IIR数字滤波器实现（mantlab+C语言）

水深无声 2022-01-27 03:21 223阅读 0赞

题目：16k采样率音频数据下采样到8k采样率

求解方案分析：直接每隔一个取一个采样值，这样就可以得到8k采样率的数据。但是这样明显会有问题。按照采样率变换理论，首先应该通过一个低通滤波器，滤掉\[pi/2, pi\]这个区间上的频率，以防止下采样造成的频率混叠。这个低通滤波器在很多书上都用FIR滤波去实现，并且可以用FIR滤波的多相结构去实现。这样滤波和下采样过程可以互换位置。即先下采样再进行多相FIR滤波。

在嵌入式设备上FIR滤波会占用较长的时间，为此，我们可以采用IIR滤波器来做。滤波器的设计采用MATLAB的FDATool工具。相关的参数如下：

![0_1291360311r38r.gif][]

![0_1291360317xfF9.gif][]

采用最小P阶范数设计IIR滤波器的幅频特性较好。滤波器阶数设定为6阶，需要滤除4k到8k的频率段，才能保证无混叠失真。实际由于滤波器的特性没法做到理想的状态，选择滤波器截止频率为3800hz，3600到3800为过渡带宽。其它选项采用默认设置。设计的滤波器幅度响应如下图：

![0_1291360322mAQ3.gif][]

生成的滤波器系数文件如下：

/\*

\* Filter Coefficients (C Source) generated by the Filter Design and Analysis Tool

\* Generated by MATLAB(R) 7.6 and the Signal Processing Toolbox 6.9.

\* Generated on: 03-Dec-2010 10:41:03

\*/

/\*

\* Discrete-Time IIR Filter (real)

\* -------------------------------

\* Filter Structure : Direct-Form II, Second-Order Sections

\* Number of Sections : 3

\* Stable : Yes

\* Linear Phase : No

\*/

/\* General type conversion for MATLAB generated C-code \*/

\#include "tmwtypes.h"

/\*

\* Expected path to tmwtypes.h

\* D:/MATLAB/R2008a/extern/include/tmwtypes.h

\*/

\#define MWSPT\_NSEC 7

const int NL\[MWSPT\_NSEC\] = \{ 1,3,1,3,1,3,1 \};

const real64\_T NUM\[MWSPT\_NSEC\]\[3\] = \{

0.09065504059673, 0, 0

\},

1, -0.1323122149853, 0.9999674089086

\},

1, 0, 0

\},

1, 0.1670351201308, 0.9999889247428

\},

1, 0, 0

\},

1, 1.417032671609, 0.9978019623105

\},

1, 0, 0

\};

const int DL\[MWSPT\_NSEC\] = \{ 1,3,1,3,1,3,1 \};

const real64\_T DEN\[MWSPT\_NSEC\]\[3\] = \{

1, 0, 0

\},

1, -0.40321085647, 0.7334254033056

\},

1, 0, 0

\},

1, -0.6868636040216, 0.2670185171768

\},

1, 0, 0

\},

1, -0.2880720042256, 0.9480010462991

\},

1, 0, 0

\};

上述系数是3个2阶节IIR结构的级联。可以转换为我们熟悉的b/a的形式如下：

double a\[3\]\[3\] = \{

\{1, -0.6868636040216, 0.2670185171768\},

\{1, -0.40321085647, 0.7334254033056\},

\{1, -0.2880720042256, 0.9480010462991\}

\};

double b\[3\]\[3\] = \{

\{1, 1.417032671609, 0.9978019623105\},

\{1, 0.1670351201308, 0.9999889247428\},

\{1, -0.1323122149853, 0.9999674089086\},

\};

注意上面系数文件中还有一个增益：

double gain = 0.09065504059673；

这个增益最好在第一级实现以后加入运算。这样可以减小误差，保证数据 动态范围不被溢出。尤其是在定点计算的时候尤为如此。

# 2阶节IIR滤波的直接实现 #

#  #

一个2阶节结构是下面这样一个表达式：

![0_1291360325Fv03.gif][]

实现上面这个表达式需要4个过去的历史值，把它定义在结构体

typedef struct tag\_ IIR\_State\_2order

float y2;

float y1;

float x1;

float x0;

\} IIR\_State\_2order;

调用下面函数之前需要把上述结构体所有值初始化为零。滤波按一帧一帧数据进行。

\#define ONE\_FRAME\_SAMPLE\_SIZE 1024

void cy\_signal\_filter\_by\_iir(signed short\* pcmIn, IIR\_State\_2order\* filter\_state, float a\[\], float b\[\], signed short\*pcmOut)

int i;

float x2;

float tmp;

for ( i = 0; i < ONE\_FRAME\_SAMPLE\_SIZE; i++ )

x2 = filter\_state->x1;

filter\_state->x1 = filter\_state->x0;

filter\_state->x0 = pcmIn\[i\];

tmp = ( float )(b\[0\]\* filter\_state->x0 + b\[1\]\* filter\_state->x1 +

b\[2\] \* x2 - a\[1\] \* filter\_state->y1 - a\[2\] \* filter\_state->y2);

if(tmp >= 32767)

tmp = 32767;

if(tmp <= -32768)

tmp = -32768;

pcmOut\[i\] = (signed short)tmp;

filter\_state->y2 = filter\_state->y1;

filter\_state->y1 = tmp;

有一个简单的技巧可以把上面的计算简化，使得历史状态数由4减少为2。定义下面的表达式：

![0_1291360330QJVI.gif][]

结构体定义如下：

typedef struct tag\_ IIR\_State\_2order

float st1;

float st2;

\} IIR\_State\_2order;

void cy\_signal\_filter\_by\_iir(signed short\* pcmIn, IIR\_State\_2order\* filter\_state, float a\[\], float b\[\], signed short\*pcmOut)

int i;

float st;

float Tmp\_fl;

for ( i = 0; i < ONE\_FRAME\_SAMPLE\_SIZE; i++ )

st = (float)(pcmIn\[i\] - a\[1\] \* filter\_state->st1 - a\[2\] \* filter\_state-> st2);

Tmp\_fl = b\[0\] \* st + b\[1\] \* filter\_state->st1 + b\[2\] \* filter\_state->st2;

filter\_state->st2= filter\_state->st1;

filter\_state->st1 = st;

if(Tmp\_fl >= 32767.0)

Tmp\_fl = 32767;

if(Tmp\_fl <= -32768)

Tmp\_fl = -32768;

pcmOut\[i\] = (signed short)Tmp\_fl;

# 6阶节IIR滤波的实现 #

#  #

有个上面的基础，我们来实现上面设计的6阶IIR滤波器。

6阶节分解为3个2阶节级联实现。每个2阶节需要2个历史状态，总共需要6个历史状态。结构体定义如下：

typedef struct tag\_IIR\_State\_3Order

double w01;

double w02;

double w11;

double w12;

double w21;

double w22;

\}IIR\_State\_6order;

代码中数组a,b,还有gain的定义见第一部分。

void cy\_signal\_filter\_by\_6th\_iir(signed short\* pcmIn, IIR\_State\_6order\* filter\_state, int sample\_size)

double x1, x2, x3, Tmp\_f00, Tmp\_f10, Tmp\_f20;

int i;

double Tmp\_pcm;

for (i = 0; i < sample\_size; i++)

Tmp\_pcm = pcmIn\[i\];

Tmp\_f00 = Tmp\_pcm - a\[0\]\[1\] \* filter\_state->w01 - a\[0\]\[2\] \* filter\_state->w02;

x1 = Tmp\_f00 + b\[0\]\[1\] \* filter\_state->w01 + b\[0\]\[2\] \* filter\_state->w02;

filter\_state->w02 = filter\_state->w01;

filter\_state->w01 = Tmp\_f00;

x1 = gain \* x1;

Tmp\_f10 = x1 - a\[1\]\[1\] \* filter\_state->w11 - a\[1\]\[2\] \* filter\_state->w12;

x2 = Tmp\_f10 + b\[1\]\[1\] \* filter\_state->w11 + b\[1\]\[2\] \* filter\_state->w12;

filter\_state->w12 = filter\_state->w11;

filter\_state->w11 = Tmp\_f10;

Tmp\_f20 = x2 - a\[2\]\[1\] \* filter\_state->w21 - a\[2\]\[2\] \* filter\_state->w22;

x3 = Tmp\_f20 \* b\[2\]\[0\] + b\[2\]\[1\] \* filter\_state->w21 + b\[2\]\[2\] \* filter\_state->w22;

filter\_state->w22 = filter\_state->w21;

filter\_state->w21 = Tmp\_f20;

if (x3 >= 32767)

x3 = 32767;

if (x3 <= -32768)

x3 = -32768;

pcmIn\[i\] = (signed short)x3;

最后看下滤波的效果：

![0_12913619466ZKe.gif][]

滤波之后的频谱：

![0_1291361984tpkk.gif][]

滤波效果不错，下面可以进行下采样了。

[0_1291360311r38r.gif]: /images/20220127/5735986950cd439fbe7aa5a761c08ceb.png
[0_1291360317xfF9.gif]: /images/20220127/c5de7c9609eb4dcbb34b60b16f2a8ac8.png
[0_1291360322mAQ3.gif]: /images/20220127/1877cc0c7efd48f09c6761feefd3e830.png
[0_1291360325Fv03.gif]: /images/20220127/7ade2ede3cf4431f8cea2695175ff3a6.png
[0_1291360330QJVI.gif]: /images/20220127/0bc9d6e44ae14859bf121f5de068b1fa.png
[0_12913619466ZKe.gif]: /images/20220127/5c5f56c9ffd44db0a97d791863b66d96.png
[0_1291361984tpkk.gif]: /images/20220127/6d03cb1f07c94571834ff44b33142e25.png