Bilingual Voice Conversion by Weighted Frequency Warping Based on Formant Space

Voice conversion is a technique that transforms the source speaker's individuality to that of the target speaker. In this paper, we propose a simple and intuitive voice conversion algorithm that does not use training data between different languages, but uses text-to-speech generated speech rather than real recorded voices. The suggested method finds the transformed frequency by formant space warping. The formant space comprises four representative monophthongs for each language. The warping functions are represented by piecewise linear equations using pairs of four formants at matched monophthongs. Experimental results show the potential of the proposed method.