
+    void calculate2(bool highPass, double fc, float *a, float *b); // two pole Chebyshev calculation
+    void cheby(bool highPass, double fc, float pr, int np, double *a, double *b);
+    void cheby_sub(bool highPass, double fc, float pr, int np, int stage,
+            double& a0, double& a1, double& a2, double& b1, double& b2);
+
     IIRFilter<float, 2> m_filterLP;
     IIRFilter<float, 2> m_filterHP;
     bool m_useHP;
+    float m_lpva[3];
+    float m_lpvb[3];
+    float m_hpva[3];
+    float m_hpvb[3];
     static const float m_lpa[3];
     static const float m_lpb[3];
     static const float m_hpa[3];
     static const float m_hpb[3];
+
 };
 
-#endif // _SDRBASE_AUDIO_AUDIOFILTER_H_
\ No newline at end of file
+#endif // _SDRBASE_AUDIO_AUDIOFILTER_H_