Sparse Auditory Reproducing Kernel (SPARK) Features for Noise-Robust Speech Recognition | IEEE Journals & Magazine | IEEE Xplore