android 音频采集、FLTP重采样与AAC编码推流

相比较视频编码，音频编码要简单很多，主要就是将采集到的音频源数据PCM编码AAC. MediaPlus中FFmpeg使用的是libfdk-aac编码器，这里有个问题需要注意下:FFmpeg已经废弃了AV_SAMPLE_FMT_S16格式PCM编码AAC，也就是说如果使用FFmpeg自带的AAC编码器，必须做音频的重采样（重采样为:AV_SAMPLE_FMT_FLTP），否则AAC编码是失败的。

接下来，看下MediaPlus中是如何采集音频与AAC编码的.

在app.mobile.nativeapp.com.libmedia.core.streamer.RtmpPushStreamer的AudioThread获取AudioRecord采集到的音频数据并传入底层：

  class AudioThread extends Thread {

        public volatile boolean m_bExit = false;

        @Override
        public void run() {
            // TODO Auto-generated method stub
            super.run();

            int[] dataLength;
            byte[] audioBuffer;
            AudioCaptureInterface.GetAudioDataReturn ret;

            dataLength = new int[1];
            audioBuffer = new byte[m_aiBufferLength[0]];


            while (!m_bExit) {
                try {
                    Thread.sleep(1, 10);
                    if (m_bExit) {
                        break;
                    }
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
                try {
                    ret = mAudioCapture.GetAudioData(audioBuffer,
                            m_aiBufferLength[0], dataLength);
                    if (ret == AudioCaptureInterface.GetAudioDataReturn.RET_SUCCESS) {
                        encodeAudio(audioBuffer, dataLength[0]);

                    }

                } catch (Exception e) {
                    e.printStackTrace();
                    stopThread();
                }
            }
        }

具体AudioRecord采集具体实现是avcapture.jar包中，代码比较简单，相关android音视频采集初始化及API调用网上都有相关Demo，这里不再赘述！

• encodeAudio(audioBuffer, dataLength[0]);将音频数据传入底层。

    /**
    * 采集的PCM音频数据
    *
    * @param audioBuffer
    * @param length
    */
   public void encodeAudio(byte[] audioBuffer, int length) {
       try {
           LiveJniMediaManager.EncodeAAC(audioBuffer, length);
       } catch (Exception e) {
           e.printStackTrace();
       }
   }

• JNI层接收到PCM音频数据，添加到AudioCapture同步队列中：

JNIEXPORT jint JNICALL
Java_app_mobile_nativeapp_com_libmedia_core_jni_LiveJniMediaManager_EncodeAAC(JNIEnv *env,
                                                                             jclass type,
                                                                             jbyteArray audioBuffer_,
                                                                             jint length) {
   if (audioCaptureInit && !isClose) {
       jbyte *audioSrc = env->GetByteArrayElements(audioBuffer_, 0);
       uint8_t *audioDstData = (uint8_t *) malloc(length);
       memcpy(audioDstData, audioSrc, length);
       OriginData *audioOriginData = new OriginData();
       audioOriginData->size = length;
       audioOriginData->data = audioDstData;
       audioCapture->PushAudioData(audioOriginData);
       env->ReleaseByteArrayElements(audioBuffer_, audioSrc, 0);
   }
   return 0;
}

以上代码，是在调用app.mobile.nativeapp.com.libmedia.core.streamer.RtmpPushStreamer>>startPushStream()开启推流前的相关调用：主要就是初始化音频采集，并将数据传入底层。

• startPushStream的调用，会重置AudioCapture::ExitCapture=false; 数据才会被加入到audioCaputureframeQueue对列中.

  /**
    * 开启推流
    * @param pushUrl
    * @return
    */
   private boolean startPushStream(String pushUrl) {
       if (nativeInt) {
           int ret = 0;
           ret = LiveJniMediaManager.StartPush(pushUrl);
           if (ret < 0) {
               Log.d("initNative", "native push failed!");
               return false;
           }
           return true;
       }
       return false;
   }

如下图：

• 重置标记后，audioCaputureframeQueue.push将数据添中到队列中.

int AudioCapture::PushAudioData(OriginData *originData) {
    if (ExitCapture) {
        return 0;
    }
    originData->pts = av_gettime();
    LOG_D(DEBUG,"audio capture pts :%lld",originData->pts);
    audioCaputureframeQueue.push(originData);
    return 0;
}

上面这些代码与视频的处理方式都是一样的流程，在调用app.mobile.nativeapp.com.libmedia.core.streamer.RtmpPushStreamer>>startPushStream()，已经开始往音频队列中添加数据，紧接着调用rtmpStreamer->StartPushStream() ，实际也就是开启了音视频的两个编码线程及推流,推流相关代码与视频一致.

int RtmpStreamer::StartPushStream() {
    videoStreamIndex = AddStream(videoEncoder->videoCodecContext);
    audioStreamIndex = AddStream(audioEncoder->audioCodecContext);
    pthread_create(&t3, NULL, RtmpStreamer::WriteHead, this);
    pthread_join(t3, NULL);

    VideoCapture *pVideoCapture = videoEncoder->GetVideoCapture();
    AudioCapture *pAudioCapture = audioEncoder->GetAudioCapture();
    pVideoCapture->videoCaputureframeQueue.clear();
    pAudioCapture->audioCaputureframeQueue.clear();

    if(writeHeadFinish) {
        pthread_create(&t1, NULL, RtmpStreamer::PushAudioStreamTask, this);
        pthread_create(&t2, NULL, RtmpStreamer::PushVideoStreamTask, this);
    }else{
        return -1;
    }
    return 0;
}

• PushAudioStreamTask中从队列中获取数据编码、推流.

rtmpStreamer->audioEncoder->EncodeAAC(&pAudioData);AAC编码.

rtmpStreamer->SendFrame(pAudioData, rtmpStreamer->audioStreamIndex);推流(与视频推流一致)

这里说明下，音频编码前获取编码器及一些参数的指定： libmedia/src/main/cpp/AudioEncoder.cpp是音频编码的核心类，int AudioEncoder::InitEncode() 方法封装了音频编码器的初始化。

int AudioEncoder::InitEncode() {
    std::lock_guard<std::mutex> lk(mut);
    avCodec = avcodec_find_encoder_by_name("libfdk_aac");
    int ret = 0;
    if (!avCodec) {
        LOG_D(DEBUG, "aac encoder not found!")
        return -1;
    }
    audioCodecContext = avcodec_alloc_context3(avCodec);
    if (!audioCodecContext) {
        LOG_D(DEBUG, "avcodec alloc context3 failed!");
        return -1;
    }
    audioCodecContext->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;
    audioCodecContext->sample_fmt = AV_SAMPLE_FMT_S16;
    audioCodecContext->sample_rate = audioCapture->GetAudioEncodeArgs()->sampleRate;
    audioCodecContext->thread_count = 8;
    audioCodecContext->bit_rate = 50*1024*8;
    audioCodecContext->channels = audioCapture->GetAudioEncodeArgs()->channels;
    audioCodecContext->frame_size = audioCapture->GetAudioEncodeArgs()->nb_samples;
    audioCodecContext->time_base = {1, 1000000};//AUDIO VIDEO 两边时间基数要相同
    audioCodecContext->channel_layout = av_get_default_channel_layout(audioCodecContext->channels);


    outputFrame = av_frame_alloc();
    outputFrame->channels = audioCodecContext->channels;
    outputFrame->channel_layout = av_get_default_channel_layout(outputFrame->channels);
    outputFrame->format = audioCodecContext->sample_fmt;
    outputFrame->nb_samples = 1024;
    ret = av_frame_get_buffer(outputFrame, 0);
    if (ret != 0) {
        LOG_D(DEBUG, "av_frame_get_buffer failed!");
        return -1;
    }
    LOG_D(DEBUG, "av_frame_get_buffer success!");

    ret = avcodec_open2(audioCodecContext, NULL, NULL);
    if (ret != 0) {
        char buf[1024] = {0};
        av_strerror(ret, buf, sizeof(buf));
        LOG_D(DEBUG, "avcodec open failed! info:%s", buf);
        return -1;
    }
    LOG_D(DEBUG, "open audio codec success!");
    LOG_D(DEBUG, "Complete init Audio Encode!")
    return 0;
}

• 指定获取libfdk_aac编码器

avCodec = avcodec_find_encoder_by_name("libfdk_aac");

• 初始化编码器上下文

audioCodecContext = avcodec_alloc_context3(avCodec);

• 创建AVFrame，并分配内存负责封装PCM源数据

outputFrame = av_frame_alloc();
    outputFrame->channels = audioCodecContext->channels;//通道数
    outputFrame->channel_layout = av_get_default_channel_layout(outputFrame->channels);
    outputFrame->format = audioCodecContext->sample_fmt;
    outputFrame->nb_samples = 1024;//默认值
    ret = av_frame_get_buffer(outputFrame, 0);
    if (ret != 0) {
        LOG_D(DEBUG, "av_frame_get_buffer failed!");
        return -1;
    }
    LOG_D(DEBUG, "av_frame_get_buffer success!");

• 打开编码器

ret = avcodec_open2(audioCodecContext, NULL, NULL);

以上是编码前必须要完成的初始化.

int AudioEncoder::EncodeAAC 方法封装了AAC编码：

int AudioEncoder::EncodeAAC(OriginData **originData) {

    int ret = 0;
    ret = avcodec_fill_audio_frame(outputFrame,
                                   audioCodecContext->channels,
                                   audioCodecContext->sample_fmt, (*originData)->data,
                                   8192, 0);
    outputFrame->pts = (*originData)->pts;
    ret = avcodec_send_frame(audioCodecContext, outputFrame);
    if (ret != 0) {
#ifdef SHOW_DEBUG_INFO
        LOG_D(DEBUG, "send frame failed!");
#endif
    }
    av_packet_unref(&audioPacket);

    ret = avcodec_receive_packet(audioCodecContext, &audioPacket);
    if (ret != 0) {
#ifdef SHOW_DEBUG_INFO
        LOG_D(DEBUG, "receive packet failed!");
#endif
    }
    (*originData)->Drop();
    (*originData)->avPacket = &audioPacket;

#ifdef SHOW_DEBUG_INFO
    LOG_D(DEBUG, "encode audio packet size:%d pts:%lld", (*originData)->avPacket->size,
          (*originData)->avPacket->pts);
    LOG_D(DEBUG, "Audio frame encode success!");
#endif
    (*originData)->avPacket->size;
    return audioPacket.size;
}

• *originData->data填充到AVFrame中，

ret = avcodec_send_frame(audioCodecContext, outputFrame);
ret = avcodec_receive_packet(audioCodecContext, &audioPacket);

audioPacket就是编码后的数据了，data是编码后的数据，size是大小，这样就完成了编码.

注意：在int AudioEncoder::InitEncode()方法中

avcodec_find_encoder_by_name("libfdk_aac");

这里使用了fdk-aac编码器，前提是你必须要将libfdk-aac库，链接到ffmpeg动态库中，否则是找不到此编码器的。FFmpeg自带有AAC编码器，可以通过：

 avcodec_find_encoder(AV_CODEC_ID_AAC);

获取到AAC编码器,当然如果使用FFmpeg的AAC编码器，就会涉及到一个问题，就是刚开始文中提到了，AV_SAMPLE_FMT_S16需要重采样为：AV_SAMPLE_FMT_FLTP的问题，由于FFmpeg废弃了AV_SAMPLE_FMT_S16格式PCM编码AAC，那么在编码前就需要多一步重采样的处理.

以下AV_SAMPLE_FMT_S16 PCM音频数据重采样相关代码仅供参考：

• 初始化SwrContext,指定输入输出参数

    swrContext = swr_alloc_set_opts(swrContext, av_get_default_channel_layout(CHANNELS),//输出通道Layout
                                    AV_SAMPLE_FMT_FLTP,//输出格式
                                    48000,//输出采样率
                                    av_get_default_channel_layout(CHANNELS),//输入通道Layout
                                    AV_SAMPLE_FMT_S16,//输入格式
                                    48000,//输入采样率
                                    NULL,//NULL
                                    NULL);//NULL
      ret = swr_init(swrContext);//初始化SwrContext

    if (ret != 0) {
        LOG_D(DEBUG, "swr_init failed!");
        return -1;
    }

• AAC编码前，将源数据重采样

   for (; ;) {
        if (encodeAAC->exit) {
            break;
        }
        if (encodeAAC->frame_queue.empty()) {
            continue;
        }
        const uint8_t *indata[AV_NUM_DATA_POINTERS] = {0};
        //PCM s16
        uint8_t *buf = *encodeAAC->frame_queue.wait_and_pop().get();//PCM 16bit
#ifdef FDK_CODEC
        //fdk-aac无需重采样
        ret = avcodec_fill_audio_frame(encodeAAC->outputFrame, encodeAAC->avCodecContext->channels,
                                       encodeAAC->avCodecContext->sample_fmt, buf, BUFFER_SIZE, 0);
        if (ret < 0) {
            LOG_D(DEBUG, "fill frame failed!");
            continue;
        }
#else
        //重采样AM_SAMPLE_FMT_FLTP
        indata[0] = buf;
        swr_convert(encodeAAC->swrContext, encodeAAC->outputFrame->data,
                    encodeAAC->outputFrame->nb_samples, indata,
                    encodeAAC->outputFrame->nb_samples);
#endif

以上代码就可以实现音频重采样，这样就可以再使用FFMPEG AAC编码器完成编码.

以上简述了android 采集音频PCM数据及AAC编码、AAC编码涉及的相关初始化、FFmpeg AAC编码器的重采样示例.android camera采集、H264编码与Rtmp推流与本文描述了音视频采集、编码过程及如何完成推流，相关文章待续......

版权声明：本文为原创文章，转载请注明出处。 代码地址：github.com/javandcpp/M…