亲宝软件园·资讯

展开

Android WebRTC 对 AudioRecord 的使用

anyRTC 人气:0

前言:

AudioRecord 是 Android 基于原始PCM音频数据录制的类,WebRCT 对其封装的代码位置位于org/webrtc/audio/WebRtcAudioRecord.java,接下来我们学习一下 AudioRecord 是如何创建启动,读取音频采集数据以及销毁等功能的。

一、创建和初始化

 private int initRecording(int sampleRate, int channels) {
    Logging.d(TAG, "initRecording(sampleRate=" + sampleRate + ", channels=" + channels + ")");
    if (audioRecord != null) {
      reportWebRtcAudioRecordInitError("InitRecording called twice without StopRecording.");
      return -1;
    }
    final int bytesPerFrame = channels * (BITS_PER_SAMPLE / 8);
    final int framesPerBuffer = sampleRate / BUFFERS_PER_SECOND;
    byteBuffer = ByteBuffer.allocateDirect(bytesPerFrame * framesPerBuffer);
    Logging.d(TAG, "byteBuffer.capacity: " + byteBuffer.capacity());
    emptyBytes = new byte[byteBuffer.capacity()];
    // Rather than passing the ByteBuffer with every callback (requiring
    // the potentially expensive GetDirectBufferAddress) we simply have the
    // the native class cache the address to the memory once.
    nativeCacheDirectBufferAddress(byteBuffer, nativeAudioRecord);

    // Get the minimum buffer size required for the successful creation of
    // an AudioRecord object, in byte units.
    // Note that this size doesn't guarantee a smooth recording under load.
    final int channelConfig = channelCountToConfiguration(channels);
    int minBufferSize =
        AudioRecord.getMinBufferSize(sampleRate, channelConfig, AudioFormat.ENCODING_PCM_16BIT);
    if (minBufferSize == AudioRecord.ERROR || minBufferSize == AudioRecord.ERROR_BAD_VALUE) {
      reportWebRtcAudioRecordInitError("AudioRecord.getMinBufferSize failed: " + minBufferSize);
      return -1;
    }
    Logging.d(TAG, "AudioRecord.getMinBufferSize: " + minBufferSize);

    // Use a larger buffer size than the minimum required when creating the
    // AudioRecord instance to ensure smooth recording under load. It has been
    // verified that it does not increase the actual recording latency.
    int bufferSizeInBytes = Math.max(BUFFER_SIZE_FACTOR * minBufferSize, byteBuffer.capacity());
    Logging.d(TAG, "bufferSizeInBytes: " + bufferSizeInBytes);
    try {
      audioRecord = new AudioRecord(audioSource, sampleRate, channelConfig,
          AudioFormat.ENCODING_PCM_16BIT, bufferSizeInBytes);
    } catch (IllegalArgumentException e) {
      reportWebRtcAudioRecordInitError("AudioRecord ctor error: " + e.getMessage());
      releaseAudioResources();
      return -1;
    }
    if (audioRecord == null || audioRecord.getState() != AudioRecord.STATE_INITIALIZED) {
      reportWebRtcAudioRecordInitError("Failed to create a new AudioRecord instance");
      releaseAudioResources();
      return -1;
    }
    if (effects != null) {
      effects.enable(audioRecord.getAudioSessionId());
    }
    logMainParameters();
    logMainParametersExtended();
    return framesPerBuffer;
  }

在初始化的方法中,主要做了两件事。

创建缓冲区:

创建 AudioRecord对象,构造函数有很多参数,分析如下:

二、启动

private boolean startRecording() {
    Logging.d(TAG, "startRecording");
    assertTrue(audioRecord != null);
    assertTrue(audioThread == null);
    try {
      audioRecord.startRecording();
    } catch (IllegalStateException e) {
      reportWebRtcAudioRecordStartError(AudioRecordStartErrorCode.AUDIO_RECORD_START_EXCEPTION,
          "AudioRecord.startRecording failed: " + e.getMessage());
      return false;
    }
    if (audioRecord.getRecordingState() != AudioRecord.RECORDSTATE_RECORDING) {
      reportWebRtcAudioRecordStartError(
          AudioRecordStartErrorCode.AUDIO_RECORD_START_STATE_MISMATCH,
          "AudioRecord.startRecording failed - incorrect state :"
          + audioRecord.getRecordingState());
      return false;
    }
    audioThread = new AudioRecordThread("AudioRecordJavaThread");
    audioThread.start();
    return true;
  }

​ 在该方法中,首先启动了 audioRecord,接着判断了读取线程事都正在录制中。

三、读数据

 private class AudioRecordThread extends Thread {
    private volatile boolean keepAlive = true;

    public AudioRecordThread(String name) {
      super(name);
    }

    // TODO(titovartem) make correct fix during webrtc:9175
    @SuppressWarnings("ByteBufferBackingArray")
    @Override
    public void run() {
      Process.setThreadPriority(Process.THREAD_PRIORITY_URGENT_AUDIO);
      Logging.d(TAG, "AudioRecordThread" + WebRtcAudioUtils.getThreadInfo());
      assertTrue(audioRecord.getRecordingState() == AudioRecord.RECORDSTATE_RECORDING);

      long lastTime = System.nanoTime();
      while (keepAlive) {
        int bytesRead = audioRecord.read(byteBuffer, byteBuffer.capacity());
        if (bytesRead == byteBuffer.capacity()) {
          if (microphoneMute) {
            byteBuffer.clear();
            byteBuffer.put(emptyBytes);
          }
          // It's possible we've been shut down during the read, and stopRecording() tried and
          // failed to join this thread. To be a bit safer, try to avoid calling any native methods
          // in case they've been unregistered after stopRecording() returned.
          if (keepAlive) {
            nativeDataIsRecorded(bytesRead, nativeAudioRecord);
          }
          if (audioSamplesReadyCallback != null) {
            // Copy the entire byte buffer array.  Assume that the start of the byteBuffer is
            // at index 0.
            byte[] data = Arrays.copyOf(byteBuffer.array(), byteBuffer.capacity());
            audioSamplesReadyCallback.onWebRtcAudioRecordSamplesReady(
                new AudioSamples(audioRecord, data));
          }
        } else {
          String errorMessage = "AudioRecord.read failed: " + bytesRead;
          Logging.e(TAG, errorMessage);
          if (bytesRead == AudioRecord.ERROR_INVALID_OPERATION) {
            keepAlive = false;
            reportWebRtcAudioRecordError(errorMessage);
          }
        }
        if (DEBUG) {
          long nowTime = System.nanoTime();
          long durationInMs = TimeUnit.NANOSECONDS.toMillis((nowTime - lastTime));
          lastTime = nowTime;
          Logging.d(TAG, "bytesRead[" + durationInMs + "] " + bytesRead);
        }
      }

      try {
        if (audioRecord != null) {
          audioRecord.stop();
        }
      } catch (IllegalStateException e) {
        Logging.e(TAG, "AudioRecord.stop failed: " + e.getMessage());
      }
    }

    // Stops the inner thread loop and also calls AudioRecord.stop().
    // Does not block the calling thread.
    public void stopThread() {
      Logging.d(TAG, "stopThread");
      keepAlive = false;
    }
  }

​ 从 AudioRecord去数据的逻辑在 AudioRecordThread 线程的 Run函数中。

四、停止和销毁

  private boolean stopRecording() {
    Logging.d(TAG, "stopRecording");
    assertTrue(audioThread != null);
    audioThread.stopThread();
    if (!ThreadUtils.joinUninterruptibly(audioThread, AUDIO_RECORD_THREAD_JOIN_TIMEOUT_MS)) {
      Logging.e(TAG, "Join of AudioRecordJavaThread timed out");
      WebRtcAudioUtils.logAudioState(TAG);
    }
    audioThread = null;
    if (effects != null) {
      effects.release();
    }
    releaseAudioResources();
    return true;
  }

​ 可以看到,这里首先把AudioRecordThread读数据循环的keepAlive条件置为false,接着调用ThreadUtils.joinUninterruptibly等待AudioRecordThread线程退出。

这里有一点值得一提,keepAlive变量加了volatile关键字进行修饰,这是因为修改和读取这个变量的操作可能发生在不同的线程,使用volatile关键字进行修饰,可以保证修改之后能被立即读取到。

AudioRecordThread线程退出循环后,会调用audioRecord.stop()停止采集;线程退出之后,会调用audioRecord.release()释放AudioRecord对象。

​ 以上,就是 Android WebRTC 音频采集 Java 层的大致流程。

加载全部内容

相关教程
猜你喜欢
用户评论