My only question is why especially since its anonymized - what benefit is there to Apple?
Improving voice recognition algorithms.
Having a large, growing pool of random audio samples from actual users in all their voices and languages and accents to analyze, over and over again as the algorithm is improved, is a huge part of how Siri improves.
Apple doesn't need to know who you are, or care what you're saying, but they and everyone else in voice recognition, have a long way to go improving this functionality yet.