Using an underspecified ASR system as an indicator for phonetic similarity

This paper presents a novel approach to the identification of phonetic similarity using properties observed during the speech recognition process. An experiment is presented whereby specific phones are removed during the training phase of a statistical speech recognition system so that the behaviour of the system can be analysed to see which alternative phone is selected. The domain of the analysis is restricted to specific contexts and the alternatively recognised (or substituted) phones are analysed with respect to a number of factors namely, the common phonetic properties, the phonetic neighbourhood and the frequency of occurrence in the complete corpus. The results indicate that a measure of phonetic similarity based on alternatively recognised observed properties can be predicted based on a combination of these factors and as such can serve as an important additional source of information for the purposes of pronunciation variation in speech recognition.