theamazingaudioengine / theamazingaudioengine Goto Github PK

Many times we have to debug functions in our code that return an OSStatus, such as AudioUnitSetParameter(). Many classes of TAAE contain a very useful checkResult function which does that, for example, in AEMixerBuffer.m:

checkResult(AudioUnitSetParameter(_mixerUnit, kMultiChannelMixerParam_Pan, kAudioUnitScope_Input, index, value, 0),
            "AudioUnitSetParameter(kMultiChannelMixerParam_Pan)");

It would be useful if checkResult would be available whenever one imports classes from TAAE!

Is there a way TAAE can be used to manage audio from videos played with AVPlayer?

Hi

I had to modify the library to re enable remote control features to work. Am I misunderstanding something?

https://github.com/TheAmazingAudioEngine/TheAmazingAudioEngine/blob/master/TheAmazingAudioEngine/AEAudioController.m#L1657

UInt32 allowMixing = YES; // this disables remote control features

Thanks for the library! Love it!

We have just migrated to Amazing Audio Engine and released a new version of our app on TestFlight. After release our beta testers report that the audio is recorded at a much higher sample rate. The current workaround we have is for the user to reinstall the app or restart the phone, after which the recordings will be made in the right sample rate.

I can't find the exact way to reproduce the bug. After the workaround the bug cannot be reproduced on same device again, even when I tried to remove, reinstall or upgrade from older version.

After looking into the source code, I found the line of code that I suspect is causing the bug. Following is the code snippet from AEAudioController.m:

- (BOOL)initAudioSession {
    NSMutableString *extraInfo = [NSMutableString string];

    ..... // code omitted

    // Set sample rate
    Float64 sampleRate = _audioDescription.mSampleRate;
    result = AudioSessionSetProperty(kAudioSessionProperty_PreferredHardwareSampleRate, sizeof(sampleRate), &sampleRate);
    checkResult(result, "AudioSessionSetProperty(kAudioSessionProperty_PreferredHardwareSampleRate)");

    // Fetch sample rate, in case we didn't get quite what we requested
    Float64 achievedSampleRate;
    UInt32 size = sizeof(achievedSampleRate);
    result = AudioSessionGetProperty(kAudioSessionProperty_CurrentHardwareSampleRate, &size, &achievedSampleRate);
    checkResult(result, "AudioSessionGetProperty(kAudioSessionProperty_CurrentHardwareSampleRate)");
    if ( achievedSampleRate != sampleRate ) {
        NSLog(@"Warning: Delivered sample rate is %f", achievedSampleRate);
        _audioDescription.mSampleRate = achievedSampleRate;  // <- Problem here
        [extraInfo appendFormat:@", sample rate %f", achievedSampleRate];
    }
}

At line 2046 AEAudioController sets the ASBD sample rate to the current hardware rate if it failed to change the preferred hardware sample rate! I don't think this is to be expected as AEAudioController should record in the sample rate that the user requested. From my understanding Core Audio would follow the sample rate we requested in the ASBD even when it is different from the hardware sample rate. (albeit with some performance penalty)

Can anyone verify that this is indeed causing the bug I described?

(P.S. Just want to thank you @michaeltyson for creating The Amazing Audio Engine, it is really amazing! :)

Regards,

Soares

Copied and pasted from TAAE forum:

I'm finding what I think is a memory leak caused by AEAudioController.

In my app, I have a custom audio file player that is modeled closely after AEAudioFilePlayer, and I add the channel to AEAudioController when I load a song. channelIsPlaying is set to NO initially, then it is set to YES when I want the file to start playing. The song is stopped by setting channelIsPlaying to NO, and the channel is removed from AEAudioController when I unload the song.

I'm finding that the dealloc function of my audio file player is not being called when I have loaded a song, played a portion of the file, and then unload the song. However it IS called if I simply load a song and unload it without playing. This leads me to believe that AEAudioController must be adding an extra retain to the channel when it is played (when channelIsPlaying is set to YES). I have scoured my code and don't see anything that would cause extra retains when a channel is played.

I should note that I have not updated my version of TheAmazingAudioEngine for a few months.

Any idea what's going on? Thanks.

Hey,

Found this critical issue today. Currently I'm trying to add iOS7 support to my app but faced with crash in AEAudioController:1725 called from AEAudioController: 355. EXC_BAD_ACCESS (code=1).

This crash could be reproduced on device only (I used iPhone 4S). Simulator works fine. Additionally on app start it asks user to grant an audio recording permissions.

Hi,

I tried to run Xcode analyzer tool and it came out with 22 issues. Some of them makes me really nervous and prevents me to stop use this really amazing audio engine in production app. Is is possible for you to run through this issues and at least reduce the number of critical ones?

Thanks.

I have a setup with one channel and one filter.
Trying to add output level metering shows the level from the channel, not the main audio coming out from the engine (through the filter).

UPDATE: This is when adding the filter as an output filter. Adding it as an input filter works fine, it seems. Perhaps output level metering reads the signal at the main mixer output instead of at the output of the last output filter?

Line 209: scratchBufferList->mBuffers[i].mDataByteSize = MIN(16384, (fileLengthInFrames-readFrames) * _targetAudioDescription.mBytesPerFrame);

Should "16384" be replaced by "4 * kIncrementalLoadBufferSize" ?

It seems that since channelIsMuted and channelIsPlaying are used internally by TAAE, an AEAudioPlayable doesn't actually need to implement anything to have these properties working, only declare them as readwriteable.

Why not change this in the AEAudioPlayable protocol? Otherwise it breaks polymorphism in cases like this:

id<AEAudioPlayable> someChannel;
...
someChannel.channelIsMuted = YES;

Not all audio formats are supported by the MultiChannelMixer - in these cases, a converter node needs to be added to convert from the channel's output format to the mixer's input format.

how can I avoid this?

It would be nice with a averagePowerLevel:peakHoldLevel:forChannel: method.

Totally minor, but should be fixed:

Line 80: * @param completionBlock The block to invoke upon completion or error

There is no completionBlock parameter in the initWithFileURL:targetAudioDescription: method.

See aso:

http://forum.theamazingaudioengine.com/discussion/comment/374#Comment_374
http://forum.theamazingaudioengine.com/discussion/comment/502#Comment_502

Anyway i can make the AEAudioFilePlayer play only a portion of the track (for eg. from 0:44 to 0:55) and loop it. with an option not to loop ?

such as pause record,and pause play?

IOKitUser-920.1.11/hid.subproj/IOHIDEventQueue.c, line: 512

2013-09-23 15:02:08.531 NORDIC[9767:60b] TAAE: Setting audio session category to PlayAndRecord

AEAudioController.m
if ( _topChannel->audiobusOutputPort && ABOutputPortGetConnectedPortAttributes(_topChannel->audiobusOutputPort) & ABInputPortAttributePlaysLiveAudio ) {
return @"Audiobus";
} else {
return _audioRoute;
}

Hi Michael - do you have any plans to switch over to the newer AVAudioSession now that AudioSession is deprecated in iOS7? Also, do you have any knowledge of when the AudioSession API is slated to be removed for good? I imagine this will impact a lot of developers.

Suggestion: The ability to set the input/output volume of the AudioUnit when created (or later) - this may help control effects that need more headroom than is available in the source audio.

Suggestion: The ability to microfade the input to the AudioUnit when you addFilter/removeFilter. In fact, I think the option of a microfade on both of these methods may prove quite useful as there are many fx that simply work better when they gradually receive a signal as opposed to all of sudden get blasted with samples.

Hi Michael,

I have just started looking at TAAE and so far I am loving it.
There's just one thing I need that it doesn't appear to support, the AUSplitter Audio Unit.
Would it be possible to get support for this added?

Also, I'd like to be able to set a specific channel or group as the source for the IOUnits output, so that not everything is output to the speakers.

Thanks.

I'm attempting to record input from the built-in mic and mix the result with my app's audio output (which is a combination of audio files loaded into a channel group).

For recording, I'm using the example code here: http://theamazingaudioengine.com/doc/_receiving-_audio.html#Recording

However, after I stop recording, the resulting audio file is twice the speed of the original audio and has some pops and clicks.

If I record without using addOutputReceiver then the audio is recorded correctly.

I've also tried changing the audio format that I'm recording to with no luck.

Any ideas of what's going on?

I am facing a problem while playing the Metronome using the audio channel and recording the audio which should not include the metronome sound in the recording. For no reason the recorded audio has lags and silence in between it. Even if AVAudioPlayer is used the same problem prevails. If any has faced this problem or knows the solution for it then please let me know.

While my app is in the background, regardless if the audio engine is running or not, plugging in the headphones gives me:

TAAE: Changed audio route to HeadphonesAndMicrophone
AEAudioController.m:2580: AudioConverterNew result 1718449215 666D743F fmt?
AEAudioController.m:2581: AudioConverterSetProperty(kAudioConverterChannelMap result -50 FFFFFFCE ˇˇˇŒ
/ABInputPort.m:243: AudioConverterNew result 1718449215 666D743F ?tmf
/ABInputPort.m:245: AudioConverterNew result 1718449215 666D743F ?tmf
TAAE: Input status updated (0 channel, non-interleaved)

Then disconnecting them again, nothing happens. Bringing the app to the foreground then shows:

TAAE: Starting Engine
TAAE: Input status updated (1 channel, non-interleaved, with converter)
ERROR:     [0x3d33d18c] 1207: AUIOClient_StartIO failed (-66628)
AEAudioController.m:924: AUGraphStart result -66628 FFFEFBBC ˇ˛˚º
TAAE: Trying to recover from system error (3 retries remain)
TAAE: Stopping Engine
TAAE: Input status updated (1 channel, non-interleaved, with converter)
TAAE: Engine setup
TAAE: Starting Engine
TAAE: Successfully recovered from system error
TAAE: Changed audio route to SpeakerAndMicrophone
TAAE: Changed audio route to SpeakerAndMicrophone

Same thing happens the other way around: plugging out the headphones while in background.

Is it possible to add audio file conversion support to your amazing audio engine? Something like this: https://github.com/michaeltyson/TPAACAudioConverter

Thanks.

It would be nice to be able to set an audioDescription on receivers and have it converted when needed. And/or expose an AECopyAndConvertBuffer(AudioBufferList *source, AudioStreamBasicDescription *sourceAudioDescription, AudioBufferList *destination, AudioStreamBasicDescription *destinationAudioDescription).

Pressing the lock button causes audio to stop playing. Audio should continue to play if all AudioUnits have kAudioUnitProperty_MaximumFramesPerSlice set to 4096. It appears from the code that this is being done for some audio units, but apparently it's not set on all of them.

I’ve recently added TheAmazingAudioEngine to the CocoaPods package manager repo.

CocoaPods is a tool for managing dependencies for OSX and iOS Xcode projects and provides a central repository for iOS/OSX libraries. This makes adding libraries to a project and updating them extremely easy and it will help users to resolve dependencies of the libraries they use.

However, TheAmazingAudioEngine doesn't have any version tags. I’ve added the current HEAD as version 0.0.1, but a version tag will make dependency resolution much easier.

Semantic version tags (instead of plain commit hashes/revisions) allow for resolution of cross-dependencies.

In case you didn’t know this yet; you can tag the current HEAD as, for instance, version 1.0.0, like so:

$ git tag -a 1.0.0 -m "Tag release 1.0.0"
$ git push --tags

Hi guys,

Is it possible for you include several predefined audio filters into your project? Something like:
1.https://github.com/bartolsthoorn/NVDSP
2.https://github.com/alexbw/novocaine/blob/master/Novocaine%20iOS%20Example/ViewController.mm

Just to have some basic filters.

Really admire job you've done!

Thank you.

I created a small demo with 1 channel (AEAudioPlayer) and it seems it takes about 1 second for the AEAudioController to actually start the playback.
Is there any way i can speed it up ?
btw, i can't actually prepare it in advance because everything is real-time.

Apple's AU3DMixerEmbedded audio unit has a poorly documented (gasp!) need for mono input to work correctly, which the AEAudioUnitFilter setup-check misses.

Would a way to pass an explicit input format to a filter solve this issue? Or is the 3D mixer better off integrated as something other than a filter, somewhere else in the chain? (Perhaps as a replacement to the default multi-channel mixer?)

I considered the latter, but obviously the default mixer is pretty tightly integrated.

This is an odd edge case, but it might be a broader bug that should be fixed.

I've tried to use my own AEAudioFilter that implements a basic time-stretching algorithm (as the AUNewTimePitch audio unit is broken on iOS6) but I've run into an odd issue. Take the following code, which calls producer(producerToken, audio, &frames) when it needs to grab more frames. If the time stretching algorithm has enough frames already, producer does not get called.

static OSStatus filterCallback(id                               filter,
                               AEAudioController                *audioController,
                               AEAudioControllerFilterProducer  producer,
                               void                             *producerToken,
                               const AudioTimeStamp             *time,
                               UInt32                           frames,
                               AudioBufferList                  *audio)
{
    RXAudioStretchFilter *THIS = (RXAudioStretchFilter*)filter;

    while (THIS->soundTouchEngine.numSamples() < frames) {
        OSStatus status = producer(producerToken, audio, &frames);
        if (status != noErr) return status;

        for (int i = 0; i < frames; i++) {
            THIS->tmp[2*i] = ((SInt16*)audio->mBuffers[0].mData)[i];
            THIS->tmp[(2*i) + 1] = ((SInt16*)audio->mBuffers[1].mData)[i];
        }

        THIS->soundTouchEngine.putSamples(THIS->tmp, frames);
    }

    UInt32 samplesWritten = THIS->soundTouchEngine.receiveSamples(THIS->tmp, frames);

    // ... more code omitted ...
}

The issue arises when producer has not been called and time has passed outside the filterCallback. The next time producer is called, the mTimeStamp.mSampleTime it receives will be equal the current host sample time - and not the mSampleTime at the last time the function was called. This makes time-stretching impossible.

As a better example, imagine producer calls an AUAudioFilePlayer somewhere to play back an audio file. This audio file is exactly 5 seconds long. However, the AEAudioFilter I've implemented for time stretching wants to extend this clip to 10 seconds. Hence, on average, every other call to filterCallback will call producer. The expected behaviour is that the mSampleTime value passed to the audio unit will start at the current host time, but then only be incremented when producer is called - in essence, allowing the filter to fake the current time and fetch samples only as they're needed. The current behaviour instead sets mSampleTime to the current real sample time every time filterCallback, making the AUAudioFilePlayer play in real time, no matter what.

I've added a quick hack of a solution to my branch (at: psobot@cdd1398) but I'm not sure if this is something that TAAE should support more broadly.

There's an issue right now in the routine that removes channels that causes other channels to pause for a buffer or two sometimes.

TAAE stores the control structures for channels within an array of channel_t, and points each AURenderCallbackStruct's inputProcRefCon to the structure in that array associated with each channel.

The problem is due to the fact that when removing channels, channel structures that come later in the channel array for the given group are shuffled back on the Core Audio thread (to keep the channels contiguous in memory) prior to reassigning the inputProcRefCon pointers for the channels that remain, back on the main thread.

The interval between shuffling the elements around in the channel array, and reassigning the pointers for each channel may be large, and in that time the channels may render the wrong actual channel data, or may be silent. Due to the way that AUGraphs can only be updated on the main thread, this latter reconfiguration can't be done at the same time as reconfiguring the channel array.

There are two potential solutions:

1. Rather than shuffling the channels array so that the channels array is contiguous (as in, channel n can be found in element n of the array), instead blank out the element of the removed channel. When adding new channels, search for the first blank entry in the channel array. That means keeping track of which channel index is associated with each record in the channel array, but removes the requirement to sync the state of the channel array and the inputProcRefCon pointers when removing channels.
2. Rather than using a fixed-size channel array to store channel control data, allocate a new structure each time a channel is added, storing a pointer to the structure in a dictionary keyed by the channel object. On removing channels, perform the graph update first, and free the structure afterwards.

The second option seems preferable.

The error on AEMixerBuffer.m:1039 is returned yet everything is still functioning as expected with one exception...sometimes (in the simulator at least, have yet to test on device) AEMixerBuffer will repeatedly spit out...

<AEMixerBuffer 0xb0bac00>: Out of buffer space

...this is logged the entire time the AERecorder is attempting to record. If I set a breakpoint and stop temporarily during the init call the memory error is less likely to occur.

UPDATE: In LIMITED testing if I put a sleepForTimeInterval immediately after AERecorder's init the buffer error does NOT occur. Removing the sleep it occurs every time.

... [[AERecorder alloc] initWithAudioController:...];
[NSThread sleepForTimeInterval:1.0];

However, the following error is returned every time (at least in simulator):
AEMixerBuffer.m:1039: AudioUnitSetProperty(kAudioUnitProperty_StreamFormat) result -10868 FFFFD58C å’ˇˇ

Details of this follow:

Having setup an AEAudioController with a number of elements that work, when I try to setup an AERecorder I get the error shown above. The error occurs in the call to AERecorder initWithAudioController.

[[AERecorder alloc] initWithAudioController:

Specifically, audioController.inputAudioDescription is zero-ed out and therefore its mChannelsPerFrame doesn't match audioController.audioDescription which is

(AudioStreamBasicDescription) $1 = {
mSampleRate = 44100
mFormatID = 'lpcm'
mFormatFlags = 12
mBytesPerPacket = 4
mFramesPerPacket = 1
mBytesPerFrame = 4
mChannelsPerFrame = 2
mBitsPerChannel = 16
mReserved = 0
}

AEMixerBuffer's setAudioDescription: is then called to fix this. The audioDescription returned from the inputCallbacks is:

Printing description of THIS->_inputCallbacks->audioDescription:
(AudioStreamBasicDescription) audioDescription = {
mSampleRate = 0
mFormatID = 0
mFormatFlags = 0
mBytesPerPacket = 0
mFramesPerPacket = 0
mBytesPerFrame = 0
mChannelsPerFrame = 0
mBitsPerChannel = 0
mReserved = 0
}

Ultimately upon the call to AudioUnitSetProperty to set the mixerUnit's stream format the above audioDescription (empty) is being passed which returns the error.

The AERecorder is still returned and functions (seemingly) normally.

Add the ability to bypass a unit in the graph by implementing an on/off mechanism ie: a filter. This would eliminate the need to manually remove the unit and add it back in just to temporarily turn it off.

It looks like +[AEAudioFilePlayer audioFilePlayerWithURL:audioController:error:] loads an entire audio file from disk synchronously. In my tests, this method takes about 1.6s on an iPhone 5 for a 7MB AAC file. That's not counting the initialization of the AEAudioController, nor adding the player as a channel.

I see the implementation starts a loading operation on the current thread but blocks until completion. Does this mean you have it in mind to do asynchronous loading with a completion block? I'd really like to be able to do that.

Failing that, is it safe to create the audio file player on a low-priority thread and add it to the controller later on the main thread?

Commit 3bcb6f1 changed AEAudioController.m:440 so that &channel->timeStamp is passed to the AEAudioControllerRenderCallback instead of &arg->inTimeStamp.

But after mHostTime is incremented in renderCallback() at AEAudioController.m:477, it never gets copied to channel. Now when channelAudioProducer() calls the AEAudioControllerRenderCallback, host time remains static.

To see this in action in the Example Project, add this line inside of renderCallback() in AEAudioUnitChannel.m:155

printf("host %lld, sample %0.f \n", time->mHostTime, time->mSampleTime);

I'm not sure it is the right place for it, but adding this line right before arg is created at AEAudioController.m:484 appears to fix the problem:

channel->timeStamp.mHostTime = timestamp.mHostTime;

Scheduler triggers like it should adding this code;

self.scheduler = [[AEBlockScheduler alloc] initWithAudioController:_audioController];
[_audioController addTimingReceiver:_scheduler];

and this:

[self.scheduler scheduleBlock:^(const AudioTimeStamp *intervalStartTime, UInt32 offsetInFrames) {
    NSLog(@"Hit Me!");
} atTime:[AEBlockScheduler timestampWithSecondsFromNow:startTimeSecondsFromNow] timingContext:AEAudioTimingContextOutput identifier:@"clip1"];

But let's say I add this line in the schedule block to play a clip at specific point in time:

[self.audioController addChannels:[NSArray arrayWithObject:clip] toChannelGroup:self.mainChannelGroup];

I then get this error:

AEAudioController.m:1003: Update graph result -10863 FFFFD591 ˇˇ’ë
TAAE: Timed out while performing message exchange

Moving the addChannels line of code out of the scheduling block, plays of the audio clip like it should.

It would be nice with a way to share signals between channels, for example introduce an AEAuxBus, where one channel can push audio and other channels read it back. The audio on the bus buffer would contain one block of audio per rendering cycle, only replacing it when a channel (or filter or receiver) writes to it. So regarding the order of execution: channels reading the bus before the writers would get it delayed one buffer cycle, which is OK and expected. I haven't tested but I assume channels execute in the order they were added. Such a feature would allow one to implement sends to global effects, etc.

Hi there,
there seems to be an issue with the -[AEAudioController setMuted: forChannelGroup:] method. I was trying to mute/unmute a channel group and did not succeed until I patched it:

diff --git a/TheAmazingAudioEngine/AEAudioController.m b/TheAmazingAudioEngine/AEAudioController.m
index 1a040aa..2716852 100644
--- a/TheAmazingAudioEngine/AEAudioController.m
+++ b/TheAmazingAudioEngine/AEAudioController.m
@@ -1162,7 +1162,8 @@ static OSStatus topRenderNotifyCallback(void *inRefCon, AudioUnitRenderActionFla
     AEChannelGroupRef parentGroup = [self searchForGroupContainingChannelMatchingPtr:group userInfo:NULL index:&index];
     NSAssert(parentGroup != NULL, @"Channel not found");

-    AudioUnitParameterValue value = group->channel->muted = muted;
+    AudioUnitParameterValue value = !(group->channel->muted = muted);
+  
     OSStatus result = AudioUnitSetParameter(parentGroup->mixerAudioUnit, kMultiChannelMixerParam_Enable, kAudioUnitScope_Input, index, value, 0);
     checkResult(result, "AudioUnitSetParameter(kMultiChannelMixerParam_Enable)");
 }

The thing is that the property kMultiChannelMixerParam_Enable is the inverse of muted.

Regards,
Thierry

This is the 4th time I have initialized the audioController, but otherwise recording multiple times works okay. Specifically it deadlocks on the main thread here: https://github.com/TheAmazingAudioEngine/TheAmazingAudioEngine/blob/master/TheAmazingAudioEngine/AEAudioFileWriter.m#L220

- (void)finishWriting {
    if ( !_writing ) return;

    _writing = NO;

    // Gets stuck here
    checkResult(ExtAudioFileDispose(_audioFile), "AudioFileClose"); 

    if ( _priorMixOverrideValue ) {
        checkResult(AudioSessionSetProperty(kAudioSessionProperty_OverrideCategoryMixWithOthers, sizeof(_priorMixOverrideValue), &_priorMixOverrideValue),
                    "AudioSessionSetProperty(kAudioSessionProperty_OverrideCategoryMixWithOthers)");
    }
}

First I used TAAE to record an AAC file two times, then used the hardware video encoder to record a video, then used TAAE to record another file. I am using a fork of TAAE that "fixes" the crash mentioned here: #27 via this method: OpenWatch@eb92e78

Here is the end of my logs:

2013-05-16 12:12:26.522 OpenWatch[482:907] TAAE: Setting audio session category to PlayAndRecord
2013-05-16 12:12:27.215 OpenWatch[482:907] TAAE: Audio session initialized (input available, audio route 'SpeakerAndMicrophone')
2013-05-16 12:12:27.217 OpenWatch[482:907] TAAE: Input status updated (1 channel, non-interleaved)
2013-05-16 12:12:27.222 OpenWatch[482:907] TAAE: Engine setup
2013-05-16 12:12:27.224 OpenWatch[482:907] TAAE: Starting Engine
2013-05-16 12:12:27.623 OpenWatch[482:907] TAAE: Changed audio route to SpeakerAndMicrophone
2013-05-16 12:12:27.626 OpenWatch[482:907] TAAE: Changed audio route to SpeakerAndMicrophone
2013-05-16 12:12:27.629 OpenWatch[482:907] TAAE: Changed audio route to SpeakerAndMicrophone
2013-05-16 12:12:37.227 OpenWatch[482:907] 12:12:37.227 <com.apple.main-thread> AudioSessionGetProperty posting message to kill mediaserverd (51)
2013-05-16 12:12:38.241 OpenWatch[482:907] TAAE: Timed out while performing message exchange
2013-05-16 12:12:39.247 OpenWatch[482:907] TAAE: Timed out while performing message exchange

Great library, by the way!

I noticed that the interruptionListener is being called even after my recording activities are complete and when I switch out of the app.

https://github.com/TheAmazingAudioEngine/TheAmazingAudioEngine/blob/master/TheAmazingAudioEngine/AEAudioController.m#L303

Specifically it crashes in this area of that function:

        if ( THIS->_runningPriorToInterruption ) {
            [THIS stop];
        }

        [[NSNotificationCenter defaultCenter] postNotificationName:AEAudioControllerSessionInterruptionBeganNotification object:THIS];

Either on "stop" or on the postNotification. The THIS data that is coming in is basically just garbage data and is usually some sort of garbled string value. It seems that the interruptionListener isn't being cleared when I'm done recording. Perhaps I'm not stopping my recording properly.

Here's how I use it if that helps:
https://github.com/OpenWatch/OpenWatch-iOS/blob/master/OpenWatch/OWAudioRecordingViewController.m#L95

UPDATE 2: This was not a bug in TAAE. It was actually a hardware issue that TAAE help me solve.

UPDATE: The link is now correct.

The AEAudioPlayable protocol has an attribute named pan. If this is set to -1.0 or 1.0 at the time a channel is created then the channel will be on the full left or right output. However, if the value is set later while the app is running the expected behavior is not seen.

Here is the TAAE example that has been modified to demonstrate the issue.

http://ericjknapp.com/samples/EJKSample.zip

The app was modified to add a new slider in the table view and code was added to the ViewController.m file. All added code has been marked with a comment line that starts with // EJK: and has detailed explanations.

Either the code is not behaving as expected or I am not understanding how the pan property is supposed to work.

It would be useful with a predefined ASBD for non-interleaved float.

The problem i am facing is that , one the user completes the recording of the audio and if he play it and at the same time he starts recording another audio , then the playing audio is also merged in the present recorded audio.
I dont want the output from speaker to be merged in the recording. It has been more than a week with this issue. All suggestions are welcome.

Puneet

In master.

It might be because of this line:

s.source_files = 'TheAmazingAudioEngine/**/*.{h,m,c}', 'Modules/**/*.{h,m,c}'

Here is the Pods > Targets > Pods-TheAmazing... > Compile Sources

When setting audioController.audiobusInputPort to receive audio from Audiobus instead of the hardware input, inputAveragePowerLevel still meters the hardware input instead of the Audiobus port.

Hi guys, not sure if this issue been reported. I got latest last week, and on iOS7 if you press home and come back, no audio is heard anymore. Playback doesn't respond for new audio also.

It works fine running on iOS6.

How I handle it:
applicationWillResignActive {
[self.audioController stop];

applicationWillEnterForeground {
[self.audioController start:NULL ];

If I don't do those calls at all, the behaviour is even worse. The whole application becomes very unresponsive, and runs at like 1 fps.