为具有立体声音频流的两个缓冲区分配AudioBufferList

时间:2018-03-03 18:14:31

标签: swift core-audio audiobufferlist

我在C中有以下代码,它们分配一个具有适当长度的AudioBufferList。

  UInt32 bufferSizeBytes = bufferSizeFrames * sizeof(Float32);
  propertySize = offsetof(AudioBufferList, mBuffers[0]) + (sizeof(AudioBuffer) * mRecordSBD.mChannelsPerFrame);

  mBufferList = (AudioBufferList *) malloc(propertySize);
  mBufferList->mNumberBuffers = mRecordSBD.mChannelsPerFrame;
  for(UInt32 i = 0; i < mBufferList->mNumberBuffers; ++i)
  {
    mBufferList->mBuffers[i].mNumberChannels = 1;
    mBufferList->mBuffers[i].mDataByteSize = bufferSizeBytes;
    mBufferList->mBuffers[i].mData = malloc(bufferSizeBytes);
  }

大多数情况下,mChannelsPerFrame为2,因此上面的代码创建了两个缓冲区,每个缓冲区一个缓冲区。每个缓冲区都有一个值bufferSizeBytes的保留内存。

如何在Swift中复制相同的行为?

2 个答案:

答案 0 :(得分:4)

不幸的是,在Swift中,C数组被视为元组,因此AudioBufferList.mBuffers导入一个带有单个AudioBuffer的元组。 C允许您通过使用指针数学(或本例中的数组下标)来访问相邻内存,以便创建可变长度结构,Swift不会。

AudioBufferList只是简单的并没有很好地转换为Swift。 Apple通过一些辅助函数和类型来缓解这个问题。他们创建了一个静态allocate函数,它返回一个UnsafeMutableAudioBufferListPointer,这是一个特殊的类型,其中下标返回audioBuffers。

let bufferSizeBytes = MemoryLayout<Float>.size * 1234

var bufferlist = AudioBufferList.allocate(maximumBuffers: 2)
bufferlist[0] = AudioBuffer(mNumberChannels: 1,
                            mDataByteSize: UInt32(bufferSizeBytes),
                            mData: malloc(bufferSizeBytes))
bufferlist[1] = AudioBuffer(mNumberChannels: 1,
                            mDataByteSize: UInt32(bufferSizeBytes),
                            mData: malloc(bufferSizeBytes))

// Free your buffers and the pointer when you're done.
for buffer in bufferlist {
    free(buffer.mData)
}
free(&bufferlist)

答案 1 :(得分:0)

您可以使用Core Audio API为2个交错流创建一个带有2个缓冲区的AudioBufferList:

import AudioUnit
import AVFoundation

var myBufferList = AudioBufferList(
              mNumberBuffers: 2,
              mBuffers: AudioBuffer(
                  mNumberChannels: UInt32(2),
                  mDataByteSize: myBufferSizeBytes,
                  mData: nil) )