用alsa录音并ffmpeg推流RTMP

Ubuntu上用alsa录音并ffmpeg推音频流RTMP到nginx服务器

1.环境需求

ffmpeg和alsa

csdn教程一大把。

2.实现方法

①通过Alsa框架进行录音，获取pcm数据；

②通过FFmpeg框架，把pcm数据重采样，编码进行推流。

3.关键代码介绍
①通过Alsa框架进行录音，获取pcm数据；

//ALSA头文件
#include 
//双声道
#define CHANNELS 2
//每个采样点2bytes
#define FSIZE 2*CHANNELS

int main()
{
    //要保存的pcm文件
    int fd;
    char *file=out_filename;
    fd = open(file,O_WRONLY|O_CREAT,0777);

    //pcm设备结构器
    snd_pcm_t *handle;
    //以录音模式打开设备
    snd_pcm_open(&handle, "default",SND_PCM_STREAM_CAPTURE, 0);
    
    //pcm参数结构体
    snd_pcm_hw_params_t *params;
    //params申请内存
    snd_pcm_hw_params_malloc(&params);
    //使用pcm设备初始化params
    snd_pcm_hw_params_any(handle, params);
    
    //设置多路数据在buffer中的存储方式
	//SND_PCM_ACCESS_RW_INTERLEAVED每个周期(period)左右声道的数据交叉存放
	snd_pcm_hw_params_set_access(handle, params,SND_PCM_ACCESS_RW_INTERLEAVED);
	//设置16位采样格式,S16代表有符号16位,LE代表小端
	snd_pcm_hw_params_set_format(handle, params,SND_PCM_FORMAT_S16_LE);
	//设置声道数
	snd_pcm_hw_params_set_channels(handle, params, CHANNELS);
	//采样率
	unsigned int val=48000;
	int dir;
	//设置采样率,如果采样率不支持，会用硬件支持最接近的采样率
	snd_pcm_hw_params_set_rate_near(handle, params,&val, &dir);
	unsigned int buffer_time,period_time;
	//获取最大的缓冲时间,buffer_time单位为us,500000us=0.5s
	snd_pcm_hw_params_get_buffer_time_max(params, &buffer_time, 0);
	printf("max_buffer_time:%d\n",buffer_time);
	if ( buffer_time >500000)
		buffer_time = 500000;
		
	//设置缓冲时间
	snd_pcm_hw_params_set_buffer_time_near(handle, params, &buffer_time, 0);
	
	//设置采样周期时间,计算方法38帧/秒，48000/38=1263点/帧
	period_time = 26315;
	snd_pcm_hw_params_set_period_time_near(handle, params, &period_time, 0);
    
    //让这些参数设置到PCM设备
    snd_pcm_hw_params(handle, params);
    
    //这个frames并不是指帧率，而是1263采样点数/帧
    snd_pcm_uframes_t frames;
	snd_pcm_hw_params_get_period_size(params,&frames, &dir);

    //缓冲区大小
    /* 2 bytes/sample, 1 channels */
    int size;
    size = frames * FSIZE; 
    printf("size:%d\n",size);
    //数据缓冲区
    char *buffer;
	buffer = (char *) malloc(size);
    
    while (1) 
	{
	    //开始录音，收集一帧数据到缓冲区
	    snd_pcm_readi(handle, buffer, frames);
        if (ret == -EPIPE) {
		// EPIPE means overrun 
			fprintf(stderr, "overrun occurred\n");
			ret=snd_pcm_prepare(handle);
			if(ret <0){
				printf("Failed to recover form overrun");
				exit(1);
			}
		}
		else if (ret < 0) {
			fprintf(stderr,"error from read: %s\n",snd_strerror(ret));
			exit(1);
		} 
		else if (ret != (int)frames) {
			fprintf(stderr, "short read, read %d frames\n", ret);
		
		}
        //保存录音
        write(fd, buffer, size);
	    	
	}
	close(fd);
	snd_pcm_drain(handle);
	snd_pcm_close(handle);
	free(buffer);
}    

②通过FFmpeg框架，把pcm数据重采样，编码进行推流。

#include 
#include 
#include 
#include 
#include 
#include 

int main()
{
    //pframePCM用于保存每帧pcm数据
    AVFrame *pframePCM;
    //AVFrame初始化
	pframePCM = av_frame_alloc();
    
	pframePCM->format = AV_SAMPLE_FMT_S16;             //S16格式
    pframePCM->channel_layout = AV_CH_LAYOUT_STEREO;   //双声道
    pframePCM->sample_rate = 48000;                    //采样率
    pframePCM->nb_samples = frames;                    //采样点/每帧
    pframePCM->channels = CHANNELS;                    
    //AVFrame应用设置
    av_frame_get_buffer(pframePCM, 0);
    
    //pframeAAC用于保存重采样数据
    AVFrame *pframeAAC;
    //AVFrame初始化
	pframeAAC = av_frame_alloc();
    
	pframeAAC->format = AV_SAMPLE_FMT_FLTP;            //fltp格式
    pframeAAC->channel_layout = AV_CH_LAYOUT_STEREO;   //双声道
    pframeAAC->sample_rate = 44100;                    //采样率
    pframeAAC->nb_samples = 1024;                      //采样点/每帧
    pframeAAC->channels = CHANNELS;
    //AVFrame应用设置
    av_frame_get_buffer(pframeAAC, 0);

    //音频格式转换上下文
    struct SwrContext *aac_convert_ctx  = swr_alloc();
    swr_alloc_set_opts(aac_convert_ctx, 
	                   AV_CH_LAYOUT_STEREO,     //dst目标
	                   AV_SAMPLE_FMT_FLTP,
	                   44100, 
	                   AV_CH_LAYOUT_STEREO,     //src原始
	                   AV_SAMPLE_FMT_S16, 
	                   48000, 
	                   0, 
	                   NULL);
    //编码器
    AVCodec* encodec = NULL;
    //设置AAC编码器
	encodec = avcodec_find_encoder(AV_CODEC_ID_AAC);

    AVCodecContext* encodec_ctx = NULL;
    //创建编码器上下文
	encodec_ctx = avcodec_alloc_context3(encodec);

	//设置编码器上下文参数
    encodec_ctx->codec_id = encodec->id;
    encodec_ctx->codec_type = AVMEDIA_TYPE_AUDIO;
    encodec_ctx->sample_fmt  = AV_SAMPLE_FMT_FLTP;
    encodec_ctx->bit_rate    = 64000;
    encodec_ctx->sample_rate = 44100;
    encodec_ctx->channel_layout = AV_CH_LAYOUT_STEREO ;
    encodec_ctx->channels = av_get_channel_layout_nb_channels(AV_CH_LAYOUT_STEREO);
    
    //打开编码器上下文
    avcodec_open2(encodec_ctx, encodec, NULL);
    
    //RTMP地址
    char *out_name="rtmp://127.0.0.1/live/stream";
    //视频格式上下文
    AVFormatContext* outfmt_ctx = NULL;
    //创建输出封装器
    avformat_alloc_output_context2(&outfmt_ctx, NULL, "flv",out_name);
    
    //视频流
    AVStream *out_stream = NULL;
    //给AVFormatContext添加AVStream
    out_stream = avformat_new_stream(outfmt_ctx,NULL);
    
    //复制参数
	avcodec_parameters_from_context(out_stream->codecpar, encodec_ctx);

    //查看输出封装内容
	av_dump_format(outfmt_ctx, 0, out_name, 1);

    //打开rtmp的网络输出IO
    avio_open(&outfmt_ctx->pb, out_name, AVIO_FLAG_WRITE);

	//写入封装头
    avformat_write_header(outfmt_ctx, NULL);
    
    int got_picture;
    int vpts=0;
    
    //视频包裹
    AVPacket enc_pkt;
    //给视频包裹申请内存
    memset(&enc_pkt, 0, sizeof(enc_pkt));

    while(1)
    {
         //alsa获取数据
         snd_pcm_readi(handle, buffer, frames);
         //复制到pframePCM中
         memcpy(pframePCM->data[0],buffer,size);
         //格式转换
         swr_convert(aac_convert_ctx,
                     pframeAAC->data,                       //dst     
                     pframeAAC->nb_samples,
                     (const uint8_t **)pframePCM->data,     //src
                     pframePCM->nb_samples);
                     
         //开始编码到视频包裹
         avcodec_encode_audio2(encodec_ctx, &enc_pkt, pframeAAC, &got_picture);
         //设置时间戳
		 enc_pkt.pts = av_rescale_q(enc_pkt.pts, encodec_ctx->time_base, out_stream->time_base);
         enc_pkt.dts = av_rescale_q(enc_pkt.dts, encodec_ctx->time_base, out_stream->time_base);
         enc_pkt.duration = av_rescale_q(enc_pkt.duration, encodec_ctx->time_base, out_stream->time_base);
         //发送RMTP流
         av_interleaved_write_frame(outfmt_ctx, &enc_pkt);
         //清空包裹
         av_free_packet(&enc_pkt);
         
         pframeAAC->pts = vpts;
		 vpts+=pframeAAC->nb_samples;
    }
}

3.完整代码
完整代码的注释没那么详细，复制黏贴后请从第二部分关键代码的介绍。

#include 
#include 
#include 
#include 
#include 
#include 
#include 
#include 
#include 
#define CHANNELS 2
#define FSIZE 2*CHANNELS

int volume_adjust(char *in_buf,float vol)
{
    short buf=0;
    buf=*in_buf+(*(in_buf+1)<<8);
    
    if(buf>=-1&&buf<=1)
    {
        buf=0;
    }
    
    buf=buf*vol;
    
    
    if(buf>=32767)
    {
        buf=0;
        *in_buf=(char)buf;
        *(in_buf+1)=buf>>8;
    }
    else if(buf<=-32768)
    {
        buf=0;
        *in_buf=(char)buf;
        *(in_buf+1)=buf>>8;
    }
    else
    {
        *in_buf=(char)buf;
        *(in_buf+1)=buf>>8;
    }
    return 0;
}


int main()
{
    int fd;
    
    char *out_filename="output.raw";
    char *file=out_filename;
    fd = open(file,O_WRONLY|O_CREAT,0777);
	if( fd ==-1)
	{
		printf("open file:%s fail.\n",out_filename);
		exit(1);
	}

    int ret=0;
    
    snd_pcm_t *handle;
    //以录音模式打开设备
    ret = snd_pcm_open(&handle, "default",SND_PCM_STREAM_CAPTURE, 0);
	if (ret < 0) 
	{
		printf("unable to open pcm device!\n");
		exit(1);
	}
	
	//配置硬件参数结构体
    snd_pcm_hw_params_t *params;
    //params申请内存
    snd_pcm_hw_params_malloc(&params);
    //使用pcm设备初始化hwparams
    ret=snd_pcm_hw_params_any(handle, params);
	if (ret < 0) 
	{
		printf("Can not configure this PCM device!\n");
		exit(1);
	}
	
	//设置多路数据在buffer中的存储方式
	//SND_PCM_ACCESS_RW_INTERLEAVED每个周期(period)左右声道的数据交叉存放
	ret=snd_pcm_hw_params_set_access(handle, params,SND_PCM_ACCESS_RW_INTERLEAVED);
	if (ret < 0) 
	{
		printf("Failed to set PCM device to interleaved!\n");
		exit(1);
	}
	
	//设置16位采样格式
	ret=snd_pcm_hw_params_set_format(handle, params,SND_PCM_FORMAT_S16_LE);
	if (ret < 0) 
	{
        printf("Failed to set PCM device to 16-bit signed PCM\n");
		exit(1);
	}
	
	//设置声道数
	ret=snd_pcm_hw_params_set_channels(handle, params, CHANNELS);
	if (ret < 0) 
	{
        printf("Failed to set PCM device CHANNELS\n");
		exit(1);
	}
	
	unsigned int val=48000;
	int dir;
	//设置采样率,如果采样率不支持，会用硬件支持最接近的采样率
	ret=snd_pcm_hw_params_set_rate_near(handle, params,&val, &dir);
	if (ret < 0) 
	{
		printf("Failed to set PCM device to sample rate\n");
		exit(1);
	}
	
	unsigned int buffer_time,period_time;
	//获取最大的缓冲时间,buffer_time单位为us,500000us=0.5s
	snd_pcm_hw_params_get_buffer_time_max(params, &buffer_time, 0);
	//printf("buffer_time:%d\n",buffer_time);
	if ( buffer_time >500000)
		buffer_time = 500000;
	
	//设置缓冲时间
	ret = snd_pcm_hw_params_set_buffer_time_near(handle, params, &buffer_time, 0);
	if (ret < 0) 
	{
		printf("Failed to set PCM device to sample rate\n");
		exit(1);
	}
	//设置周期时间
	period_time = 26315;
	ret = snd_pcm_hw_params_set_period_time_near(handle, params, &period_time, 0);
	if (ret < 0) 
	{
		printf("Failed to set PCM device to period time\n");
		exit(1);
	}
	
	//让这些参数作用于PCM设备
	ret = snd_pcm_hw_params(handle, params);
	if (ret < 0) 
	{
		printf("unable to set hw parameters\n");
		exit(1);
	}
	
	snd_pcm_uframes_t frames;
	snd_pcm_hw_params_get_period_size(params,&frames, &dir);
	printf("period_size:%ld\n",frames);
	int size;
	// 1 frame = channels * sample_size.
	size = frames * FSIZE; /* 2 bytes/sample, 1 channels */
	printf("size:%d\n",size);
	char *buffer;
	buffer = (char *) malloc(size);
	
	AVFrame *pframePCM;
	pframePCM = av_frame_alloc();

	pframePCM->format = AV_SAMPLE_FMT_S16;
    pframePCM->channel_layout = AV_CH_LAYOUT_STEREO;
    pframePCM->sample_rate = 48000;
    pframePCM->nb_samples = frames;
    pframePCM->channels = CHANNELS;
    av_frame_get_buffer(pframePCM, 0);
    
    AVFrame *pframeAAC;
	pframeAAC = av_frame_alloc();

	pframeAAC->format = AV_SAMPLE_FMT_FLTP;
    pframeAAC->channel_layout = AV_CH_LAYOUT_STEREO;
    pframeAAC->sample_rate = 44100;
    pframeAAC->nb_samples = 1024;
    pframeAAC->channels = CHANNELS;
    av_frame_get_buffer(pframeAAC, 0);

    struct SwrContext *aac_convert_ctx  = swr_alloc();
    if (!aac_convert_ctx) 
    {
        fprintf(stderr, "Could not allocate resampler context\n");
        return -1;
    }
    
	swr_alloc_set_opts(aac_convert_ctx, 
	                   AV_CH_LAYOUT_STEREO, 
	                   AV_SAMPLE_FMT_FLTP,
	                   44100, 
	                   AV_CH_LAYOUT_STEREO, 
	                   AV_SAMPLE_FMT_S16, 
	                   48000, 
	                   0, 
	                   NULL);
    
	
	if ((ret = swr_init(aac_convert_ctx)) < 0) 
	{
        fprintf(stderr, "Failed to initialize the resampling context\n");
        return -1;
    }
    
    
    AVCodec* encodec = NULL;
    //找到编码器
	encodec = avcodec_find_encoder(AV_CODEC_ID_AAC);
	if (!encodec) 
	{
		printf("not find encoder\n");
		return -1;
	}
	
    AVCodecContext* encodec_ctx = NULL;
    //创建编码器
	encodec_ctx = avcodec_alloc_context3(encodec);
	if (!encodec_ctx) 
	{
		printf("not alloc context3\n\n");
		return -1;
	}
	
    encodec_ctx->codec_id = encodec->id;
    encodec_ctx->codec_type = AVMEDIA_TYPE_AUDIO;
    encodec_ctx->sample_fmt  = AV_SAMPLE_FMT_FLTP;
    encodec_ctx->bit_rate    = 64000;
    encodec_ctx->sample_rate = 44100;
    encodec_ctx->channel_layout = AV_CH_LAYOUT_STEREO ;
    encodec_ctx->channels = av_get_channel_layout_nb_channels(AV_CH_LAYOUT_STEREO);
    
    //打开解码器
	ret = avcodec_open2(encodec_ctx, encodec, NULL);
    if (ret < 0) {
        fprintf(stderr, "Could not open codec: %s\n", av_err2str(ret));
        return -1;
    }
    char *out_name="rtmp://47.101.62.167/live/stream";
    AVFormatContext* outfmt_ctx = NULL;
    //创建输出封装器
    ret=avformat_alloc_output_context2(&outfmt_ctx, NULL, "flv",out_name);
    if (ret != 0) 
    {
        printf("failed alloc output context\n");
        return -1;;
    }
    AVStream *out_stream = NULL;
    //添加视频流
    out_stream = avformat_new_stream(outfmt_ctx,NULL);
	if (!out_stream) {
		printf("failed new stream\n");
		return -1;
	}
	
	//复制参数
	avcodec_parameters_from_context(out_stream->codecpar, encodec_ctx);
	
	//查看输出封装内容
	av_dump_format(outfmt_ctx, 0, out_name, 1);
	
	//打开rtmp的网络输出IO
	ret=avio_open(&outfmt_ctx->pb, out_name, AVIO_FLAG_WRITE);
	if (ret!=0) {
		printf("failed to open outfile\n");
		return -1;
	}

	//写入封装头
	ret=avformat_write_header(outfmt_ctx, NULL);
	if (ret!=0) {
		printf("failed to write header\n");
		avio_close(outfmt_ctx->pb);
		return -1;
	}
	
    AVPacket enc_pkt;
    memset(&enc_pkt, 0, sizeof(enc_pkt));
    
    int got_picture;
    int i,vpts=0;
    char *p;
    struct timeval start, end;
    gettimeofday( &start, NULL );
	while (1) 
	{
		ret = snd_pcm_readi(handle, buffer, frames);

		if (ret == -EPIPE) {
		// EPIPE means overrun 
			fprintf(stderr, "overrun occurred\n");
			ret=snd_pcm_prepare(handle);
			if(ret <0){
				printf("Failed to recover form overrun");
				exit(1);
			}
		}
		else if (ret < 0) {
			fprintf(stderr,"error from read: %s\n",snd_strerror(ret));
			exit(1);
		} 
		else if (ret != (int)frames) {
			fprintf(stderr, "short read, read %d frames\n", ret);
		
		}
		
		
        memcpy(pframePCM->data[0],buffer,size);
        
        ret=swr_convert(aac_convert_ctx,pframeAAC->data, pframeAAC->nb_samples,(const uint8_t **)pframePCM->data, pframePCM->nb_samples);
        
        avcodec_encode_audio2(encodec_ctx, &enc_pkt, pframeAAC, &got_picture);
		if(!got_picture)
        {
            printf("123\n");
            continue;
        }

        
		//推流
		enc_pkt.pts = av_rescale_q(enc_pkt.pts, encodec_ctx->time_base, out_stream->time_base);
        enc_pkt.dts = av_rescale_q(enc_pkt.dts, encodec_ctx->time_base, out_stream->time_base);
        enc_pkt.duration = av_rescale_q(enc_pkt.duration, encodec_ctx->time_base, out_stream->time_base);
        
        ret = av_interleaved_write_frame(outfmt_ctx, &enc_pkt);
		if (ret < 0) {
            fprintf(stderr, "Error muxing packet\n");
            break;
        }	
        av_free_packet(&enc_pkt);
        
        pframeAAC->pts = vpts;
		vpts+=pframeAAC->nb_samples;
                
		ret = write(fd, buffer, size);
		if (ret <0){
			perror("fail to write to audio file\n");
		}

        gettimeofday( &end, NULL );
        printf("%ld",end.tv_sec-start.tv_sec);
        printf("\r\033[k");
        fflush(stdout); 
	}
    
	close(fd);
	snd_pcm_drain(handle);
	snd_pcm_close(handle);
	free(buffer);
	

    return 0;
}

不知不觉就搞定了FFmpeg框架下实现的视频编解码和音频编解码，过几天我把他们整合到一起，成为完整的音视频RTMP流。
还是要感谢雷大佬的帮助。