I have an AUv3 plugin which uses an FFT - which requires n samples before it can produce any output - so, depending on the relation between the host's buffer size and the FFT window size, it may receive a several buffers of samples, producing no output, and then dumping out what it has once a sufficient number of samples have been received. This means that output is produced in fits and starts, in batches that match the FFT size (modulo oversampling) - e.g. if being fed buffers of 256 samples with an fft size of 1024, the output buffer sizes will be 0 for the first 3 buffers, and upon the fourth, the first 256 processed samples are returned and the remaining 768 cached; the next three buffers will return the remaining cached samples while processing and buffering subsequent ones, and so forth. The internal mechanics of that I have solved, caching output if the current output buffer is too small, and so forth - so it all works as advertised, and the plugin reports its latency correctly. And when run as an app in demo-mode, playback works as expected. In the plugin's render block, it captures the number of frames written, and if it is less than the number of frames passed in, adjusts the mDataByteSize of the output buffers to match the actual quantity of data being returned: unsigned int framesWritten = (unsigned int) processHelper->processWithEvents(inAudioBufferList, outAudioBufferList, timestamp, frameCount, realtimeEventListHead); if (framesWritten < frameCount) { for (UInt32 i = 0; i < outAudioBufferList->mNumberBuffers; ++i) { outAudioBufferList->mBuffers[i].mDataByteSize = framesWritten * 4; // assume 4 byte floats } } However, there are a couple of serious issues: auval -v fails it with - Render Test at 64 frames, sample rate: 22050 Hz ERROR: Output Buffer Size does not match requested When connected to Logic Pro, it appears that mDataByteSize is ignored, and the entire allocated buffer is read - audio has sections of silence snipped into it which corresponds the number of empty buffers being returned If I set Logic's buffer size to 1024 and use a 1024 sample FFT window, the plugin works correctly - but of course a plugin cannot dictate buffer size, and `1024 is too small a window size to be useful for anything but filtering very high frequencies This seems like it has to be a solvable problem, and most likely the issue is in how my code reports the number of usable samples in the returned buffer. So, what is the correct way for a plugin to report that it has no samples to return, but will, uh, real soon now? I know I could convert this plugin to be one that does offline rendering of the entire input, but this is real-time processing, just with a fixed amount of latency, so that should not be necessary.

Hi, On macOS I used to open MP3 and MP4 files with ExtAudioFile. For a few years it doesn't work anymore. So I decided to try different macOS API using the AudioFileID of AudioToolbox framework. I decided to write a test: https://gist.github.com/joelkraehemann/7f5b241b52ca38c3a765c138fb647588 It fails right here: AudioFileOpenWithCallbacks() By telling OSStatus error 1954115647, which means kAudioFileUnsupportedFileTypeError. The filename was set to an MP4 file: ~/Music/test.mp4 Howto fix this? regards, Joël

Hi all, I’ve implemented the new Core Audio Tap API (AudioHardwareCreateProcessTap with CATapDescription) and I’m seeing consistent level attenuation that scales with the number of stereo output pairs exposed by the target device. What I observe Device with 4 stereo pairs (8 outs) → tap shows −12.04 dB relative to source. True 2-ch devices (built-in speakers, AirPods) → ~0 dB attenuation. The attenuation appears regardless of whether I: Create a global (default-output) tap via initStereoGlobalTapButExcludeProcesses: Or create a per-process/per-device tap via initWithProcesses:andDeviceUID:withStream: Additionally, the routing choice inside the sending app matters: App output to “System/Default Output” → I often see no attenuation. App output directly to a multi-out interface (e.g., RME Fireface) → I see the pair-count-scaled attenuation. I can query Core Audio for the number of output channels/pairs and gain-compensate (+20·log10(N_pairs) dB) and that matches my measurements for many cases. However, this compensation is not universally correct because it seems to depend on where each process routes its audio (Default Output vs. direct device), even when those processes are included in the same tap aggregate. Question Is there a supported way to obtain the raw, unattenuated streams for all processes through the Tap API—i.e., to bypass this automatic headroom/attenuation behavior entirely? If this attenuation is expected by design: Is there a documented rule for when it applies (global vs. device taps, per-process taps, stream selection, etc.)? Is there a property/flag to disable it, or a reliable, official method to compute the exact compensation (beyond counting stereo pairs)? Any guidance on ensuring consistent levels when multiple processes route differently (Default Output vs. direct device) but are captured by the same tap? Environment API: AudioHardwareCreateProcessTap + CATapDescription Devices: built-in output (2-ch), RME Fireface (8+ outs / 4+ stereo pairs) Behavior reproducible with both global and per-process/per-device tap descriptions. Attenuation example: 4 stereo pairs → −12.04 dB observed. Happy to provide a minimal sample, measurements, and device logs. Thanks! — David

I have integrated the ShazamKit SDK into my iOS app and would like to implement the same functionality in my Android app. My question is: Can I use the Android version of the ShazamKit SDK for commercial purposes? After extensive research, I could not find any official information regarding the license of the Android version of the ShazamKit SDK. Could you please provide a formal license statement?

Hi, I try to record audio on the iPhone with the AVAudioRecorder and Xcode 26.0.1. Maybe the problem is that I can not record audio with the simulator. But there's a menu for audio. In the plist I added 'Privacy - Microphone Usage Description' and I ask for permission before recording. if await AVAudioApplication.requestRecordPermission() { print("permission granted") recordPermission = true } else { print("permission denied") } Permission is granted. let settings: [String : Any] = [ AVFormatIDKey: kAudioFormatMPEG4AAC, AVSampleRateKey: 12000, AVNumberOfChannelsKey: 1, AVEncoderAudioQualityKey: AVAudioQuality.high.rawValue ] recorder = try AVAudioRecorder(url: filename, settings: settings) let prepared = recorder.prepareToRecord() print("prepared started: \(prepared)") let started = recorder.record() print("recording started: \(started)") started is always false and I tried many settings. Error messages AddInstanceForFactory: No factory registered for id <CFUUID 0x600000211480> F8BB1C28-BAE8-11D6-9C31-00039315CD46 AudioConverter.cpp:1052 Failed to create a new in process converter -> from 0 ch, 12000 Hz, .... (0x00000000) 0 bits/channel, 0 bytes/packet, 0 frames/packet, 0 bytes/frame to 1 ch, 12000 Hz, aac (0x00000000) 0 bits/channel, 0 bytes/packet, 1024 frames/packet, 0 bytes/frame, with status -50 AudioQueueObject.cpp:1892 BuildConverter: AudioConverterNew returned -50 from: 0 ch, 12000 Hz, .... (0x00000000) 0 bits/channel, 0 bytes/packet, 0 frames/packet, 0 bytes/frame to: 1 ch, 12000 Hz, aac (0x00000000) 0 bits/channel, 0 bytes/packet, 1024 frames/packet, 0 bytes/frame prepared started: true AudioQueueObject.cpp:7581 ConvertInput: aq@0x10381be00: AudioConverterFillComplexBuffer returned -50, packetCount 5 recording started: false All examples I find are the same, but apparently there must be something different.

Hello. My app uses AVAudioRecorder to generate recording files, which are consistently only 4kb in size. Most users generate audio files normally, with only a few users experiencing this phenomenon occasionally. After uninstalling and installing the app, it will work normally, but it will reappear after a period of time. I have compared that the problematic audio files generated each time are fixed and cannot be played. Added the audioRecorderDidFinishRecording proxy method, which shows that the recording was completed normally. The user also reported that the recording is normal, but there is a problem with the generated file. How should I handle this issue? Look forward to your reply. - (void)startRecordWithOrderID:(NSString *)orderID { AVAudioSession *audioSession = [AVAudioSession sharedInstance]; [audioSession setCategory:AVAudioSessionCategoryRecord error:nil]; [audioSession setActive:YES error:nil]; NSMutableDictionary *settings = [[NSMutableDictionary alloc] init]; [settings setObject:[NSNumber numberWithFloat: 8000.0] forKey:AVSampleRateKey]; [settings setObject:[NSNumber numberWithInt: kAudioFormatLinearPCM] forKey:AVFormatIDKey]; [settings setObject:[NSNumber numberWithInt:16] forKey:AVLinearPCMBitDepthKey]; [settings setObject:[NSNumber numberWithInt: 1] forKey:AVNumberOfChannelsKey]; [settings setObject:[NSNumber numberWithBool:NO] forKey:AVLinearPCMIsBigEndianKey]; [settings setObject:[NSNumber numberWithBool:NO] forKey:AVLinearPCMIsFloatKey]; NSString *path = [WDUtility createDirInDocument:@"audios" withOrderID:orderID withPathExtension:@"wav"]; NSURL *tmpFile = [NSURL fileURLWithPath:path]; recorder = [[AVAudioRecorder alloc] initWithURL:tmpFile settings:settings error:nil]; [recorder setDelegate:self]; [recorder prepareToRecord]; [recorder record]; }

My app Balletrax is a music player for people to use while they teach ballet. Used to be you could silence notifications during use, but now the customer seems to have to know how to use Focus mode, remember to turn it on and off, and have to check the notifications one does and doesn't want to use. Is there no way to silence all notifications when the app is in use?

Hey everyone, I'm encountering an issue with audio sample rate conversion that I'm hoping someone can help with. Here's the breakdown: Issue Description: I've installed a tap on an input device to convert audio to an optimal sample rate. There's a converter node added on top of this setup. The problem arises when joining Zoom or FaceTime calls—the converter gets deallocated from memory, causing the program to crash. Symptoms: The converter node is being deallocated during video calls. The program crashes entirely when this happens. Traditional methods of monitoring sample rate changes (tracking nominal or actual sample rates) aren't working as expected. The Big Challenge: I can't figure out how to properly monitor sample rate changes. Listeners set up to track these changes don't trigger when the device joins a Zoom or FaceTime call. Please, if anyone has experience with this or knows a solution, I'd really appreciate your help. Thanks in advance! ⁠

The device is connected to Bluetooth A and Bluetooth B, currently the audio is played through Bluetooth A, click the interface button, how to realize the code to switch to Bluetooth B？

I’m running the iOS 26.2 Public Beta update and my album artwork is missing from the music app (I’m not using Apple Music). I use google to get my album artwork. Do I need to wait for a new update?

We have the necessary background recording entitlements, and for many users... do not run into any issues. However, there is a subset of users that routinely get recordings ending.. we have narrowed this down and believe it to be the work of the watch dog. First we removed the entire view hierarchy when app is backgrounded. There is just 'Text("Recording")' This got the CPU usage in profiler down to 0%. We saw massive improvements to recording success rate. We walked away assuming that was enough. However we are still seeing the same sort of crashes. All in the background. We're using Observation to drive audio state changes to a Live Activity. Are those Observations causing the problem? Why doesn't apple provide a better API to background audio? The internet is full of weird issues https://stackoverflow.com/questions/76010213/why-is-my-react-native-app-sometimes-terminated-in-the-background-while-tracking https://stackoverflow.com/questions/71656047/why-is-my-react-native-app-terminating-in-the-background-while-recording-ios-r https://github.com/expo/expo/issues/16807 This is such a terrible user experience. And we have very little visibility into what is happening and why. No where in apple documentation states that in order for background recording to work, the app can only be 'Text("Recording")' It does not outline a CPU or memory threshold. It just kills us.

My current app implements a custom video player, based on a AVSampleBufferRenderSynchronizer synchronising two renderers: an AVSampleBufferDisplayLayer receiving decoded CVPixelBuffer-based video CMSampleBuffers, and an AVSampleBufferAudioRenderer receiving decoded lpcm-based audio CMSampleBuffers. The AVSampleBufferRenderSynchronizer is started when the first image (in presentation order) is decoded and enqueued, using avSynchronizer.setRate(_ rate: Float, time: CMTime), with rate = 1 and time the presentation timestamp of the first decoded image. Presentation timestamps of video and audio sample buffers are consistent, and on most streams, the audio and video are correctly synchronized. However on some network streams, on iOS, the audio and video aren't synchronized, with a time difference that seems to increase with time. On the other hand, with the same player code and network streams on macOS, the synchronization always works fine. This reminds me of something I've read, about cases where an AVSampleBufferRenderSynchronizer could not synchronize audio and video, causing them to run with independent and potentially drifting clocks, but I cannot find it again. So, any help / hints on this sync problem will be greatly appreciated! :)

According to the header file the outputVolume properties supported range is 0.0-1.0: /*! @property outputVolume @abstract The mixer's output volume. @discussion This accesses the mixer's output volume (0.0-1.0, inclusive). @property (nonatomic) float outputVolume; However when setting the volume to 2.0 the audio does indeed play louder. Is the header file out of date and if so, what is the supported range for outputVolume? Thanks

I have the new iOS 26 SpeechTranscriber working in my application. The issue I am facing is how to determine if the device I am running on supports SpeechTranscriber. I was able to create code that tests if the device supports transcription but it takes a bit of time to run and thus the results are not available when the app launches. What I am looking for is a list of what iOS 26 devices it doesn't run on. I think its safe to assume any new devices will support it so if we can just have a list of what devices that can run iOS 26 and not able to do transcription it would be much faster for the app. I have determined it doesn't work on a SE 2nd Gen, it works on iPhone 12, SE 3rd Gen, iPhone 14 Pro, 15 Pro. As the SpeechTranscriber doesn't work in the simulator I can't determine that way. I have checked the docs and it doesn't list the devices it doesn't work on.

Two issues: No matter what I set in try audioSession.setPreferredSampleRate(x) the sample rate on both iOS and macOS is always 48000 when the output goes through the speaker, and 24000 when my Airpods connect to an iPhone/iPad. Now, I'm checking the current output loudness to animate a 3D character, using mixerNode.installTap(onBus: 0, bufferSize: y, format: nil) { [weak self] buffer, time in Task { @MainActor in // calculate rms and animate character accordingly but any buffer size under 4800 is just ignored and the buffers I get are 4800 sized. This is ok, when the sampleRate is currently 48000, as 10 samples per second lead to decent visual results. But when AirPods connect, the samplerate is 24000, which means only 5 samples per second, so the character animation looks lame. My AVAudioEngine setup is the following: audioEngine.connect(playerNode, to: pitchShiftEffect, format: format) audioEngine.connect(pitchShiftEffect, to: mixerNode, format: format) audioEngine.connect(mixerNode, to: audioEngine.outputNode, format: nil) Now, I'd be fine if the outputNode runs at whatever if it needs, as long as my tap would get at least 10 samples per second. PS: Specifying my favorite format in the let format = AVAudioFormat(standardFormatWithSampleRate: 48_000, channels: 2)! mixerNode.installTap(onBus: 0, bufferSize: y, format: format) doesn't change anything either

Hello, I've discovered a buffer initialization bug in AVAudioUnitSampler that happens when loading presets with multiple zones referencing different regions in the same audio file (monolith/concatenated samples approach). Almost all zones output silence (i.e. zeros) at the beginning of playback instead of starting with actual audio data. The Problem Setup: Single audio file (monolith) containing multiple concatenated samples Multiple zones in an .aupreset, each with different sample start and sample end values pointing to different regions of the same file All zones load successfully without errors Expected Behavior: All zones should play their respective audio regions immediately from the first sample. Actual Behavior: Last zone in the zone list: Works perfectly - plays audio immediately All other zones: Output [0, 0, 0, 0, ..., _audio_data] instead of [real_audio_data] The number of zeros varies from event to event for each zone. It can be a couple of samples (<30) up to several buffers. After the initial zeros, the correct audio plays normally, so there is no shift in audio playback, just missing samples at the beginning. Minimal Reproduction 1. Create Test Monolith Audio File Create a single Wav file with 3 concatenated 1-second samples (44.1kHz): Sample 1: frames 0-44099 (constant amplitude 0.3) Sample 2: frames 44100-88199 (constant amplitude 0.6) Sample 3: frames 88200-132299 (constant amplitude 0.9) 2. Create Test Preset Create an .aupreset with 3 zones all referencing the same file: Pseudo code <Zone array> <zone 1> start : 0, end: 44099, note: 60, waveform: ref_to_monolith.wav; <zone 2> start sample: 44100, note: 62, end sample: 88199, waveform: ref_to_monolith.wav; <zone 3> start sample: 88200, note: 64, end sample: 132299, waveform: ref_to_monolith.wav; </Zone array> 3. Load and Test // Load preset into AVAudioUnitSampler let sampler = AVAudioUnitSampler() try sampler.loadAudioFiles(from: presetURL) // Play each zone (MIDI notes C4=60, D4=62, E4=64) sampler.startNote(60, withVelocity: 64, onChannel: 0) // Zone 1 sampler.startNote(62, withVelocity: 64, onChannel: 0) // Zone 2 sampler.startNote(64, withVelocity: 64, onChannel: 0) // Zone 3 4. Observed Result Zone 1 (C4): [0, 0, 0, ..., 0.3, 0.3, 0.3] ❌ Zeros at beginning Zone 2 (D4): [0, 0, 0, ..., 0.6, 0.6, 0.6] ❌ Zeros at beginning Zone 3 (E4): [0.9, 0.9, 0.9, ...] ✅ Works correctly (last zone) What I've Extensively Tested What DOES Work Separate files per zone: Each zone references its own individual audio file All zones play correctly without zeros Problem: Not viable for iOS apps with 500+ sample libraries due to file handle limitations What DOESN'T Work (All Tested) 1. Different Audio Formats: CAF (Float32 PCM, Int16 PCM, both interleaved and non-interleaved) M4A (AAC compressed) WAV (uncompressed) SF2 (SoundFont2) Bug persists across all formats 2. CAF Region Chunks: Created CAF files with embedded region chunks defining zone boundaries Set zones with no sampleStart/sampleEnd in preset (nil values) AVAudioUnitSampler completely ignores CAF region metadata Bug persists 3. Unique Waveform IDs: Gave each zone a unique waveform ID (268435456, 268435457, 268435458) Each ID has its own file reference entry (all pointing to same physical file) Hypothesized this might trigger separate buffer initialization Bug persists - no improvement 4. Different Sample Rates: Tested: 44.1kHz, 48kHz, 96kHz Bug occurs at all sample rates 5. Mono vs Stereo: Bug occurs with both mono and stereo files Environment macOS: Sonoma 14.x (tested across multiple minor versions) iOS: Tested on iOS 17.x with same results Xcode: 16.x Frameworks: AVFoundation, AudioToolbox Reproducibility: 100% reproducible with setup described above Impact & Use Case This bug severely impacts professional music applications that need: Small file sizes: Monolith files allow sharing compressed audio data (AAC/M4A) iOS file handle limits: Opening 400+ individual sample files is not viable on iOS Performance: Single file loading is much faster than hundreds of individual files Standard industry practice: Monolith/concatenated samples are used by EXS24, Kontakt, and most professional samplers Current Impact: Cannot use monolith files with AVAudioUnitSampler on iOS Forced to choose between: unusable audio (zeros at start) OR hitting iOS file limits No viable workaround exists Root Cause Hypothesis The bug appears to be in AVAudioUnitSampler's internal buffer initialization when: Multiple zones share the same source audio file Each zone specifies different sampleStart/sampleEnd offsets Key observation: The last zone in the zone array always works correctly. This is NOT related to: File permissions or security-scoped resources (separate files work fine) Audio codec issues (happens with uncompressed PCM too) Preset parsing (preset loads correctly, all zones are valid) Questions Is this a known issue? I couldn't find any documentation, bug reports, or discussions about this. Is there ANY workaround that allows monolith files to work with AVAudioUnitSampler? Alternative APIs? Is there a different API or approach for iOS that properly supports monolith sample files?

Hello, I'm trying to receive parquet files using the example that provided in documentation. I've done all required steps but receive constantly error 500 with "Upstream Service Error". By looking into the issues list, seems this error exists for months. Is it possible to get it working?

I wrote a Swift macOS app to control a PCI audio device. The code switches between the default output and input channels. As soon as I launch the Audio-Midi Setup utility, channel switching stops working. The driver properties allow switching, but the system doesn't respond. I have to delete the contents of /Library/Preferences/Audio and reset Core Audio. What am I missing? func setDefaultChannelsOutput() { guard let deviceID = getDeviceIDByName(deviceName: "PCI-424") else { return } let selectedIndex = DefaultChannelsOutput.indexOfSelectedItem if selectedIndex < 0 || selectedIndex >= 24 { return } let channel1 = UInt32(selectedIndex * 2 + 1) let channel2 = UInt32(selectedIndex * 2 + 2) var channels: [UInt32] = [channel1, channel2] var propertyAddress = AudioObjectPropertyAddress( mSelector: kAudioDevicePropertyPreferredChannelsForStereo, mScope: kAudioDevicePropertyScopeOutput, mElement: kAudioObjectPropertyElementWildcard ) let dataSize = UInt32(MemoryLayout<UInt32>.size * channels.count) let status = AudioObjectSetPropertyData(deviceID, &propertyAddress, 0, nil, dataSize, &channels) if status != noErr { print("Error setting default output channels: \(status)") } }

Hi everyone, I’m testing audio recording on an iPhone 15 Plus using AVFoundation. Here’s a simplified version of my setup: let settings: [String: Any] = [ AVFormatIDKey: Int(kAudioFormatLinearPCM), AVSampleRateKey: 8000, AVNumberOfChannelsKey: 1, AVLinearPCMBitDepthKey: 16, AVLinearPCMIsFloatKey: false ] audioRecorder = try AVAudioRecorder(url: fileURL, settings: settings) audioRecorder?.record() When I check the recorded file’s sample rate, it logs: Actual sample rate: 8000.0 However, when I inspect the hardware sample rate: try session.setCategory(.playAndRecord, mode: .default) try session.setActive(true) print("Hardware sample rate:", session.sampleRate) I consistently get: `Hardware sample rate: 48000.0 My questions are: Is the iPhone mic actually capturing at 8 kHz, or is it recording at 48 kHz and then downsampling to 8 kHz internally? Is there any way to force the hardware to record natively at 8 kHz? If not, what’s the recommended approach for telephony-quality audio (true 8 kHz) on iOS devices? Thanks in advance for your guidance!

Hi all, I have been quite stumped on this behavior for a little bit now, so thought it best to share here and see if someone more experience with AVAudioEngine / AVAudioSession can weigh in. Right now I have a AVAudioEngine that I am using to perform some voice chat with and give buffers to play. This works perfectly until route changes start to occur, which causes the AVAudioEngine to reset itself, which then causes all players attached to this engine to be stopped. Once a AVPlayerNode gets stopped due to this (but also any other time), all samples that were scheduled to be played then get purged. Where this becomes confusing for me is the completion handler gets called every time regardless of the sound actually being played. Is there a reliable way to know if a sample needs to be rescheduled after a player has been reset? I am not quite sure in my case what my observer of AVAudioEngineConfigurationChange needs to be doing, as this engine only handles output. All input is through a separate engine for simplicity. Currently I am storing a queue of samples as they get sent to the AVPlayerNode for playback, and after that completion checking if the player isPlaying or not. If it's playing I assume that the sound actually was played- and if not then I leave it in the queue and assume that an observer on the route change or the configuration change will realize there are samples in the queue and reset them Thanks for any feedback!