Open Radar

 
Our apps at JoyTunes listen to an acoustic piano using the iPad’s/iPhone’s microphone while playing background music through the device’s speakers.

Since the background music we play is picked up by the device’s microphone, our challenge of recognizing the user’s acoustic piano and differentiating it from the sounds of the background music becomes harder.

The options of setting stuff like kAUVoiceIOProperty_VoiceProcessingQuality are interesting as they really help in canceling the played background music. However - they also cancel some (if not most) of the piano playing itself, so we can’t use it “as is” for our purposes.

It would really help us though if we could have access to 2 streams in parallel in our apps - both before the voice processing echo cancellation is applied, and after it.
Then we would have more information for our own background music cancellation algorithms.

Alternatively, an option to have echo cancellation that isn’t optimized only for voice, will also be very handy!

Attached is a project proving that you can’t sample from microphone twice both with and without voice processing enabled.