概述

一个简单的混音demo，把文件a和文件b的音频混为一个音频流输出并存为文件，只处理每个文件的第一个音频流
注：源代码是网友Larry_Liang(1085803139)写的，我帮其调试通过，存在这里以免备忘

流程介绍

流程如下：

打开输入1->打开输入2->打开输出->初始化混音filter->开启采集线程和编码写文件循环

线程之间数据走向如下：

采集线程1采集到数据->解码->写入fifo1 |
采集线程2采集到数据->解码->写入fifo2 | ->主线程编码循环读取2个fifo->把数据压人filter->从filter中读取混音后数据->编码->写入文件

代码

/* 
*一个简单的混音demo，把文件a和文件b的音频混为一个音频流输出并存为文件，只处理每个文件的第一个音频流 
*源代码是网友Larry_Liang(1085803139)写的，我帮其调试通过 
*MK(821486004@qq.com) 
*/ 
 
 
extern "C" 
{ 
#include "libavcodec/avcodec.h" 
#include "libavformat/avformat.h" 
#include "libavdevice/avdevice.h" 
#include "libavfilter/avfilter.h" 
#include "libavfilter/avfiltergraph.h" 
#include "libavfilter/buffersink.h" 
#include "libavfilter/buffersrc.h" 
#include "libavutil/audio_fifo.h" 
#include "libavutil/avutil.h" 
#include "libavutil/fifo.h" 
} 
 
#pragma comment(lib, "avcodec.lib") 
#pragma comment(lib, "avformat.lib") 
#pragma comment(lib, "avutil.lib") 
#pragma comment(lib, "avdevice.lib") 
#pragma comment(lib, "avfilter.lib") 
 
//#pragma comment(lib, "avfilter.lib") 
//#pragma comment(lib, "postproc.lib") 
//#pragma comment(lib, "swresample.lib") 
#pragma comment(lib, "swscale.lib") 
 
#include <windows.h> 
#include <conio.h> 
#include <time.h> 
 
enum CaptureState 
{ 
    PREPARED, 
    RUNNING, 
    STOPPED, 
    FINISHED 
}; 
 
 typedef struct BufferSourceContext { 
    const AVClass    *bscclass; 
    AVFifoBuffer     *fifo; 
    AVRational        time_base;     ///< time_base to set in the output link 
    AVRational        frame_rate;    ///< frame_rate to set in the output link 
    unsigned          nb_failed_requests; 
    unsigned          warning_limit; 
 
    /* video only */ 
    int               w, h; 
    enum AVPixelFormat  pix_fmt; 
    AVRational        pixel_aspect; 
    char              *sws_param; 
 
    AVBufferRef *hw_frames_ctx; 
 
    /* audio only */ 
    int sample_rate; 
    enum AVSampleFormat sample_fmt; 
    int channels; 
    uint64_t channel_layout; 
    char    *channel_layout_str; 
 
    int got_format_from_params; 
    int eof; 
 } BufferSourceContext; 
 
AVFormatContext* _fmt_ctx_spk = NULL; 
AVFormatContext* _fmt_ctx_mic = NULL; 
AVFormatContext* _fmt_ctx_out = NULL; 
int _index_spk = -1; 
int _index_mic = -1; 
int _index_a_out = -1; 
 
AVFilterGraph* _filter_graph = NULL; 
AVFilterContext* _filter_ctx_src_spk = NULL; 
AVFilterContext* _filter_ctx_src_mic = NULL; 
AVFilterContext* _filter_ctx_sink = NULL; 
 
CaptureState _state = CaptureState::PREPARED; 
 
CRITICAL_SECTION _section_spk; 
CRITICAL_SECTION _section_mic; 
AVAudioFifo* _fifo_spk = NULL; 
AVAudioFifo* _fifo_mic = NULL; 
 
void InitRecorder() 
{ 
    av_register_all(); 
    avdevice_register_all(); 
    avfilter_register_all(); 
} 
 
int OpenSpeakerInput(char* inputForamt, char* url) 
{ 
    AVInputFormat* ifmt = av_find_input_format(inputForamt); 
    AVDictionary* opt1 = NULL; 
    av_dict_set(&opt1, "rtbufsize", "10M", 0); 
 
    int ret = 0; 
    ret = avformat_open_input(&_fmt_ctx_spk, url, ifmt, &opt1); 
    if (ret < 0) 
    { 
        printf("Speaker: failed to call avformat_open_input\n"); 
        return -1; 
    } 
    ret = avformat_find_stream_info(_fmt_ctx_spk, NULL); 
    if (ret < 0) 
    { 
        printf("Speaker: failed to call avformat_find_stream_info\n"); 
        return -1; 
    } 
    for (int i = 0; i < _fmt_ctx_spk->nb_streams; i++) 
    { 
        if (_fmt_ctx_spk->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) 
        { 
            _index_spk = i; 
            break; 
        } 
    } 
    if (_index_spk < 0) 
    { 
        printf("Speaker: negative audio index\n"); 
        return -1; 
    } 
    AVCodecContext* codec_ctx = _fmt_ctx_spk->streams[_index_spk]->codec; 
    AVCodec* codec = avcodec_find_decoder(codec_ctx->codec_id); 
    if (codec == NULL) 
    { 
        printf("Speaker: null audio decoder\n"); 
        return -1; 
    } 
    ret = avcodec_open2(codec_ctx, codec, NULL); 
    if (ret < 0) 
    { 
        printf("Speaker: failed to call avcodec_open2\n"); 
        return -1; 
    } 
    av_dump_format(_fmt_ctx_spk, _index_spk, url, 0); 
 
    return 0; 
} 
 
int OpenMicrophoneInput(char* inputForamt, char* url) 
{ 
    AVInputFormat* ifmt = av_find_input_format(inputForamt); 
    AVDictionary* opt1 = NULL; 
    av_dict_set(&opt1, "rtbufsize", "10M", 0); 
 
    int ret = 0; 
    ret = avformat_open_input(&_fmt_ctx_mic, url, ifmt, &opt1); 
    if (ret < 0) 
    { 
        printf("Microphone: failed to call avformat_open_input\n"); 
        return -1; 
    } 
    ret = avformat_find_stream_info(_fmt_ctx_mic, NULL); 
    if (ret < 0) 
    { 
        printf("Microphone: failed to call avformat_find_stream_info\n"); 
        return -1; 
    } 
    for (int i = 0; i < _fmt_ctx_mic->nb_streams; i++) 
    { 
        if (_fmt_ctx_mic->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) 
        { 
            _index_mic = i; 
            break; 
        } 
    } 
    if (_index_mic < 0) 
    { 
        printf("Microphone: negative audio index\n"); 
        return -1; 
    } 
    AVCodecContext* codec_ctx = _fmt_ctx_mic->streams[_index_mic]->codec; 
    AVCodec* codec = avcodec_find_decoder(codec_ctx->codec_id); 
    if (codec == NULL) 
    { 
        printf("Microphone: null audio decoder\n"); 
        return -1; 
    } 
    ret = avcodec_open2(codec_ctx, codec, NULL); 
    if (ret < 0) 
    { 
        printf("Microphone: failed to call avcodec_open2\n"); 
        return -1; 
    } 
 
    av_dump_format(_fmt_ctx_mic, _index_mic, url, 0); 
 
    return 0; 
} 
 
int OpenFileOutput(char* fileName) 
{ 
    int ret = 0; 
    ret = avformat_alloc_output_context2(&_fmt_ctx_out, NULL, NULL, fileName); 
    if (ret < 0) 
    { 
        printf("Mixer: failed to call avformat_alloc_output_context2\n"); 
        return -1; 
    } 
    AVStream* stream_a = NULL; 
    stream_a = avformat_new_stream(_fmt_ctx_out, NULL); 
    if (stream_a == NULL) 
    { 
        printf("Mixer: failed to call avformat_new_stream\n"); 
        return -1; 
    } 
    _index_a_out = 0; 
 
    stream_a->codec->codec_type = AVMEDIA_TYPE_AUDIO; 
    AVCodec* codec_mp3 = avcodec_find_encoder(AV_CODEC_ID_MP3); 
    stream_a->codec->codec = codec_mp3; 
    stream_a->codec->sample_rate = 44100; 
    stream_a->codec->channels = 2; 
    stream_a->codec->channel_layout = av_get_default_channel_layout(2); 
    stream_a->codec->sample_fmt = codec_mp3->sample_fmts[0]; 
    stream_a->codec->bit_rate = 320000; 
    stream_a->codec->time_base.num = 1; 
    stream_a->codec->time_base.den = stream_a->codec->sample_rate; 
    stream_a->codec->codec_tag = 0; 
 
 
    if (_fmt_ctx_out->oformat->flags & AVFMT_GLOBALHEADER) 
        stream_a->codec->flags |= CODEC_FLAG_GLOBAL_HEADER; 
 
    if (avcodec_open2(stream_a->codec, stream_a->codec->codec, NULL) < 0) 
    { 
        printf("Mixer: failed to call avcodec_open2\n"); 
        return -1;  
    } 
    if (!(_fmt_ctx_out->oformat->flags & AVFMT_NOFILE)) 
    { 
        if (avio_open(&_fmt_ctx_out->pb, fileName, AVIO_FLAG_WRITE) < 0) 
        { 
            printf("Mixer: failed to call avio_open\n"); 
            return -1; 
        } 
    } 
 
    if (avformat_write_header(_fmt_ctx_out, NULL) < 0) 
    { 
        printf("Mixer: failed to call avformat_write_header\n"); 
        return -1; 
    } 
 
    bool b = (!_fmt_ctx_out->streams[0]->time_base.num && _fmt_ctx_out->streams[0]->codec->time_base.num); 
 
    av_dump_format(_fmt_ctx_out, _index_a_out, fileName, 1); 
 
    _fifo_spk = av_audio_fifo_alloc(_fmt_ctx_spk->streams[_index_spk]->codec->sample_fmt, _fmt_ctx_spk->streams[_index_spk]->codec->channels, 30*_fmt_ctx_spk->streams[_index_spk]->codec->frame_size); 
    _fifo_mic = av_audio_fifo_alloc(_fmt_ctx_mic->streams[_index_mic]->codec->sample_fmt, _fmt_ctx_mic->streams[_index_mic]->codec->channels, 30*_fmt_ctx_mic->streams[_index_spk]->codec->frame_size); 
 
    return 0; 
} 
 
int InitFilter(char* filter_desc) 
{ 
    char args_spk[512]; 
    char* pad_name_spk = "in0"; 
    char args_mic[512]; 
    char* pad_name_mic = "in1"; 
 
    AVFilter* filter_src_spk = avfilter_get_by_name("abuffer"); 
    AVFilter* filter_src_mic = avfilter_get_by_name("abuffer"); 
    AVFilter* filter_sink = avfilter_get_by_name("abuffersink"); 
    AVFilterInOut* filter_output_spk = avfilter_inout_alloc(); 
    AVFilterInOut* filter_output_mic = avfilter_inout_alloc(); 
    AVFilterInOut* filter_input = avfilter_inout_alloc(); 
    _filter_graph = avfilter_graph_alloc(); 
 
    sprintf_s(args_spk, sizeof(args_spk), "time_base=%d/%d:sample_rate=%d:sample_fmt=%s:channel_layout=0x%I64x",  
        _fmt_ctx_spk->streams[_index_spk]->codec->time_base.num,  
        _fmt_ctx_spk->streams[_index_spk]->codec->time_base.den,  
        _fmt_ctx_spk->streams[_index_spk]->codec->sample_rate,  
        av_get_sample_fmt_name(_fmt_ctx_spk->streams[_index_spk]->codec->sample_fmt),  
        _fmt_ctx_spk->streams[_index_spk]->codec->channel_layout); 
    sprintf_s(args_mic, sizeof(args_mic), "time_base=%d/%d:sample_rate=%d:sample_fmt=%s:channel_layout=0x%I64x",  
        _fmt_ctx_mic->streams[_index_mic]->codec->time_base.num,  
        _fmt_ctx_mic->streams[_index_mic]->codec->time_base.den,  
        _fmt_ctx_mic->streams[_index_mic]->codec->sample_rate,  
        av_get_sample_fmt_name(_fmt_ctx_mic->streams[_index_mic]->codec->sample_fmt),  
        _fmt_ctx_mic->streams[_index_mic]->codec->channel_layout); 
 
    //sprintf_s(args_spk, sizeof(args_spk), "time_base=%d/%d:sample_rate=%d:sample_fmt=%s:channel_layout=0x%I64x", _fmt_ctx_out->streams[_index_a_out]->codec->time_base.num, _fmt_ctx_out->streams[_index_a_out]->codec->time_base.den, _fmt_ctx_out->streams[_index_a_out]->codec->sample_rate, av_get_sample_fmt_name(_fmt_ctx_out->streams[_index_a_out]->codec->sample_fmt), _fmt_ctx_out->streams[_index_a_out]->codec->channel_layout); 
    //sprintf_s(args_mic, sizeof(args_mic), "time_base=%d/%d:sample_rate=%d:sample_fmt=%s:channel_layout=0x%I64x", _fmt_ctx_out->streams[_index_a_out]->codec->time_base.num, _fmt_ctx_out->streams[_index_a_out]->codec->time_base.den, _fmt_ctx_out->streams[_index_a_out]->codec->sample_rate, av_get_sample_fmt_name(_fmt_ctx_out->streams[_index_a_out]->codec->sample_fmt), _fmt_ctx_out->streams[_index_a_out]->codec->channel_layout); 
 
 
    int ret = 0; 
    ret = avfilter_graph_create_filter(&_filter_ctx_src_spk, filter_src_spk, pad_name_spk, args_spk, NULL, _filter_graph); 
    if (ret < 0) 
    { 
        printf("Filter: failed to call avfilter_graph_create_filter -- src spk\n"); 
        return -1; 
    } 
    ret = avfilter_graph_create_filter(&_filter_ctx_src_mic, filter_src_mic, pad_name_mic, args_mic, NULL, _filter_graph); 
    if (ret < 0) 
    { 
        printf("Filter: failed to call avfilter_graph_create_filter -- src mic\n"); 
        return -1; 
    } 
 
    ret = avfilter_graph_create_filter(&_filter_ctx_sink, filter_sink, "out", NULL, NULL, _filter_graph); 
    if (ret < 0) 
    { 
        printf("Filter: failed to call avfilter_graph_create_filter -- sink\n"); 
        return -1; 
    } 
    AVCodecContext* encodec_ctx = _fmt_ctx_out->streams[_index_a_out]->codec; 
    ret = av_opt_set_bin(_filter_ctx_sink, "sample_fmts", (uint8_t*)&encodec_ctx->sample_fmt, sizeof(encodec_ctx->sample_fmt), AV_OPT_SEARCH_CHILDREN); 
    if (ret < 0) 
    { 
        printf("Filter: failed to call av_opt_set_bin -- sample_fmts\n"); 
        return -1; 
    } 
    ret = av_opt_set_bin(_filter_ctx_sink, "channel_layouts", (uint8_t*)&encodec_ctx->channel_layout, sizeof(encodec_ctx->channel_layout), AV_OPT_SEARCH_CHILDREN); 
    if (ret < 0) 
    { 
        printf("Filter: failed to call av_opt_set_bin -- channel_layouts\n"); 
        return -1; 
    } 
    ret = av_opt_set_bin(_filter_ctx_sink, "sample_rates", (uint8_t*)&encodec_ctx->sample_rate, sizeof(encodec_ctx->sample_rate), AV_OPT_SEARCH_CHILDREN); 
    if (ret < 0) 
    { 
        printf("Filter: failed to call av_opt_set_bin -- sample_rates\n"); 
        return -1; 
    } 
 
    filter_output_spk->name = av_strdup(pad_name_spk); 
    filter_output_spk->filter_ctx = _filter_ctx_src_spk; 
    filter_output_spk->pad_idx = 0; 
    filter_output_spk->next = filter_output_mic; 
 
    filter_output_mic->name = av_strdup(pad_name_mic); 
    filter_output_mic->filter_ctx = _filter_ctx_src_mic; 
    filter_output_mic->pad_idx = 0; 
    filter_output_mic->next = NULL; 
 
    filter_input->name = av_strdup("out"); 
    filter_input->filter_ctx = _filter_ctx_sink; 
    filter_input->pad_idx = 0; 
    filter_input->next = NULL; 
 
    AVFilterInOut* filter_outputs[2]; 
    filter_outputs[0] = filter_output_spk; 
    filter_outputs[1] = filter_output_mic; 
 
    ret = avfilter_graph_parse_ptr(_filter_graph, filter_desc, &filter_input, filter_outputs, NULL); 
    if (ret < 0) 
    { 
        printf("Filter: failed to call avfilter_graph_parse_ptr\n"); 
        return -1; 
    } 
 
    ret = avfilter_graph_config(_filter_graph, NULL); 
    if (ret < 0) 
    { 
        printf("Filter: failed to call avfilter_graph_config\n"); 
        return -1; 
    } 
 
    avfilter_inout_free(&filter_input); 
    av_free(filter_src_spk); 
    av_free(filter_src_mic); 
    avfilter_inout_free(filter_outputs); 
    //av_free(filter_outputs); 
 
    char* temp = avfilter_graph_dump(_filter_graph, NULL); 
    printf("%s\n", temp); 
 
    return 0; 
} 
 
DWORD WINAPI SpeakerCapThreadProc(LPVOID lpParam) 
{ 
    AVFrame* pFrame = av_frame_alloc(); 
    AVPacket packet; 
    av_init_packet(&packet); 
 
    int got_sound; 
 
    while (_state == CaptureState::RUNNING) 
    { 
        packet.data = NULL; 
        packet.size = 0; 
 
        if (av_read_frame(_fmt_ctx_spk, &packet) < 0) 
        { 
            continue; 
        } 
        if (packet.stream_index == _index_spk) 
        { 
            if (avcodec_decode_audio4(_fmt_ctx_spk->streams[_index_spk]->codec, pFrame, &got_sound, &packet) < 0) 
            { 
                break; 
            } 
            av_free_packet(&packet); 
 
            if (!got_sound) 
            { 
                continue; 
            } 
 
            int fifo_spk_space = av_audio_fifo_space(_fifo_spk); 
            while(fifo_spk_space < pFrame->nb_samples && _state == CaptureState::RUNNING) 
            { 
                Sleep(10); 
                printf("_fifo_spk full !\n"); 
                fifo_spk_space = av_audio_fifo_space(_fifo_spk); 
            } 
 
            if (fifo_spk_space >= pFrame->nb_samples) 
            { 
                EnterCriticalSection(&_section_spk); 
                int nWritten = av_audio_fifo_write(_fifo_spk, (void**)pFrame->data, pFrame->nb_samples); 
                LeaveCriticalSection(&_section_spk); 
            } 
        } 
    } 
    av_frame_free(&pFrame); 
 
    return 0; 
} 
 
DWORD WINAPI MicrophoneCapThreadProc(LPVOID lpParam) 
{ 
    AVFrame* pFrame = av_frame_alloc(); 
    AVPacket packet; 
    av_init_packet(&packet); 
 
    int got_sound; 
 
    while (_state == CaptureState::PREPARED) 
    { 
 
    } 
 
    while (_state == CaptureState::RUNNING) 
    { 
        packet.data = NULL; 
        packet.size = 0; 
 
        if (av_read_frame(_fmt_ctx_mic, &packet) < 0) 
        { 
            continue; 
        } 
        if (packet.stream_index == _index_mic) 
        { 
            if (avcodec_decode_audio4(_fmt_ctx_mic->streams[_index_mic]->codec, pFrame, &got_sound, &packet) < 0) 
            { 
                break; 
            } 
            av_free_packet(&packet); 
 
            if (!got_sound) 
            { 
                continue; 
            } 
 
            int fifo_mic_space = av_audio_fifo_space(_fifo_mic); 
            while(fifo_mic_space < pFrame->nb_samples && _state == CaptureState::RUNNING) 
            { 
                Sleep(10); 
                printf("_fifo_mic full !\n"); 
                fifo_mic_space = av_audio_fifo_space(_fifo_mic); 
            } 
 
            if (fifo_mic_space >= pFrame->nb_samples) 
            { 
                EnterCriticalSection(&_section_mic); 
                int temp = av_audio_fifo_space(_fifo_mic); 
                int temp2 = pFrame->nb_samples; 
                int nWritten = av_audio_fifo_write(_fifo_mic, (void**)pFrame->data, pFrame->nb_samples); 
                LeaveCriticalSection(&_section_mic); 
            } 
        } 
    } 
    av_frame_free(&pFrame); 
 
    return 0; 
} 
 
int main() 
{ 
    int ret = 0; 
 
    InitRecorder(); 
 
    char fileName[128]; 
    char* outFileType = ".mp3"; 
 
    time_t rawtime; 
    tm* timeInfo; 
    time(&rawtime); 
    timeInfo = localtime(&rawtime); 
    sprintf_s(fileName, sizeof(fileName), "%d_%d_%d_%d_%d_%d%s", 
        timeInfo->tm_year + 1900, timeInfo->tm_mon + 1, timeInfo->tm_mday, 
        timeInfo->tm_hour, timeInfo->tm_min, timeInfo->tm_sec, outFileType); 
 
    char* filter_desc = "[in0][in1]amix=inputs=2[out]"; 
 
    //ret = OpenSpeakerInput("dshow", "audio=virtual-audio-capturer"); 
    ret = OpenSpeakerInput(NULL, "故乡.mp3"); 
    if (ret < 0) 
    { 
        goto Release; 
    } 
    //ret = OpenMicrophoneInput("dshow", "audio=External Microphone (Conexant S"); 
    ret = OpenMicrophoneInput(NULL, "旅行.mp3"); 
    if (ret < 0) 
    { 
        goto Release; 
    } 
    ret = OpenFileOutput(fileName); 
    if (ret < 0) 
    { 
        goto Release; 
    } 
    ret = InitFilter(filter_desc); 
    if (ret < 0) 
    { 
        goto Release; 
    } 
 
    _state = CaptureState::RUNNING; 
 
    InitializeCriticalSection(&_section_spk); 
    InitializeCriticalSection(&_section_mic); 
 
    CreateThread(NULL, 0, SpeakerCapThreadProc, 0, 0, NULL); 
    CreateThread(NULL, 0, MicrophoneCapThreadProc, 0, 0, NULL); 
 
    int tmpFifoFailed = 0; 
    int64_t frame_count = 0; 
 
    while (_state != CaptureState::FINISHED) 
    { 
        if (_kbhit()) 
        { 
            _state = CaptureState::STOPPED; 
            break; 
        } 
        else 
        { 
            int ret = 0; 
            AVFrame* pFrame_spk = av_frame_alloc(); 
            AVFrame* pFrame_mic = av_frame_alloc(); 
 
 
            AVPacket packet_out; 
 
            int got_packet_ptr = 0; 
 
            int fifo_spk_size = av_audio_fifo_size(_fifo_spk); 
            int fifo_mic_size = av_audio_fifo_size(_fifo_mic); 
            int frame_spk_min_size = _fmt_ctx_spk->streams[_index_spk]->codec->frame_size; 
            int frame_mic_min_size = _fmt_ctx_mic->streams[_index_mic]->codec->frame_size; 
            if (fifo_spk_size >= frame_spk_min_size && fifo_mic_size >= frame_mic_min_size) 
            { 
                tmpFifoFailed = 0; 
 
                pFrame_spk->nb_samples = frame_spk_min_size; 
                pFrame_spk->channel_layout = _fmt_ctx_spk->streams[_index_spk]->codec->channel_layout; 
                pFrame_spk->format = _fmt_ctx_spk->streams[_index_spk]->codec->sample_fmt; 
                pFrame_spk->sample_rate = _fmt_ctx_spk->streams[_index_spk]->codec->sample_rate; 
                av_frame_get_buffer(pFrame_spk, 0); 
 
                pFrame_mic->nb_samples = frame_mic_min_size; 
                pFrame_mic->channel_layout = _fmt_ctx_mic->streams[_index_mic]->codec->channel_layout; 
                pFrame_mic->format = _fmt_ctx_mic->streams[_index_mic]->codec->sample_fmt; 
                pFrame_mic->sample_rate = _fmt_ctx_mic->streams[_index_mic]->codec->sample_rate; 
                av_frame_get_buffer(pFrame_mic, 0); 
 
                EnterCriticalSection(&_section_spk); 
                ret = av_audio_fifo_read(_fifo_spk, (void**)pFrame_spk->data, frame_spk_min_size); 
                LeaveCriticalSection(&_section_spk); 
 
                EnterCriticalSection(&_section_mic); 
                ret = av_audio_fifo_read(_fifo_mic, (void**)pFrame_mic->data, frame_mic_min_size); 
                LeaveCriticalSection(&_section_mic); 
 
                pFrame_spk->pts = av_frame_get_best_effort_timestamp(pFrame_spk); 
                pFrame_mic->pts = av_frame_get_best_effort_timestamp(pFrame_mic); 
 
                BufferSourceContext* s = (BufferSourceContext*)_filter_ctx_src_spk->priv; 
                bool b1 = (s->sample_fmt != pFrame_spk->format); 
                bool b2 = (s->sample_rate != pFrame_spk->sample_rate); 
                bool b3 = (s->channel_layout != pFrame_spk->channel_layout); 
                bool b4 = (s->channels != pFrame_spk->channels); 
 
                ret = av_buffersrc_add_frame(_filter_ctx_src_spk, pFrame_spk); 
                if (ret < 0) 
                { 
                    printf("Mixer: failed to call av_buffersrc_add_frame (speaker)\n"); 
                    break; 
                } 
 
                ret = av_buffersrc_add_frame(_filter_ctx_src_mic, pFrame_mic); 
                if (ret < 0) 
                { 
                    printf("Mixer: failed to call av_buffersrc_add_frame (microphone)\n"); 
                    break; 
                } 
 
                while (1) 
                { 
                    AVFrame* pFrame_out = av_frame_alloc(); 
 
                    ret = av_buffersink_get_frame_flags(_filter_ctx_sink, pFrame_out, 0); 
                    if (ret < 0) 
                    { 
                        printf("Mixer: failed to call av_buffersink_get_frame_flags\n"); 
                        break; 
                    } 
                    if (pFrame_out->data[0] != NULL) 
                    { 
                        av_init_packet(&packet_out); 
                        packet_out.data = NULL; 
                        packet_out.size = 0; 
 
                        ret = avcodec_encode_audio2(_fmt_ctx_out->streams[_index_a_out]->codec, &packet_out, pFrame_out, &got_packet_ptr); 
                        if (ret < 0) 
                        { 
                            printf("Mixer: failed to call avcodec_decode_audio4\n"); 
                            break; 
                        } 
                        if (got_packet_ptr) 
                        { 
                            packet_out.stream_index = _index_a_out; 
                            packet_out.pts = frame_count * _fmt_ctx_out->streams[_index_a_out]->codec->frame_size; 
                            packet_out.dts = packet_out.pts; 
                            packet_out.duration = _fmt_ctx_out->streams[_index_a_out]->codec->frame_size; 
 
                            packet_out.pts = av_rescale_q_rnd(packet_out.pts,  
                                _fmt_ctx_out->streams[_index_a_out]->codec->time_base, 
                                _fmt_ctx_out->streams[_index_a_out]->time_base, 
                                (AVRounding)(AV_ROUND_NEAR_INF | AV_ROUND_PASS_MINMAX)); 
                            packet_out.dts = packet_out.pts; 
                            packet_out.duration = av_rescale_q_rnd(packet_out.duration, 
                                _fmt_ctx_out->streams[_index_a_out]->codec->time_base, 
                                _fmt_ctx_out->streams[_index_a_out]->time_base, 
                                (AVRounding)(AV_ROUND_NEAR_INF | AV_ROUND_PASS_MINMAX)); 
 
                            frame_count++; 
 
                            ret = av_interleaved_write_frame(_fmt_ctx_out, &packet_out); 
                            if (ret < 0) 
                            { 
                                printf("Mixer: failed to call av_interleaved_write_frame\n"); 
                            } 
                            printf("Mixer: write frame to file\n"); 
                        } 
                        av_free_packet(&packet_out);                     
                    } 
                    av_frame_free(&pFrame_out); 
                } 
            } 
            else 
            { 
                tmpFifoFailed++; 
                Sleep(20); 
                if (tmpFifoFailed > 300) 
                { 
                    _state = CaptureState::STOPPED; 
                    Sleep(30); 
                    break; 
                } 
            } 
            av_frame_free(&pFrame_spk); 
            av_frame_free(&pFrame_mic); 
        } 
    } 
 
    av_write_trailer(_fmt_ctx_out); 
 
Release: 
    av_audio_fifo_free(_fifo_spk); 
    av_audio_fifo_free(_fifo_mic); 
 
    avfilter_free(_filter_ctx_src_spk); 
    avfilter_free(_filter_ctx_src_mic); 
    avfilter_free(_filter_ctx_sink); 
 
    avfilter_graph_free(&_filter_graph); 
 
 
    if (_fmt_ctx_out) 
    { 
        avio_close(_fmt_ctx_out->pb); 
    } 
 
    avformat_close_input(&_fmt_ctx_spk); 
    avformat_close_input(&_fmt_ctx_mic); 
    avformat_free_context(_fmt_ctx_out); 
 
    return ret; 
}