Robust speaker recognition through acoustic array processing and spectral normalization