Learning Representations for Nonspeech Audio Events Through Their Similarities to Speech Patterns | IEEE Journals & Magazine | IEEE Xplore