autotools initial...

1 files changed, 562 insertions, 0 deletions

diff --git a/src/filter.c b/src/filter.c
new file mode 100644
index 0000000..69d7bb6
--- /dev/null
+++ b/src/filter.c
@@ -0,0 +1,562 @@
+/*
+ * filter.c
+ *
+ * Copyright (c) 2018 Disi A
+ * 
+ * Author: Disi A
+ * Email: adis@live.cn
+ *  https://www.mathworks.com/matlabcentral/profile/authors/3734620-disi-a
+ */
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <math.h>
+#include "filter.h"
+#include <stddef.h>
+
+#if DOUBLE_PRECISION
+#define COS cos
+#define SIN sin
+#define TAN tan
+#define COSH cosh
+#define SINH sinh
+#define SQRT sqrt
+#define LOG log
+#else
+#define COS cosf
+#define SIN sinf
+#define TAN tanf
+#define COSH coshf
+#define SINH sinhf
+#define SQRT sqrtf
+#define LOG logf
+#endif
+
+void update_bw_low_pass_filter(BWLowPass* filter, FTR_PRECISION s, FTR_PRECISION f, FTR_PRECISION q)
+{
+    FTR_PRECISION a = TAN((FTR_PRECISION)(M_PI * f / s));
+    FTR_PRECISION a2 = a * a;
+    FTR_PRECISION r;
+    
+    int i;
+    
+    for(i=0; i < filter -> n; ++i){
+      r = SIN((FTR_PRECISION)(M_PI * (2.0 * i + 1.0) / (4.0 * filter->n)));
+      s = (FTR_PRECISION) (a2 + 2.0 * a * r + 1.0);
+      filter->A[i] = a2 / s;
+      filter->d1[i] = (FTR_PRECISION) (2.0 * (1 - a2) / s);
+      filter->d2[i] = (FTR_PRECISION)(-(a2 - 2.0 * a * r + 1.0) / s);
+    }
+}
+
+BWLowPass* create_bw_low_pass_filter(int order, FTR_PRECISION s, FTR_PRECISION f) {
+    BWLowPass* filter = (BWLowPass *) malloc(sizeof(BWLowPass));
+    filter -> n = order/2;
+    filter -> A = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d1 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d2 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> w0 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w1 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w2 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    
+    if (filter->d2 == NULL)
+    {
+        free_bw_low_pass(filter);
+        return NULL;
+    }
+
+    FTR_PRECISION a = TAN((FTR_PRECISION)(M_PI * f / s));
+    FTR_PRECISION a2 = a * a;
+    FTR_PRECISION r;
+    
+    int i;
+    
+    for(i=0; i < filter -> n; ++i){
+        r = SIN((FTR_PRECISION)(M_PI * (2.0 * i + 1.0) / (4.0 * filter->n)));
+        s = (FTR_PRECISION) (a2 + 2.0 * a * r + 1.0);
+        filter->A[i] = a2 / s;
+        filter->d1[i] = (FTR_PRECISION) (2.0 * (1 - a2) / s);
+        filter->d2[i] = (FTR_PRECISION)(-(a2 - 2.0 * a * r + 1.0) / s);
+    }
+    return filter;
+}
+
+BWHighPass* create_bw_high_pass_filter(int order, FTR_PRECISION s, FTR_PRECISION f){
+    BWHighPass* filter = (BWHighPass *) malloc(sizeof(BWHighPass));
+    filter -> n = order/2;
+    filter -> A = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d1 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d2 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> w0 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w1 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w2 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+
+    if (filter->d2 == NULL)
+    {
+        free_bw_high_pass(filter);
+        return NULL;
+    }
+
+    FTR_PRECISION a = TAN((FTR_PRECISION) (M_PI * f/s));
+    FTR_PRECISION a2 = a * a;
+    FTR_PRECISION r;
+    
+    int i;
+
+    for(i=0; i < filter -> n; ++i){
+        r = SIN((FTR_PRECISION)(M_PI * (2.0 * i + 1.0) / ( 4.0 * filter->n)));
+        s = (FTR_PRECISION) (a2 + 2.0 * a * r + 1.0);
+        filter -> A[i] = (FTR_PRECISION) (1.0 / s);
+        filter -> d1[i] = (FTR_PRECISION) (2.0 * (1 - a2) / s);
+        filter -> d2[i] = (FTR_PRECISION) (- (a2 - 2.0 * a * r + 1.0) / s);
+    }
+    return filter;
+}
+
+BWBandPass* create_bw_band_pass_filter(int order, FTR_PRECISION s, FTR_PRECISION fl, FTR_PRECISION fu){
+    if(fu <= fl){
+        printf("ERROR:Lower half-power frequency is smaller than higher half-power frequency");
+        return NULL;
+    }
+    BWBandPass* filter = (BWBandPass *) malloc(sizeof(BWBandPass));
+    filter -> n = order/4;
+    filter -> A = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d1 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d2 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d3 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d4 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+
+    filter -> w0 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w1 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w2 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w3 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w4 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+
+  
+    FTR_PRECISION a = COS(M_PI*(fu+fl)/s)/COS(M_PI*(fu-fl)/s);
+    FTR_PRECISION a2 = a*a;
+    FTR_PRECISION b = TAN(M_PI*(fu-fl)/s);
+    FTR_PRECISION b2 = b*b;
+    FTR_PRECISION r;
+    int i;
+    for(i=0; i<filter->n; ++i){
+        r = SIN( (FTR_PRECISION) (M_PI * (2.0 * i + 1.0) / (4.0 * filter->n)));
+        s = (FTR_PRECISION)(b2 + 2.0 * b * r + 1.0);
+        filter->A[i] = b2/s;
+        filter->d1[i] = (FTR_PRECISION) (4.0 * a * (1.0 + b * r) / s);
+        filter->d2[i] = (FTR_PRECISION) (2.0 * (b2 - 2.0 * a2 - 1.0)/ s);
+        filter->d3[i] = (FTR_PRECISION) (4.0 * a * (1.0 - b * r ) / s);
+        filter->d4[i] = (FTR_PRECISION) (- (b2 - 2.0 * b * r + 1.0) / s);
+    }
+    return filter;
+}
+
+BWBandStop* create_bw_band_stop_filter(int order, FTR_PRECISION s, FTR_PRECISION fl, FTR_PRECISION fu){
+    if(fu <= fl){
+        printf("ERROR:Lower half-power frequency is smaller than higher half-power frequency");
+        return NULL;
+    }
+
+    BWBandStop* filter = (BWBandStop *) malloc(sizeof(BWBandStop));
+    filter -> n = order/4;
+    filter -> A = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d1 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d2 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d3 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+    filter -> d4 = (FTR_PRECISION *)malloc(filter -> n*sizeof(FTR_PRECISION));
+
+    filter -> w0 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w1 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w2 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w3 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+    filter -> w4 = (FTR_PRECISION *)calloc(filter -> n, sizeof(FTR_PRECISION));
+
+    FTR_PRECISION a = COS(M_PI*(fu+fl)/s)/COS(M_PI*(fu-fl)/s);
+    FTR_PRECISION a2 = a*a;
+    FTR_PRECISION b = TAN(M_PI*(fu-fl)/s);
+    FTR_PRECISION b2 = b*b;
+    FTR_PRECISION r;
+
+    int i;
+    for(i=0; i<filter->n; ++i){
+        r = SIN((FTR_PRECISION) (M_PI * (2.0 * i + 1.0 ) / ( 4.0 * filter->n)));
+        s = (FTR_PRECISION) (b2 + 2.0 * b * r + 1.0);
+        filter->A[i] = (FTR_PRECISION) (1.0 / s);
+        filter->d1[i] = (FTR_PRECISION) (4.0 * a * (1.0 + b * r ) / s);
+        filter->d2[i] = (FTR_PRECISION) (2.0 * (b2 - 2.0 * a2- 1.0) / s);
+        filter->d3[i] = (FTR_PRECISION) (4.0 * a * (1.0 - b * r) / s);
+        filter->d4[i] = (FTR_PRECISION) (- (b2 - 2.0 * b * r + 1.0) / s);
+    }
+    filter->r = (FTR_PRECISION) (4.0 * a);
+    filter->s = (FTR_PRECISION) (4.0 * a2 + 2.0);
+    return filter;
+}
+
+CHELowPass* create_che_low_pass_filter(int n, FTR_PRECISION epsilon, FTR_PRECISION s, FTR_PRECISION f){
+    CHELowPass* filter = (CHELowPass *) malloc(sizeof(CHELowPass));
+    filter -> m = n/2;
+    filter -> A = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d1 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d2 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> w0 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w1 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w2 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+
+    if (filter->d2 == NULL)
+    {
+        free_che_low_pass(filter);
+        return NULL;
+    }
+
+    FTR_PRECISION a = TAN((FTR_PRECISION) (M_PI * f/ s));
+    FTR_PRECISION a2 = a * a;
+    
+    FTR_PRECISION u = LOG((FTR_PRECISION) (1.0 + SQRT((FTR_PRECISION) (1.0 + epsilon * epsilon)) / epsilon));
+    FTR_PRECISION su = SINH(u/(FTR_PRECISION)n);
+    FTR_PRECISION cu = COSH(u/(FTR_PRECISION)n);
+    FTR_PRECISION b,c;
+    
+    int i;
+    for(i=0; i<filter->m; ++i){
+        b = SIN((FTR_PRECISION) (M_PI * (2.0 * i + 1.0) / (2.0 * n))) * su;
+        c = COS((FTR_PRECISION) (M_PI * (2.0 * i + 1.0) / (2.0 * n))) * cu;
+        c = b*b + c*c;
+        s = (FTR_PRECISION) (a2 * c + 2.0 * a * b + 1.0);
+        filter->A[i] = (FTR_PRECISION) (a2 / (4.0 * s));
+        filter->d1[i] = (FTR_PRECISION) (2.0 * (1 - a2 * c) / s);
+        filter->d2[i] = (FTR_PRECISION) (- (a2 * c - 2.0 * a * b + 1.0) / s);
+    }
+    filter->ep = (FTR_PRECISION) (2.0 / epsilon);  // used to normalize
+    
+    return filter;
+}
+
+CHEHighPass* create_che_high_pass_filter(int n, FTR_PRECISION epsilon, FTR_PRECISION s, FTR_PRECISION f){
+    CHEHighPass* filter = (CHELowPass *) malloc(sizeof(CHEHighPass));
+    filter -> m = n/2;
+    filter -> A = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d1 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d2 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> w0 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w1 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w2 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+
+    if (filter->d2 == NULL)
+    {
+        free_che_high_pass(filter);
+        return NULL;
+    }
+
+    FTR_PRECISION a = TAN((FTR_PRECISION) (M_PI * f/ s));
+    FTR_PRECISION a2 = a * a;
+
+    FTR_PRECISION u = LOG((FTR_PRECISION) (1.0 + SQRT((FTR_PRECISION) (1.0 + epsilon*epsilon))) / epsilon);
+    FTR_PRECISION su = SINH(u/(FTR_PRECISION)n);
+    FTR_PRECISION cu = COSH(u/(FTR_PRECISION)n);
+    FTR_PRECISION b,c;
+    
+    int i;
+    for(i=0; i<filter->m; ++i){
+        b = SIN((FTR_PRECISION) (M_PI * (2.0 * i + 1.0) / (2.0 * n))) * su;
+        c = COS((FTR_PRECISION) (M_PI * (2.0 * i + 1.0) / (2.0 * n))) * cu;
+        c = b*b + c*c;
+        s = (FTR_PRECISION) (a2 + 2.0 * a * b + c);
+        filter->A[i] = (FTR_PRECISION) (1.0 / (4.0 * s));
+        filter->d1[i] = (FTR_PRECISION) (2.0 * (c - a2)/s);
+        filter->d2[i] = (FTR_PRECISION) (- (a2 - 2.0 * a * b + c) / s);
+    }
+
+    filter->ep = (FTR_PRECISION) (2.0 / epsilon); // used to normalize
+    return filter;
+}
+
+CHEBandPass* create_che_band_pass_filter(int n, FTR_PRECISION epsilon, FTR_PRECISION s, FTR_PRECISION f_lower, FTR_PRECISION f_upper){
+    CHEBandPass* filter = (CHEBandPass *) malloc(sizeof(CHEBandPass));
+    filter -> m = n/4;
+    filter -> A = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d1 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d2 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d3 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d4 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> w0 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w1 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w2 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w3 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w4 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+
+    FTR_PRECISION a = COS(M_PI*(f_lower+f_upper)/s)/COS(M_PI*(f_upper-f_lower)/s);
+    FTR_PRECISION a2 = a*a;
+    FTR_PRECISION b = TAN(M_PI*(f_upper-f_lower)/s);
+    FTR_PRECISION b2 = b*b;
+    FTR_PRECISION u = LOG((FTR_PRECISION) (1.0 + SQRT((FTR_PRECISION) (1.0 + epsilon * epsilon))) / epsilon);
+    FTR_PRECISION su = SINH((FTR_PRECISION) (2.0 * u / (FTR_PRECISION)n));
+    FTR_PRECISION cu = COSH((FTR_PRECISION) (2.0 * u / (FTR_PRECISION)n));
+    FTR_PRECISION r,c;
+
+    int i;
+    for(i=0; i < filter->m; ++i){
+        r = SIN((FTR_PRECISION) (M_PI * (2.0 * i + 1.0) / n)) * su;
+        c = COS((FTR_PRECISION) (M_PI * (2.0 * i + 1.0) / n)) * cu;
+        c = r*r + c*c;
+        s = (FTR_PRECISION) (b2 * c + 2.0 * b * r + 1.0);
+        filter->A[i] = (FTR_PRECISION) (b2 / (4.0 * s));
+        filter->d1[i] = (FTR_PRECISION) (4.0 * a * (1.0 + b * r) / s);
+        filter->d2[i] = (FTR_PRECISION) (2.0 * (b2 * c - 2.0 * a2 - 1.0) / s);
+        filter->d3[i] = (FTR_PRECISION) (4.0 * a * (1.0 - b * r) / s);
+        filter->d4[i] = (FTR_PRECISION) (- (b2 * c - 2.0 * b * r + 1.0) / s);
+    }
+    filter->ep = (FTR_PRECISION) (2.0 / epsilon);  // used to normalize
+    return filter;
+}
+
+CHEBandStop* create_che_band_stop_filter(int n, FTR_PRECISION epsilon, FTR_PRECISION s, FTR_PRECISION f_lower, FTR_PRECISION f_upper){
+    CHEBandStop* filter = (CHEBandStop *) malloc(sizeof(CHEBandStop));
+    filter -> m = n/4;
+    filter -> A = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d1 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d2 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d3 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> d4 = (FTR_PRECISION *)malloc(filter -> m*sizeof(FTR_PRECISION));
+    filter -> w0 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w1 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w2 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w3 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+    filter -> w4 = (FTR_PRECISION *)calloc(filter -> m, sizeof(FTR_PRECISION));
+
+    FTR_PRECISION a = COS(M_PI*(f_lower+f_upper)/s)/COS(M_PI*(f_upper-f_lower)/s);
+    FTR_PRECISION a2 = a*a;
+    FTR_PRECISION b = TAN(M_PI*(f_upper-f_lower)/s);
+    FTR_PRECISION b2 = b*b;
+    FTR_PRECISION u = LOG((FTR_PRECISION) (1.0 + SQRT((FTR_PRECISION) (1.0 + epsilon * epsilon))) / epsilon);
+    FTR_PRECISION su = SINH((FTR_PRECISION) (2.0*u/(FTR_PRECISION)n));
+    FTR_PRECISION cu = COSH((FTR_PRECISION) (2.0*u/(FTR_PRECISION)n));
+    FTR_PRECISION r,c;
+    int i;
+    for(i=0; i < filter->m; ++i){
+        r = SIN((FTR_PRECISION) (M_PI * (2.0 * i + 1.0) / n)) * su;
+        c = COS((FTR_PRECISION) (M_PI * (2.0 * i + 1.0) / n))*cu;
+        c = r*r + c*c;
+        s = (FTR_PRECISION) (b2 + 2.0 * b * r + c);
+        filter->A[i] = (FTR_PRECISION) (1.0 / (4.0 * s));
+        filter->d1[i] = (FTR_PRECISION) (4.0 * a * (c + b * r)/s);
+        filter->d2[i] = (FTR_PRECISION) (2.0 * (b2 - 2.0 * a2 * c - c) / s);
+        filter->d3[i] = (FTR_PRECISION) (4.0 * a * (c - b * r ) / s);
+        filter->d4[i] = (FTR_PRECISION) (- (b2 - 2.0 * b * r + c) / s);
+    }
+    filter->r = (FTR_PRECISION) (4.0 * a);
+    filter->s = (FTR_PRECISION) (4.0 * a2 + 2.0);
+    filter->ep = (FTR_PRECISION) (2.0 / epsilon);  // used to normalize
+    return filter;
+}
+
+void free_bw_low_pass(BWLowPass* filter){
+    free(filter -> A);
+    free(filter -> d1);
+    free(filter -> d2);
+    free(filter -> w0);
+    free(filter -> w1);
+    free(filter -> w2);
+    free(filter);
+}
+void free_bw_high_pass(BWHighPass* filter){
+    free(filter -> A);
+    free(filter -> d1);
+    free(filter -> d2);
+    free(filter -> w0);
+    free(filter -> w1);
+    free(filter -> w2);
+    free(filter);
+}
+void free_bw_band_pass(BWBandPass* filter){
+    free(filter -> A);
+    free(filter -> d1);
+    free(filter -> d2);
+    free(filter -> d3);
+    free(filter -> d4);
+    free(filter -> w0);
+    free(filter -> w1);
+    free(filter -> w2);
+    free(filter -> w3);
+    free(filter -> w4);
+    free(filter);
+}
+void free_bw_band_stop(BWBandStop* filter){
+    free(filter -> A);
+    free(filter -> d1);
+    free(filter -> d2);
+    free(filter -> d3);
+    free(filter -> d4);
+    free(filter -> w0);
+    free(filter -> w1);
+    free(filter -> w2);
+    free(filter -> w3);
+    free(filter -> w4);
+    free(filter);
+}
+
+void free_che_low_pass(CHELowPass* filter){
+    free(filter -> A);
+    free(filter -> d1);
+    free(filter -> d2);
+    free(filter -> w0);
+    free(filter -> w1);
+    free(filter -> w2);
+    free(filter);
+}
+
+void free_che_high_pass(CHEHighPass* filter){
+    free(filter -> A);
+    free(filter -> d1);
+    free(filter -> d2);
+    free(filter -> w0);
+    free(filter -> w1);
+    free(filter -> w2);
+    free(filter);
+}
+void free_che_band_pass(CHEBandPass* filter){
+    free(filter -> A);
+    free(filter -> d1);
+    free(filter -> d2);
+    free(filter -> d3);
+    free(filter -> d4);
+    free(filter -> w0);
+    free(filter -> w1);
+    free(filter -> w2);
+    free(filter -> w3);
+    free(filter -> w4);
+    free(filter);
+}
+void free_che_band_stop(CHEBandStop* filter){
+    free(filter -> A);
+    free(filter -> d1);
+    free(filter -> d2);
+    free(filter -> d3);
+    free(filter -> d4);
+    free(filter -> w0);
+    free(filter -> w1);
+    free(filter -> w2);
+    free(filter -> w3);
+    free(filter -> w4);
+    free(filter);
+}
+
+FTR_PRECISION bw_low_pass(BWLowPass* filter, FTR_PRECISION x){
+    int i;
+    for(i=0; i<filter->n; ++i){
+        filter->w0[i] = filter->d1[i]*filter->w1[i] + filter->d2[i]*filter->w2[i] + x;
+        x = filter->A[i] * (filter->w0[i] + 2.0f * filter->w1[i] + filter->w2[i]);
+        filter->w2[i] = filter->w1[i];
+        filter->w1[i] = filter->w0[i];
+    }
+    return x;
+}
+FTR_PRECISION bw_high_pass(BWHighPass* filter, FTR_PRECISION x){
+    int i;
+    for(i=0; i<filter->n; ++i){
+        filter->w0[i] = filter->d1[i]*filter->w1[i] + filter->d2[i]*filter->w2[i] + x;
+        x = filter->A[i] * (filter->w0[i] - 2.0f * filter->w1[i] + filter->w2[i]);
+        filter->w2[i] = filter->w1[i];
+        filter->w1[i] = filter->w0[i];
+    }
+    return x;
+}
+FTR_PRECISION bw_band_pass(BWBandPass* filter, FTR_PRECISION x){
+    int i;
+    for(i=0; i<filter->n; ++i){
+        filter->w0[i] = filter->d1[i]*filter->w1[i] + filter->d2[i]*filter->w2[i]+ filter->d3[i]*filter->w3[i]+ filter->d4[i]*filter->w4[i] + x;
+        x = filter->A[i] * (filter->w0[i] - 2.0f * filter->w2[i] + filter->w4[i]);
+        filter->w4[i] = filter->w3[i];
+        filter->w3[i] = filter->w2[i];
+        filter->w2[i] = filter->w1[i];
+        filter->w1[i] = filter->w0[i];
+    }
+    return x;
+}
+FTR_PRECISION bw_band_stop(BWBandStop* filter, FTR_PRECISION x){
+    int i;
+    for(i=0; i<filter->n; ++i){
+        filter->w0[i] = filter->d1[i]*filter->w1[i] + filter->d2[i]*filter->w2[i]+ filter->d3[i]*filter->w3[i]+ filter->d4[i]*filter->w4[i] + x;
+        x = filter->A[i]*(filter->w0[i] - filter->r*filter->w1[i] + filter->s*filter->w2[i]- filter->r*filter->w3[i] + filter->w4[i]);
+        filter->w4[i] = filter->w3[i];
+        filter->w3[i] = filter->w2[i];
+        filter->w2[i] = filter->w1[i];
+        filter->w1[i] = filter->w0[i];
+    }
+    return x;
+}
+
+FTR_PRECISION che_low_pass(CHELowPass* filter, FTR_PRECISION x){
+    int i;
+    for(i=0; i<filter->m; ++i){
+        filter->w0[i] = filter->d1[i]*filter->w1[i] + filter->d2[i]*filter->w2[i] + x;
+        x = filter->A[i] * (filter->w0[i] + 2.0f * filter->w1[i] + filter->w2[i]);
+        filter->w2[i] = filter->w1[i];
+        filter->w1[i] = filter->w0[i];
+    }
+    return x * filter->ep;
+}
+FTR_PRECISION che_high_pass(CHEHighPass* filter, FTR_PRECISION x){
+    int i;
+    for(i=0; i<filter->m; ++i){
+        filter->w0[i] = filter->d1[i]*filter->w1[i] + filter->d2[i]*filter->w2[i] + x;
+        x = filter->A[i] * (filter->w0[i] - 2.0f * filter->w1[i] + filter->w2[i]);
+        filter->w2[i] = filter->w1[i];
+        filter->w1[i] = filter->w0[i];
+    }
+    return x * filter->ep;
+}
+
+FTR_PRECISION che_band_pass(CHEBandPass* filter, FTR_PRECISION x){
+    int i;
+    for(i=0; i<filter->m; ++i){
+        filter->w0[i] = filter->d1[i]*filter->w1[i] + filter->d2[i]*filter->w2[i]+ filter->d3[i]*filter->w3[i]+ filter->d4[i]*filter->w4[i] + x;
+        x = filter->A[i] * (filter->w0[i] - 2.0f * filter->w2[i] + filter->w4[i]);
+        filter->w4[i] = filter->w3[i];
+        filter->w3[i] = filter->w2[i];
+        filter->w2[i] = filter->w1[i];
+        filter->w1[i] = filter->w0[i];
+    }
+    return x * filter->ep;
+}
+
+FTR_PRECISION che_band_stop(CHEBandStop* filter, FTR_PRECISION x){
+    int i;
+    for(i=0; i<filter->m; ++i){
+        filter->w0[i] = filter->d1[i]*filter->w1[i] + filter->d2[i]*filter->w2[i]+ filter->d3[i]*filter->w3[i]+ filter->d4[i]*filter->w4[i] + x;
+        x = filter->A[i]*(filter->w0[i] - filter->r*filter->w1[i] + filter->s*filter->w2[i]- filter->r*filter->w3[i] + filter->w4[i]);
+        filter->w4[i] = filter->w3[i];
+        filter->w3[i] = filter->w2[i];
+        filter->w2[i] = filter->w1[i];
+        filter->w1[i] = filter->w0[i];
+    }
+    return x * filter->ep;
+}
+
+FTR_PRECISION softmax(FTR_PRECISION* data, int size, int target_ind){
+    FTR_PRECISION sum = 0;
+    for(int i = 0; i < size; i++) sum += data[i];
+    return data[target_ind]/sum;
+}
+
+static const FTR_PRECISION SPIKE_KERNEL[] = {-1.0, 2.0, -1.0};
+void spike_filter_upward(FTR_PRECISION * input, int size, FTR_PRECISION * output, FTR_PRECISION strength){
+    FTR_PRECISION mean = 0.0;
+    FTR_PRECISION std = 0.0;
+    FTR_PRECISION diff = 0.0;
+    for(int i=0; i < size; i++) mean += input[i];
+    mean /= size;
+
+    for(int i=0; i < size; i++){
+        diff = input[i] - mean;
+        std += diff * diff;
+    }
+    std = SQRT(std/size);
+      
+    output[0] = 0.0;
+    output[size - 1] = 0.0;
+    for(int i=1; i<size-1; i++){
+        FTR_PRECISION val = input[i-1] * SPIKE_KERNEL[0] + input[i] * SPIKE_KERNEL[1] + input[i+1] * SPIKE_KERNEL[2];
+        if(val < strength * std) output[i] = 0.0;
+        else output[i] = val;
+    }
+}