How to listen to mic input and analyse in real time?

Question

Hi unfortunately I've not been able to figure out audio on the iPhone. The best I've come close to are the AVAudioRecorder/Player classes and I know that they are no good fo audio processing.

So i'm wondering if someone would be able to explain to me how to "listen" to the iPhone's mic input in chunks of say 1024 samples, analyse the samples and do stuff. And just keep going like that until my app terminates or tells it to stop. I'm not looking to save any data, all I want is to analyse the data in real time and do stuff in real time with it.

I've attempted to try and understand apples "aurioTouch" example but it's just way too complicated for me to understand.

So can someone explain to me how I should go about this?

Solution 1

If you want to analyze audio input in real-time, it doesn't get a lot simpler than Apple's aurioTouch iOS sample app with source code (there is also a mirror site). You can google a bit more info on using the Audio Unit RemoteIO API for recording, but you'll still have to figure out the real-time analysis DSP portion.

The Audio Queue API is a slight bit simpler for getting input buffers of raw PCM audio data from the mic, but not much simpler, and it has a higher latency.

Added later: There's also a version of aurioTouch converted to Swift here: https://github.com/ooper-shlab/aurioTouch2.0-Swift

Solution 2

AVAudioPlayer/Recorder class won't take you there if you wanna do any real time audio processing. The Audio Toolbox and Audio Unit frameworks are the way to go. Check here for apple's audio programming guide to see which framework suits your need. And believe me, these low level stuff is not easy and is poorly documented. CocoaDev has some tutorials where you can find sample codes. Also, there is an audio DSP library DIRAC I recently discovered for tempo and pitch manipulation. I haven't looked into it much but you might find it useful.

Solution 3

If all you want is samples with a minimum amount of processing by the OS, you probably want the Audio Queue API; see Audio Queue Services Programming Guide.

AVAudioRecorder is designed for recording to a file, and AudioUnit is more for "pluggable" audio processing (and on the Mac side of things, AU Lab is actually pretty cool).