Speaker adaptation by hierarchical EigenVoice

We propose a novel speaker adaptation method, hierarchical EigenVoice (HEV). This method extends the eigenvoice through clustering the Gaussian components of HMMs into a hierarchical tree structure. It enables one to autonomously control a number of adaptation parameters (model complexity) depending on the amount of adaptation utterances from a new speaker. The experimental results of Japanese large vocabulary continuous speech recognition confirmed the significant performance increase in all range of the adaptation utterance amounts compared with the conventional speaker adaptation methods.