Design and Implementation of Voice Conversion System Based on GMM and ANN

Voice conversion has become one current researching hotspot in speech signal processing. The article designs and establishes a voice conversion system based on Gaussian mixture model (GMM) and Artificial Neural Network (ANN) after researching the existing voice conversion algorithms. The core is to obtain respectively three mapping rules by training spectral envelope and its residual with GMM, and pitch contrail with BP network. And the voice is transformed according to the three mapping rules above. Finally, the algorithm simulation, the system implementation and algorithm performance evaluating of voice conversion is completed. The theoretic analysis and simulating results reveal that the algorithm and the system of voice conversion are effective.

[1]  Peng Song,et al.  Improving the Performance of GMM Based Voice Conversion Method , 2008, 2008 IEEE Pacific-Asia Workshop on Computational Intelligence and Industrial Application.

[2]  Chung-Hsien Wu,et al.  Voice conversion using duration-embedded bi-HMMs for expressive speech synthesis , 2006, IEEE Transactions on Audio, Speech, and Language Processing.

[3]  Xiongwei Zhang,et al.  Voice Conversion based on GMM and Artificial Neural Network , 2010, 2010 IEEE 12th International Conference on Communication Technology.

[4]  Qiang Huo,et al.  An Environment-Compensated Minimum Classification Error Training Approach Based on Stochastic Vector Mapping , 2006, IEEE Transactions on Audio, Speech, and Language Processing.

[5]  Ling-hua Zhang,et al.  Research on voice conversion based codebook and GMM , 2010, 2010 IEEE 12th International Conference on Communication Technology.

[6]  Elliot Moore,et al.  Voice conversion with linear prediction residual estimaton , 2008, 2008 IEEE International Conference on Acoustics, Speech and Signal Processing.