単語セットの音声認識難易度推定 : 記号領域における単語間距離計算の応用

概要

論文の詳細を見る
This paper presents a between-word distance(BWD)calculation in a symbolic domain and discusses its typical application in estimating the degree of the speech recognition difficulty for given word sets. The first part of the paper describes a method for the distance calculation which employs matching by DP(dynamic programming)on subphonemic segment sequences to take phonemic-context-dependent characteristics into account. In order to test the usefulness of the method, two types of word sets are composed using a distance-based clustering technique. Vocabularies of one type have dense sample distributions while the others have sparse sample distributions in a BWD sense. Speaker-independent word recognition is examined for these word sets using a common phone-HMM-based speech recognition technique. We compare the recognition results and the statistical characteristics of individual word sets, and present criteria for relative order of the recognition difficulty of given word sets. One criterion using between-word distance distributions of n-nearest neighbor words provides a reasonable index for the recognition diffculty.
社団法人日本音響学会の論文