Please wait a minute...
 首页  期刊介绍 期刊订阅 联系我们 横山亮次奖 百年刊庆
 
最新录用  |  预出版  |  当期目录  |  过刊浏览  |  阅读排行  |  下载排行  |  引用排行  |  横山亮次奖  |  百年刊庆
清华大学学报(自然科学版)  2015, Vol. 55 Issue (5): 508-513    
  电子工程 本期目录 | 过刊浏览 | 高级检索 |
基于网格的语音关键词检索算法改进
肖熙, 王竞千
清华大学 电子工程系, 北京 100084
Improved lattice-based speech keyword spotting algorithm
XIAO Xi, WANG Jingqian
Department of Electronic Engineer, Tsinghua University, Beijing 100084, China
全文: PDF(995 KB)  
输出: BibTeX | EndNote (RIS)      
摘要 针对多候选汉语音节网格语音关键词检索任务,在Gauss混合模型以及多候选识别算法方面进行了研究改进。首先探讨了Gauss混合模型的不同简化策略并用实验进行了验证, 证明了全协方差矩阵在识别性能上的优越性; 随后对经典的多候选令牌传递算法做出了针对汉语特点的改进。实验表明这2方面的研究不仅提高了以音节作为输出的语音识别引擎的单候选识别效果, 也大幅提高了多候选的识别性能。最后搭建了一个基于多候选网格的语音关键词检索系统, 在该系统中验证了上述改进的效果。
服务
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章
肖熙
王竞千
关键词 语音关键词检索多候选网格Gauss混合模型CUDA三音子模型    
Abstract:An improved lattice-based speech keyword spotting system was developed from the Gaussian mixture model and an improved N-best speech recognition algorithm. First, tests were used to evaluate different simplified structures of Gaussian mixture models. Then, an N-best token passing algorithm was developed from the classic token passing algorithm using some unique pronunciation rules for the Chinese language. These two modifications improve the performance of both the 1-best and N-best speech recognition candidates. Finally, a key word spotting system was developed based on an N-best lattice to show the effectiveness of these improvements.
Key wordsspeech keyword spotting    multi-candidate lattice    Gaussian mixture model    compute unified device architecture (CUDA)    triphone model
收稿日期: 2015-03-05      出版日期: 2015-08-04
ZTFLH:  TP391.4  
引用本文:   
肖熙, 王竞千. 基于网格的语音关键词检索算法改进[J]. 清华大学学报(自然科学版), 2015, 55(5): 508-513.
XIAO Xi, WANG Jingqian. Improved lattice-based speech keyword spotting algorithm. Journal of Tsinghua University(Science and Technology), 2015, 55(5): 508-513.
链接本文:  
http://jst.tsinghuajournals.com/CN/  或  &nb