To get yourself started: start by reading this https://developer.apple.com/library/ios/samplecode/SpeakHere/Introduction/Intro.html
and then go for this library: https://code.google.com/p/improved-mistral/
This is a matlab library that does exactly what you intend: you can make a possible duplicate for iOS based on this library: https://github.com/codyaray/speaker-recognition