| 基于神经网络语音合成中的单词重音标注技术 |
| |
| 引用本文： | 冯华君.基于神经网络语音合成中的单词重音标注技术[J].浙江师范大学学报(自然科学版),2006,29(1):56-60. |
| |
| 作者姓名： | 冯华君 |
| |
| 作者单位： | 浙江师范大学,信息科学与工程学院,浙江省,金华,321004 |
| |
| 摘 要： | 目前多媒体技术要求机器对英语语音的合成接近人的自然声音.提出了用神经网络技术来标注英语单词内重音的新方法,讨论了输入-输出样本的编码转换问题,利用MATLAB仿真平台设计了神经网络结构和训练方法.实验证明,该方法是正确和有效的,并且具有更高的精度.
|
| 关 键 词： | 语音合成 重音标注 神经网络 编码 |
| 文章编号： | 1001-5051（2006）01-0056-05 |
| 收稿时间： | 2005-09-07 |
| 修稿时间： | 2005-10-24 |
Neural network based stress label for letter-to-sound of words |
| |
| FENG Huajun.Neural network based stress label for letter-to-sound of words[J].Journal of Zhejiang Normal University Natural Sciences,2006,29(1):56-60. |
| |
| Authors: | FENG Huajun |
| |
| Institution: | College of Information Science and Engineering, Zhejiang Normal University, Jinhua Zhejiang 321004, China |
| |
| Abstract: | A new approach based on neural network technique to label stress of English word was introduced.The encoding and transform of input-output samples were discussed.The structure and training method of neural network was designed by using MATLAB simulation software.The experiment demonstrated the correctness and effectiveness of the proposed strategy with high accuracy. |
| |
| Keywords: | MATLAB |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
