| 基于符号化和语言模型方法的汉语方言自动辨识 |
| |
| 引用本文： | 沈兆勇,顾明亮.基于符号化和语言模型方法的汉语方言自动辨识[J].徐州师范大学学报(自然科学版),2006,24(2):54-57. |
| |
| 作者姓名： | 沈兆勇 顾明亮 |
| |
| 作者单位： | 1. 徐州师范大学,语言研究所,江苏,徐州,221116
2. 徐州师范大学,物理系,江苏,徐州,221116 |
| |
| 基金项目： | 江苏省“十五”社科基金资助项目(K3-013)，江苏省高校自然科学基金资助项目(99KJB510002) |
| |
| 摘 要： | 在借鉴和改进了一种新的语种辨识方法———基于高斯混合模型(GMM)符号化和语言模型方法———的基础上,建立了一个汉语方言自动辨识系统.实验使用了一个多说话人、非特定文本、连续语音汉语方言语音库进行系统测试.在3种主要汉语方言的辨识中,15 s语料测试平均辨识率达到了90.7%.
|
| 关 键 词： | GMM符号化 语言模型 汉语方言自动辨识 |
| 文章编号： | 1007-6573(2006)02-0054-04 |
| 收稿时间： | 07 5 2005 12:00AM |
| 修稿时间： | 2005年7月5日 |
A Study About Chinese Dialect Identification Based on Tokenization and Language Model |
| |
| SHEN Zhao-yong,GU Ming-liang.A Study About Chinese Dialect Identification Based on Tokenization and Language Model[J].Journal of Xuzhou Normal University(Natural Science Edition),2006,24(2):54-57. |
| |
| Authors: | SHEN Zhao-yong GU Ming-liang |
| |
| Institution: | 1. Linguistics Institute, Xuzhou Normal University, Xuzhou, Jiangsu, 221116, China; 2. Department of Physics, Xuzhou Normal University, Xuzhou, Jiangsu, 221116, China |
| |
| Abstract: | In this paper,based on using for reference and impr ov ing the techniques recently developed for language identification,the Gaussian mixture model tokenization and language model,a system of automatic identificat ion for Chinese dialects has been established.This system was evaluated (using) a multi-speaker speech corpus of text-independent spontaneous speech data.An accuracy of 90.7% is achieved in three main Chinese dialects classification test using 15 s speech sessions. |
| |
| Keywords: | GMM tokenization language model Chinese dialect identification |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
