Segmentation of acoustic events using time encoded speech (TES) descriptors