Voice conversion for non-parallel datasets using dynamic kernel partial least squares regression
Voice conversion for non-parallel datasets using dynamic kernel partial least squares regression
复制标题
使用动态核偏最小二乘回归对非并行数据集进行语音转换
DOI：
10.21437/interspeech.2013-103
复制
发表时间：
2013
期刊：
影响因子：
--
通讯作者：
M. Gabbouj
中科院分区：
文献类型：
--
作者：
Hanna Silén;J. Nurminen;E. Helander;M. Gabbouj
语音转换旨在将一个说话者的语音转换为另一特定说话者所说的声音。基于高斯混合建模的最流行的语音转换方法往往会受到模型过度拟合或过度平滑的影响。为了克服传统方法的缺点，我们最近提出在并行数据语音转换框架中使用动态核偏最小二乘（DKPLS）回归。然而，并不总是能保证来自源和目标说话者的并行训练数据的可用性。在本文中，我们将基于 DKPLS 的非并行数据转换方法与著名的 INCA 对齐算法相结合来扩展。聆听测试结果表明，所提出的组合可以实现高质量的转换。此外，还使用语言内和跨语言数据评估了 INCA 两种变体的性能。
Voice conversion aims at converting speech from one speaker to sound as if it was spoken by another specific speaker. The most popular voice conversion approach based on Gaussian mixture modeling tends to suffer either from model overfitting or oversmoothing. To overcome the shortcomings of the traditional approach, we recently proposed to use dynamic kernel partial least squares (DKPLS) regression in the framework of parallel-data voice conversion. However, the availability of parallel training data from both the source and target speaker is not always guaranteed. In this paper, we extend the DKPLS-based conversion approach for non-parallel data by combining it with a well-known INCA alignment algorithm. The listening test results indicate that high-quality conversion can be achieved with the proposed combination. Furthermore, the performance of two variations of INCA are evaluated with both intra-lingual and cross-lingual data.
(2008) 历史和地理过程导致介壳虫的多样化，这些介壳虫与蚂蚁植物马卡兰加 (Macaranga) 共生。
DOI：
--
发表时间：
2008
期刊：
影响因子：
--
作者：
上田昇平・Quek;S.-P.・市岡孝朗・村瀬香・Kondo;T・Gullan;P. J.・市野隆雄
通讯作者：
P. J.・市野隆雄