Speex 一个双声道回声消除的小demo

Speex中的回声消除可以对双声道(aec stereo)或多声道的数据进行直接处理。
以下以双声道进行举例:
主要注意事项有三点:
1. 用speex_echo_state_init_mc来进行初始化

echo_state       = speex_echo_state_init_mc(frame_size, filter_length, 2, 2);

2. frame_size的大小
在Speex 双声道回声消除中frame_size的含义中已经详细说明，frame_size就是单声道每次处理的采样点的个数。
以48k，取10ms为例:

    int sampleRate     = 48000;
    int nFrameSizeInMS = 10;
    int frame_size    = (nFrameSizeInMS  * sampleRate  * 1.0) / 1000;

计算结果为480.
3. 每次读取多少数据给speex_echo_cancellation
通过对源码进行分析，如果声道数为N，则应该读取N*frame_size的数据给回声消除的模块。
比如双声道，则每次近端数据和远端数据应该读取2 * frame_size个采样点的数据给回声消除。

双声道回声消除的小demo如下:
https://github.com/ZhaoliangGuo/speex/blob/master/testecho_stereo/testecho_stereo.cpp

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#include "targetver.h"

#include <stdio.h>
#include <tchar.h>
#include "arch.h"

#ifdef __cplusplus  
extern "C" {
#endif

#include "speex/speex_echo.h"
#include "speex/speex_preprocess.h"

#ifdef __cplusplus
}
#endif

#include <windows.h>

int main(int argc, char **argv)
{
    int sampleRate     = 48000;
    int nFrameSizeInMS = 10;
    int nFilterLenInMS = 120;

    SpeexEchoState       *echo_state       = NULL;
    SpeexPreprocessState *preprocess_state = NULL; 

    SYSTEMTIME st;
    ::GetLocalTime(&st);

    int frame_size    = (nFrameSizeInMS  * sampleRate  * 1.0) / 1000;
    int filter_length = (nFilterLenInMS  * sampleRate  * 1.0) / 1000;

    char szMicFileName[]     = ".\\near48_stereo.pcm";
    char szSpeakerFileName[] = ".\\far48_stereo.pcm";

    char szOutputFileName[512];
    sprintf(szOutputFileName, ".\\output_stereo_samplerate_%d_NN_%4d_TAIL_%4d_%04d%02d%02d_%02d%02d%02d.pcm",
        sampleRate,
        frame_size, filter_length,
        st.wYear, st.wMonth, st.wDay, 
        st.wHour, st.wMinute, st.wSecond);
    printf("%s\n", szOutputFileName);

    LARGE_INTEGER timeStartCount;
    LARGE_INTEGER timeEndCount;
    LARGE_INTEGER timeFreq;
    QueryPerformanceFrequency(&timeFreq);
    QueryPerformanceCounter(&timeStartCount);

    FILE *echo_fd,  *ref_fd, *e_fd;

    ref_fd  = fopen(szMicFileName,     "rb");
    echo_fd = fopen(szSpeakerFileName, "rb"); 
    e_fd    = fopen(szOutputFileName,  "wb");

    echo_state       = speex_echo_state_init_mc(frame_size, filter_length, 2, 2); // for stereo
    frame_size *= 2; // length read each time

    preprocess_state = speex_preprocess_state_init(frame_size, sampleRate);

    speex_echo_ctl(echo_state,             SPEEX_ECHO_SET_SAMPLING_RATE,    &sampleRate);
    speex_preprocess_ctl(preprocess_state, SPEEX_PREPROCESS_SET_ECHO_STATE, echo_state);

    short *echo_buf; 
    short *ref_buf;  
    short *e_buf;  

    echo_buf = new short[frame_size];
    ref_buf  = new short[frame_size];
    e_buf    = new short[frame_size];

    while (!feof(ref_fd) && !feof(echo_fd))
    {
        int nLen = fread(ref_buf,  sizeof(short), frame_size, ref_fd);
        if (nLen < 0)
        {
            break;
        }

        nLen = fread(echo_buf, sizeof(short), frame_size, echo_fd);
        if (nLen < 0)
        {
            break;
        }

        speex_echo_cancellation(echo_state, ref_buf, echo_buf, e_buf);
        speex_preprocess_run(preprocess_state, e_buf);

        fwrite(e_buf, sizeof(short), frame_size, e_fd);
    }

    // Destroys an echo canceller state
    speex_echo_state_destroy(echo_state);
    speex_preprocess_state_destroy(preprocess_state);

    fclose(e_fd);
    fclose(echo_fd);
    fclose(ref_fd);

    if (echo_buf)
    {
        delete [] echo_buf;
        echo_buf = NULL;
    }

    if (ref_buf)
    {
        delete [] ref_buf;
        ref_buf = NULL;
    }

    if (e_buf)
    {
        delete [] e_buf;
        e_buf = NULL;
    }

    QueryPerformanceCounter(&timeEndCount);
    double elapsed = (((double)(timeEndCount.QuadPart - timeStartCount.QuadPart) * 1000/ timeFreq.QuadPart));

    printf("AEC Done. TimeDuration: %.2f ms\n", elapsed);

    getchar();

    return 0;
}