C格式的FIR低通滤波器

Question

我正在用C语言实现FIR低通滤波器。音频来自 Zynq Zedboard，并且是实时的。如果我运行该应用程序，我会得到很多噪音。但我不知道信号是如何失真的。我是否需要音频缓冲区来保存处理后的样本？还是我只有蹩脚的代码？我分析了代码中的几乎所有元素，但无法理解它......

这是函数调用

#define FILTER_LEN_LP 44
int32_t coeffsLPF[ FILTER_LEN_LP ] =
{
    87, 76, 106, 143, 185, 234, 289, 349,
    414, 483, 555, 628, 701, 773, 842, 907,
    966, 1017, 1060, 1093, 1115, 1126, 1126,
    1115, 1093, 1060, 1017, 966, 907, 842, 773,
    701, 628, 555, 483, 414, 349, 289, 234, 185,
    143, 106, 76, 87
};
void low_pass_filter(){
    int k = 0;
    int filter_length = FILTER_LEN_LP;
    int32_t *coeffp; // pointer to coefficients
    int32_t in_left;
    while (!XUartPs_IsReceiveData(UART_BASEADDR)) {
        int32_t out_left;
        int32_t acc; // accumulator
        in_left = Xil_In32(I2S_DATA_RX_L_REG);
        //printf("%drn", in_left);
        coeffp = coeffsLPF;
        acc = 1 << 14 ;
        for ( k = 0; k < filter_length; k++ )
        {
             acc += (int32_t)(*coeffp++) * (int32_t)(in_left);
        }
        //printf("%drn", acc);
        //printf("rn");
        // saturate the result
        // 32768^2
        if ( acc > 1073741823)
        {
             acc = 1073741823;
        } else if ( acc < -1073741824 ){
            acc = -1073741824;
        }
       out_left = (int32_t) (acc >> 13);
       //xil_printf("%dnr", out_left);
       //in_right = Xil_In32(I2S_DATA_RX_R_REG);
       Xil_Out32(I2S_DATA_TX_L_REG, out_left);
       //Xil_Out32(I2S_DATA_TX_R_REG, in_right);
    }
// break
if(XUartPs_ReadReg(UART_BASEADDR, XUARTPS_FIFO_OFFSET) == 'q') menu();
else low_pass_filter();
}

Answer 1

FIR 滤波器的公式为：

y[n] = b0*x[n] + b1*x[

n-1] + b2*x[n-2]...

要计算一个输出，您需要最后 44 个输入样本，并将其与滤波器系数相乘和累加。我的建议是使用大小为 44 的环形缓冲区来存储输入样本。分配一个环形缓冲区并将其初始化为零。

ring_buffer[FILTER_LEN_LP] = {0};
rb_index = 0;

，然后使用以下示例代码存储输入示例并计算输出。

ring_buffer[rb_idx] = Xil_In32(I2S_DATA_RX_L_REG);
rb_idx = (rb_idx + 1) % filter_length;
for (k=0; k<filter_length; k++) {
    acc += (int32_t)(*coeffp++) * (int32_t)(ring_buffer[(k+rb_idx)%filter_length]);
}