ISCA Archive Eurospeech 1999
ISCA Archive Eurospeech 1999

Voice conversion between UK and US accented English

Ching-Hsiang Ho, Saeed Vaseghi, Aimin Chen

This paper presents an HMM-based method and ex-perimental results for voice conversion between UK and US accented English. Phonetic-tree based tied-state triphone HMMs are used to map equivalent states of the source and target spectra. Then a linear transformation method is incorporated to estimate the most likely target spectra for a given input. The map-ping is between two different sets of phoneme i.e. the 44-phoneme UK English BEEP phone set and 39-phoneme US CMU phone set. Finally, a prosody ad-aptation is applied to tune the prosodic parameters. The experiments are based on voice conversion be-tween speakers speaking different unrestricted texts. Acoustic-phonetic mapping between two different ac-cents database enables us to attempt to deconstruct accents to investigate how they are distributed among different parameters such as spectra, energy contour, pitch, and duration.