EUSIPCO 2015, 23rd European Signal Processing Conference, August 31st-September 4th, 2015, Nice, France
Phone adaptive training (PAT) aims to derive a new acoustic feature space in which the influence of phone variation is minimised while that of speaker variation is maximised. Originally proposed in the context of speaker diarization, our most recent work showed the utility of PAT in short-duration, automatic speaker verification where phone variation typically degrades performance. New to this contribution is the assessment of PAT utilising automatically generated acoustic class transcriptions whose number is controlled by regression tree analysis. Experimental results using a standard database show that PAT delivers significant improvements in the performance of a state-of-the-art iVector speaker verification system.
© 2015 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.