An Intelligent System for Spoken Term Detection That Uses Belief Combination

Spoken term detection (STD) can be considered a subpart of automatic speech recognition that aims to extract partial information from speech signals in the form of query utterances. A variety of STD techniques available in the literature employ a single source of evidence for query utterance match/mismatch determination. In this article, the authors develop an acoustic signal processing-based approach for STD that incorporates several techniques for silence removal, dynamic noise filtration, and evidence combination using the Dempster-Shafer theory. Spectral-temporal features-based voiced segment detection and energy and zero cross rate-based unvoiced segment detection remove silence segments in the speech signal. Comprehensive experiments have been performed on large speech datasets and satisfactory results have been achieved with the proposed approach, which improves existing speaker-dependent STD approaches, specifically the reliability of query utterance spotting, by combining the results from multiple belief sources.