Voice Conversion Adopting SOLAFS

An improved method of voice conversion is proposed to make the speech of a source speaker sound like uttered by a target speaker. Speaker individuality transformation is achieved by altering the spectral envelope and prosodic information. The main advantage of this method is to firstly apply the synchronized overlap-add fixed synthesis (SOLAFS) to modify the source speaker's speaking rate to match that of the target speaker, which enhances the performance of the whole conversion system compared with conventional systems without such a procedure. Besides, a precise estimation for the target excitation is advanced only with the information of the matched source's excitation and the average pitch period of the target speaker. The proposed scheme is evaluated using both subjective and objective measures. The experimental results show that the system is capable of effectively transforming speaker identity whilst the converted speech maintains high quality.