中文摘要：

为了在语音转换过程中充分考虑语音的帧问相关性，提出了一种基于卷积非负矩阵分解的语音转换方法。卷积非负矩阵分解得到的时频基可较好地保存语音信号中的个人特征信息及帧间相关性。利用这一特性，在训练阶段，通过卷积非负矩阵分解从训练数据中提取源说话人和目标说话人相匹配的时频基。在转换阶段，通过时频基替换实现对源说话人语音的转换。相对于传统方法，本方法能够更好地保存和转换语音帧间相关性。实验仿真及主、客观评价结果表明，与基于高斯混合模型、状态空间模型的语音转换方法相比，该方法具有更好的转换语音质量和转换相似度。

英文摘要：

In order to fully consider the inter-frame correlation in voice conversion, a voice con- version method based on convolutive nonnegative matrix factorization is proposed. The person- al characteristics and inter-frame correlation in voice can be well preserved in the time-frequen-cy bases obtained from convolutive nonnegative matrix factorization. With this feature, during the training phase of voice conversion, the matching time-frequency bases of source and target speakers can be extracted from training data through convolutive nonnegative matrix factoriza-tion. Then in the conversion phase, the voice of source speaker is converted through time-fre-quency bases substitution. Compared with traditional methods, the inter-frame correlation in voice can be better preserved and converted in the proposed method. Experimental results using objective and subjective evaluations show that the proposed method outperforms the methods based on Gaussian mixture model and the state space model in the view of both speech quality and conversion similarity to the target speech.