macOS下AUHAL单元压缩输出流AudioBufferList相关技术问询

Hey there, let's break down this niche AUHAL compressed format issue you're facing—these low-level CoreAudio quirks can be tricky, so let's start with key context and actionable fixes.

First, a quick recap of your setup: you're using CoreAudio/AudioToolbox on macOS, leaning into AUHAL (Audio Unit Host Audio Layer) to record from any input device (totally the right call, since simpler APIs like AVFoundation don't offer the same granular device control). Your pain point is handling the AudioBufferList when the AUHAL unit is configured to output a compressed audio format (unlike uncompressed PCM, which works as expected).

Common Pitfalls with Compressed Formats & AudioBufferList #

Compressed audio behaves fundamentally differently from linear PCM in AUHAL, which is almost certainly where your issues are coming from:

• Buffer Structure: Unlike PCM (which uses 1-2 buffers for mono/stereo interleaved data), compressed formats almost always use a single buffer in the AudioBufferList. Compressed data is packed into a continuous stream, not split per channel like PCM.
• Dynamic Sizing: Compressed data doesn't have a fixed size per frame (unlike PCM, where you calculate size as framesPerBuffer * bytesPerFrame). You need to query the AUHAL unit's maximum packet size to know how much buffer space to allocate.
• Callback Logic: In your render callback, you can't rely on fixed frame counts—you need to work with packet counts and check the actual byte size of the compressed data in the buffer.

Step-by-Step Fixes #

Let's adjust your code to handle compressed formats properly:

1. Validate Format Support:
   First, confirm your target compressed format (like AAC kAudioFormatMPEG4AAC or Apple Lossless kAudioFormatAppleLossless) is supported by the input device and AUHAL. Use this check:

   BOOL formatIsSupported = NO;
   UInt32 size = sizeof(formatIsSupported);
   AudioFormatGetProperty(kAudioFormatProperty_FormatSupported,
                          sizeof(AudioStreamBasicDescription),
                          &yourCompressedFormat,
                          &formatIsSupported,
                          &size);
   

2. Allocate Buffers Correctly:
   Query the AUHAL unit's maximum packet size to size your buffer list properly:

   UInt32 maxPacketSize = 0;
   UInt32 propSize = sizeof(maxPacketSize);
   AudioUnitGetProperty(auhalUnit,
                        kAudioUnitProperty_MaximumOutputPacketSize,
                        kAudioUnitScope_Output,
                        0,
                        &maxPacketSize,
                        &propSize);
   
   // Allocate a single-buffer list for compressed data
   AudioBufferList* bufferList = (AudioBufferList*)malloc(sizeof(AudioBufferList) + sizeof(AudioBuffer));
   bufferList->mNumberBuffers = 1;
   bufferList->mBuffers[0].mNumberChannels = yourCompressedFormat.mChannelsPerFrame;
   bufferList->mBuffers[0].mDataByteSize = maxPacketSize;
   bufferList->mBuffers[0].mData = malloc(maxPacketSize);
   

3. Adjust Render Callback:
   Focus on packet counts and actual byte size instead of frame counts in your callback:

   OSStatus renderCallback(void* inRefCon,
                          AudioUnitRenderActionFlags* ioActionFlags,
                          const AudioTimeStamp* inTimeStamp,
                          UInt32 inBusNumber,
                          UInt32 inNumberFrames,
                          AudioBufferList* ioData) {
      // For compressed formats, use the actual byte size in the buffer
      UInt32 compressedDataSize = ioData->mBuffers[0].mDataByteSize;
      if (compressedDataSize > 0) {
          // Process the compressed data here (write to file, stream, etc.)
          void* compressedData = ioData->mBuffers[0].mData;
      }
      return noErr;
   }
   

4. Double-Check AUHAL Configuration:
   Ensure you've set the AUHAL unit's output format correctly—if the unit can't output compressed data directly, you may need to capture PCM first and use AudioConverter to compress it post-recording.

Key Note #

AUHAL doesn't natively encode compressed audio for all devices. Some hardware has built-in encoders, but others require you to handle compression separately. If you're hitting walls with direct compressed output, switching to PCM capture + post-compression is a reliable fallback.

内容的提问来源于stack exchange，提问作者Pierre P.