Using VoiceProcessingIO along with AUSampler

Question

I'm trying to implement something similar to flute that when you blow the mic it's going to start play one note from the speaker. For this reason I was trying to use VoiceProcessingIO to substract playbacked note (output speaker) from input mic.

For some reason I noticed VoiceProcessingIO doesn't work together with AUSampler. I modified Apple Sampled "LoadPresetDemo" and just changed line:

cd.componentSubType = kAudioUnitSubType_RemoteIO 

with

cd.componentSubType = kAudioUnitSubType_VoiceProcessingIO;

and then you cannot hear anything played

I had the same effect (no sound) when I used the following scheme of audiounits:

VoiceProcessingIO (input mic) --> processing function (if audio level > thrs 
                                              then feed MusicDeviceMIDIEvent)

AUSampler --> VoiceProcessingIO (output speaker)

However when I used the scheme below:

VoiceProcessingIO (input mic) --> processing function (if audio level > thrs 
                                              then feed MusicDeviceMIDIEvent)

AUSampler --> RemoteIO (output speaker)

The output speaker volume was much lowered. The issue It seams is related to this: Is it safe to use two audio units for simultaneous I/O in iOS?

I've tested also example from here: https://code.google.com/p/ios-coreaudio-example/downloads/detail?name=Aruts.zip&can=2&q=

And it works with the following scheme without AUSampler:

VoiceProcessingIO (input mic) --> processing function --> VoiceProcessingIO (output speaker)

Questions is:

Is there any way to use VoiceProcessingIO along with AUSampler? Or is the only way to feed some data to VoiceProcessing output via render callback?