We propose a voice conversion method based on sparse convolutive non-negative matrix factorization. The method utilizes the Itakura-Saito distance as the objective cost function, making the smaller matrix element with a smaller reconstruction error due to the property of scale invariant of the cost function. The time-frequency basis of the source and target were extracted during the training phase, and the speech is converted through time-frequency basis substitution. The eifect of whisper-to-normal speech conversion experiment is also conducted. Experimental results show that the proposed voice conversion method outperforms the method based on the conventional convolutive non-negative matrix factorization and the method based on the Kullback-Leibler (K-L) cost function in the aspects of speech intelligibility.
展开▼
