CMReadySampleBuffer.copyPCMData() fails with -12731 (kCMSampleBufferError_RequiredParameterMissing) error

Question

I'm trying to figure out how to use the new outputProvider API on AVAssetReader. Specifically, I want to open a media file (video or audio), extract its audio, and write it back out as PCM. I have the following code:

import Foundation
import AVFoundation
enum AudioExtractorError: Error {
 case noAudio
 case bufferAllocationFailed
}
@available(macOS 26.0, iOS 26.0, *)
class AudioExtractor {
 init(fileURL: URL) {
 self.fileURL = fileURL
 }
 func extractAudio(from fileURL: URL, to outputFolder: URL) async throws -> URL {
 let asset = AVAsset(url: fileURL)
 guard let audioTrack = try await asset.loadTracks(withMediaType: .audio).first else {
 throw AudioExtractorError.noAudio
 }
 let reader = try AVAssetReader(asset: asset)
 let outputSettings: [String: Any] = [
 AVFormatIDKey: kAudioFormatLinearPCM,
 AVLinearPCMIsBigEndianKey: false,
 AVLinearPCMBitDepthKey: 32,
 AVLinearPCMIsFloatKey: true,
 AVLinearPCMIsNonInterleaved: false,
 AVNumberOfChannelsKey: 2
 ]
 let trackOutput = AVAssetReaderTrackOutput(track: audioTrack, outputSettings: outputSettings)
 let outputProvider = reader.outputProvider(for: trackOutput)
 try reader.start()
 let outputURL = outputFolder
 .appendingPathComponent(UUID().uuidString)
 .appendingPathExtension("caf")
 let outputFile = try AVAudioFile(forWriting: outputURL, settings: outputSettings)
 while let sampleBuffer = try await outputProvider.next() {
 guard case let CMSampleBuffer.DynamicContent.dataBuffer(blockBuffer) = sampleBuffer.content else {
 continue
 }
 let blockSampleBuffer = CMReadySampleBuffer(dataBuffer: blockBuffer,
 formatDescription: sampleBuffer.formatDescription!,
 sampleProperties: sampleBuffer.sampleProperties)
 guard let pcmBuffer = AVAudioPCMBuffer(pcmFormat: outputFile.processingFormat, frameCapacity: AVAudioFrameCount(sampleBuffer.sampleCount)) else {
 throw AudioExtractorError.bufferAllocationFailed
 }
 try blockSampleBuffer.copyPCMData(fromRange: 0..<sampleBuffer.sampleCount, into: pcmBuffer.mutableAudioBufferList)
 try outputFile.write(from: pcmBuffer)
 }
 return outputURL
 }
 let fileURL: URL
}

When I run this, the call to blockSampleBuffer.copyPCMData() fails with error code -12731 which is kCMSampleBufferError_RequiredParameterMissing. Has anyone managed to get this API working?

• Gordon Childs: I see that these APIs are new, but are they really new or just swift annotations for existing APIs? e.g. CMSampleBufferCopyPCMDataIntoAudioBufferList(). I guess those structs and enums go a little beyond the usual syntactic sugar. TIL!
• Andrew Madsen: I'm pretty sure they're actually new APIs, though they certainly do seem designed to be more Swifty, at least in that they include richer information for the type system and therefore more type safety. Not that it's enough to save you from the trademark inscrutable Core Audio/Core Media error codes...
• Gordon Childs: Agreed, kCMSampleBufferError_RequiredParameterMissing was completely misleading.

Answer 1

It's great to see a new question on here!

There are two sources of copyPCMData() returning -12731 = kCMSampleBufferError_RequiredParameterMissing in your code:

1. The AVAudioFile initializer you used gives you an interleaved fileFormat but a deinterleaved processingFormat which does not match AVAssetReaderTrackOutput's configuration (it's not obvious, I've mentioned this before)
2. You're not setting the AVAudioPCMBuffer.frameLength which results in a 0 AudioBuffer.mDataByteSize

With the following changes, the code works for me:

- let outputFile = try AVAudioFile(forWriting: outputURL, settings: outputSettings)
+ let outputFile = try AVAudioFile(forWriting: outputURL, settings: outputSettings, commonFormat: .pcmFormatFloat32, interleaved: true) // note the inverted senses of interleaved vs AVLinearPCMIsNonInterleaved!
...
+ pcmBuffer.frameLength = AVAudioFrameCount(sampleBuffer.sampleCount)
 try blockSampleBuffer.copyPCMData(fromRange: 0..<sampleBuffer.sampleCount,
 into: pcmBuffer.mutableAudioBufferList)

• kakaiikaka: Impressive! I was suspecting the difference between processing format and file format, and I know apple really don't like planar(non-interleave) audio formats, that's why we need the second AVAudioFile initializer to set interleaved to true, right?
• Gordon Childs: Thanks! I’d phrase it slightly differently: planar (non-interleaved) formats are more common in Apple’s capture APIs and less common in file I/O. Planarity ties closely to buffer sizing, which tends to matter more during capture and less during disk operations. With that in mind, I think the AVAudioFile initializer Andrew used was designed primarily for the workflow of writing captured audio to disk.
• Gordon Childs: AVAudioFile has some funny API corner cases, some inherited fromExtAudioFile, some not. My favourite is that it lacked a close method for a decade or so: stackoverflow.com/a/52122691/22147
• kakaiikaka: Many thanks for the answer!
• Andrew Madsen: Thanks! I figured out the frameLength problem (stupid mistake), but wouldn't have figured out the interleaving mismatch without your answer. And you mentioned it before in an answer to a question from me, no less! I wish I had a better memory...
• Gordon Childs: Ha! I missed that that was you! Wish I had a better memory too.