Elastic spectral distortion for low resource speech recognition with deep neural networks | IEEE Conference Publication | IEEE Xplore