PCM操作

简单说明

Webrtc中有一个类专门用于音频处理-AudioFrameOperations，提供了很实用的功能。

• MonoToStereo：单声道转立体声，简单的复制。
• StereoToMono：立体声转单声道，两个声道相加除以2。
• SwapStereoChannels：左右声道交换。
• Mute：哑音，把所有通道对应的采样点全部置0，采样点个数就要看采样率了。
• Scale：立体声音量控制，可以单独控制左右声道的音量，范围是[0.0, n.0]，这个方法效率高，但是效果不好，因为音量改变好直接取低16bit，这样得到的值不够真实。
• ScaleWithSat：多声道音量改变，范围是[0.0, n.0]，这个改变要平滑一些。Android源码中也是采用这样的做法，不够采用位移的方式。

其他阅读

1. PCM混音
2. 音频术语

代码

• 头文件在webrtc/modules/utility/include/audio_frame_operations.h

#ifndef WEBRTC_MODULES_UTILITY_INCLUDE_AUDIO_FRAME_OPERATIONS_H_
#define WEBRTC_MODULES_UTILITY_INCLUDE_AUDIO_FRAME_OPERATIONS_H_

#include "webrtc/typedefs.h"

namespace webrtc {

class AudioFrame;

// TODO(andrew): consolidate this with utility.h and audio_frame_manipulator.h.
// Change reference parameters to pointers. Consider using a namespace rather
// than a class.
class AudioFrameOperations {
 public:
  // Upmixes mono |src_audio| to stereo |dst_audio|. This is an out-of-place
  // operation, meaning src_audio and dst_audio must point to different
  // buffers. It is the caller's responsibility to ensure that |dst_audio| is
  // sufficiently large.
  static void MonoToStereo(const int16_t* src_audio, size_t samples_per_channel,
                           int16_t* dst_audio);
  // |frame.num_channels_| will be updated. This version checks for sufficient
  // buffer size and that |num_channels_| is mono.
  static int MonoToStereo(AudioFrame* frame);

  // Downmixes stereo |src_audio| to mono |dst_audio|. This is an in-place
  // operation, meaning |src_audio| and |dst_audio| may point to the same
  // buffer.
  static void StereoToMono(const int16_t* src_audio, size_t samples_per_channel,
                           int16_t* dst_audio);
  // |frame.num_channels_| will be updated. This version checks that
  // |num_channels_| is stereo.
  static int StereoToMono(AudioFrame* frame);

  // Swap the left and right channels of |frame|. Fails silently if |frame| is
  // not stereo.
  static void SwapStereoChannels(AudioFrame* frame);

  // Zeros out the audio and sets |frame.energy| to zero.
  static void Mute(AudioFrame& frame);

  static int Scale(float left, float right, AudioFrame& frame);

  static int ScaleWithSat(float scale, AudioFrame& frame);
};

}  // namespace webrtc

#endif  // #ifndef WEBRTC_MODULES_UTILITY_INCLUDE_AUDIO_FRAME_OPERATIONS_H_

• 实现文件的路径在webrtc/modules/utility/source/audio_frame_operations.cc

#include "webrtc/modules/include/module_common_types.h"
#include "webrtc/modules/utility/include/audio_frame_operations.h"

namespace webrtc {

void AudioFrameOperations::MonoToStereo(const int16_t* src_audio,
                                        size_t samples_per_channel,
                                        int16_t* dst_audio) {
  for (size_t i = 0; i < samples_per_channel; i++) {
    dst_audio[2 * i] = src_audio[i];
    dst_audio[2 * i + 1] = src_audio[i];
  }
}

int AudioFrameOperations::MonoToStereo(AudioFrame* frame) {
  if (frame->num_channels_ != 1) {
    return -1;
  }
  if ((frame->samples_per_channel_ * 2) >= AudioFrame::kMaxDataSizeSamples) {
    // Not enough memory to expand from mono to stereo.
    return -1;
  }

  int16_t data_copy[AudioFrame::kMaxDataSizeSamples];
  memcpy(data_copy, frame->data_,
         sizeof(int16_t) * frame->samples_per_channel_);
  MonoToStereo(data_copy, frame->samples_per_channel_, frame->data_);
  frame->num_channels_ = 2;

  return 0;
}

void AudioFrameOperations::StereoToMono(const int16_t* src_audio,
                                        size_t samples_per_channel,
                                        int16_t* dst_audio) {
  for (size_t i = 0; i < samples_per_channel; i++) {
    dst_audio[i] = (src_audio[2 * i] + src_audio[2 * i + 1]) >> 1;
  }
}

int AudioFrameOperations::StereoToMono(AudioFrame* frame) {
  if (frame->num_channels_ != 2) {
    return -1;
  }

  StereoToMono(frame->data_, frame->samples_per_channel_, frame->data_);
  frame->num_channels_ = 1;

  return 0;
}

void AudioFrameOperations::SwapStereoChannels(AudioFrame* frame) {
  if (frame->num_channels_ != 2) return;

  for (size_t i = 0; i < frame->samples_per_channel_ * 2; i += 2) {
    int16_t temp_data = frame->data_[i];
    frame->data_[i] = frame->data_[i + 1];
    frame->data_[i + 1] = temp_data;
  }
}

void AudioFrameOperations::Mute(AudioFrame& frame) {
  memset(frame.data_, 0, sizeof(int16_t) *
      frame.samples_per_channel_ * frame.num_channels_);
}

int AudioFrameOperations::Scale(float left, float right, AudioFrame& frame) {
  if (frame.num_channels_ != 2) {
    return -1;
  }

  for (size_t i = 0; i < frame.samples_per_channel_; i++) {
    frame.data_[2 * i] =
        static_cast<int16_t>(left * frame.data_[2 * i]);
    frame.data_[2 * i + 1] =
        static_cast<int16_t>(right * frame.data_[2 * i + 1]);
  }
  return 0;
}

int AudioFrameOperations::ScaleWithSat(float scale, AudioFrame& frame) {
  int32_t temp_data = 0;

  // Ensure that the output result is saturated [-32768, +32767].
  for (size_t i = 0; i < frame.samples_per_channel_ * frame.num_channels_;
       i++) {
    temp_data = static_cast<int32_t>(scale * frame.data_[i]);
    if (temp_data < -32768) {
      frame.data_[i] = -32768;
    } else if (temp_data > 32767) {
      frame.data_[i] = 32767;
    } else {
      frame.data_[i] = static_cast<int16_t>(temp_data);
    }
  }
  return 0;
}

}  // namespace webrtc

• ScaleWithSat可以优化一下

static inline int16_t clamp16(int32_t sample)
{
    if ((sample>>15) ^ (sample>>31))
        sample = 0x7FFF ^ (sample>>31);
    return sample;
}