Soft feature decoding in a distributed automatic speech recognition system for use over wireless channels

摘要: A method and apparatus for performing automatic speech recognition (ASR) in a distributed ASR system use over wireless channel takes advantage of probabilistic information concerning the likelihood that given, portion data has been accurately decoded to particular value. The probability error each feature transmitted set is employed improve performance under adverse conditions. Bit probabilities bits which are used encode given compute confidence level may have value feature. Features corrupted with high advantageously either not or weighted less acoustic distance computation performed by recognizer. This novel approach decoding features referred herein as “soft decoding,” produces dramatic improvements certain