ABSTRACT
Speech consists of an alternating sequence of opening and closing gestures of the vocal tract. The closing gestures correspond to consonants, and the open phases correspond to vowels. Although (linguistic) distinctive features provide a ready explanation for many of the observed facts of speech perception, close inspection suggests that there may always be primitive acoustic variables that can provide even better explanations. When considering the classification of auditory signals, it is obviously appropriate to include the area in which human auditory-classification skills are most highly and widely developed: speech perception. A simple bottom-up approach would describe the recognition process as starting with determination of primitive auditory properties, which are then combined to yield classifications as phonemes, which in turn are combined to form syllables, words, phrases, and so on up. Identification of the temporal order of speech sounds is dependent on speech forming a coherent auditory stream.
