Vocabulary distance matrix analysis – based reference template update technique

Abstract

Intra-reference and inter-reference distances can be used for evaluation and comparison of reference templates in template based speech recognition. Hence decision on quality of reference template can be done and the reference set can be updated if necessary. In this paper a new reference set update technique based on analysis of vocabulary distance matrix is proposed. Reference templates with small intra-reference and inter-reference distance values are substituted with candidate patterns giving bigger distance values. The substitution can be performed in supervised or unsupervised manner. The technique was tested for Dynamic time warping based isolated word recognizer. The reference substitution using proposed technique increased the intra-reference and inter-reference distances by 6.8 % and 13.6 % respectively.