Automatic speech recognition experiments show that, depending on the task performed and how speech variability is modeled, automatic speech recognizers are more or less sensitive to the Lombard reflex. To gain an understanding about the Lombard effect with the prospect of improving performance of automatic speech recognizers, (1) an analysis was made of the acoustic-phonetic changes occurring in Lombard speech, and (2) the influence of the Lombard effect on speech perception was studied. Both acoustic and perceptual analyses suggest that the influence of the Lombard effect on male and female speakers is different. The analyses also bring to light that, even if some tendencies across speakers can be observed consistently, the Lombard reflex is highly variable from speaker to speaker. Based on the results of the acoustic and perceptual studies, some ways of dealing with Lombard speech variability in automatic speech recognition are also discussed.
This paper describes a new model-based speaker adaptation algorithm called the eigenvoice approach. The approach constrains the adapted model to be a linear combination of a small number of basis vectors obtained offline from a set of reference speakers, and thus greatly reduces the number of free parameters to be estimated from adaptation data. These "eigenvoice" basis vectors are orthogonal to each other and guaranteed to represent the most important components of variation between the reference speakers. Experimental results for a small-vocabulary task (letter recognition) given in the paper show that the approach yields major improvements in performance for tiny amounts of adaptation data. For instance, we obtained 16% relative improvement in error rate with one letter of supervised adaptation data, and 26% relative improvement with four letters of supervised adaptation data. After a comparison of the eigenvoice approach with other speaker adaptation algorithms, the paper concludes with a discussion of future work.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.