Digital Signal Sorting for Digitized Speech in an Environment of Common Digital Data Signals

摘要

This report discusses a digital signal sorting technique for identifying digitized speech signals in an environment of common digital data signals. The technique is modular in that two levels of recognition are possible: (1) a linear prediction based scheme which provides a fine-grain recognition; and (2) a digital word synchronization scheme which provides a coarse, or first cut recognition. In the fine-grain scheme, a discrimination measure defined herein is shown to provide good sorting of the digitized speech sequence - this is demonstrated through examples. (Author)