我不认为两年后您仍在寻找解决方案,但以防万一有人处于类似情况并找到这个问题(就像我一样),这是我的解决方案。
我的音频队列回调函数调用下面的 appendAudioBuffer 函数,将 AudioQueueBufferRef 及其长度 (mAudioDataByteSize) 传递给它。
void appendAudioBuffer(void* pBuffer, long pLength)
{
// CMSampleBuffers require a CMBlockBuffer to hold the media data; we
// create a blockBuffer here from the AudioQueueBuffer's data.
CMBlockBufferRef blockBuffer;
OSStatus status = CMBlockBufferCreateWithMemoryBlock(kCFAllocatorDefault,
pBuffer,
pLength,
kCFAllocatorNull,
NULL,
0,
pLength,
kCMBlockBufferAssureMemoryNowFlag,
&blockBuffer);
// Timestamp of current sample
CFAbsoluteTime currentTime = CFAbsoluteTimeGetCurrent();
CFTimeInterval elapsedTime = currentTime - mStartTime;
CMTime timeStamp = CMTimeMake(elapsedTime * mTimeScale, mTimeScale);
// Number of samples in the buffer
long nSamples = pLength / mWaveRecorder->audioFormat()->mBytesPerFrame;
CMSampleBufferRef sampleBuffer;
OSStatus err = CMAudioSampleBufferCreateWithPacketDescriptions(kCFAllocatorDefault,
blockBuffer,
true,
NULL,
NULL,
mAudioFormatDescription,
nSamples,
timeStamp,
NULL,
&sampleBuffer);
// Add the audio sample to the asset writer input
if ([mAudioWriterInput isReadyForMoreMediaData]) {
if(![mAudioWriterInput appendSampleBuffer:sampleBuffer])
// print an error
}
else
// either do nothing and just print an error, or queue the CMSampleBuffer
// somewhere and add it later, when the AVAssetWriterInput is ready
CFRelease(sampleBuffer);
CFRelease(blockBuffer);
}
请注意,当我调用appendAudioBuffer 时,声音没有被压缩;音频格式被指定为 LPCM(这就是我不使用数据包描述符的原因,因为 LPCM 没有)。 AVAssetWriterInput 处理压缩。
我最初尝试将 AAC 数据传递给 AVAssetWriter,但这导致了太多的复杂性,我无法让它工作。