Acoustic Cues in Natural Speech: Their Nature and Potential Uses in Speech Recognition.

摘要

This final report summarizes the work carried out to study those acoustic cues in natural speech that are of potential use for speech-recognition purposes. The principal objectives of the program were: (a) to carry out basic research on automatically identifying the acoustic cues of natural speech. This research was aimed to bear directly on the scientific problems encountered in building a Speech Understanding System. (b) To work closely with the principal contractors in the program whose tasks were to build complete speech-understanding systems. It was expected that this close cooperation would allow the research results to be quickly incorporated in these systems and their usefulness to be readily evaluated.