Voice conversion using Gaussian Mixture Models

Voice conversion is an emergent problem in voice and speech processing with increasing commercial interest. The main aim of the voice conversion system is to modify the speaker specific characteristics with respect to target specific characteristics. The vocal tract transfer function,shape of the glottal pulse and the prosodic features uniquely characterize a particular speaker. This work present a method of extracting features of vocal tract level i.e. Line Spectrum Frequencies (LSFs) using Regressions Model and the source characteristics level feature i.e. LP Residual is mapped using Gaussian Mixture Models for improving the performance of voice conversion system. The Performance of Gaussian Based Models based voice conversion system is conducted using objective measures and subjective measures.