| 关键词检测系统中基于音素网格的置信度计算 |
| |
| 引用本文： | 张鹏远,韩疆,颜永红.关键词检测系统中基于音素网格的置信度计算[J].电子与信息学报,2007,29(9):2063-2066. |
| |
| 作者姓名： | 张鹏远 韩疆 颜永红 |
| |
| 作者单位： | 中国科学院声学研究所中科信利语音实验室,北京,100080 |
| |
| 基金项目： | 国家重点基础研究发展计划(973计划) |
| |
| 摘 要： | 该文提出了一种基于音素网格的置信度计算方法。与传统的基于整个声学模型的置信度不同的是，这种方法在解码器生成的音素网格上计算关键词的置信度，从而具有更好的拒识能力。另外，针对两种置信度取值范围的不同，该文采用权重因子的方法综合利用两种置信度，取得了较好的效果。在自然对话的电话数据测试中，与传统的置信度计算方式相比，混和置信度的FOM(Figure Of Merit)值相对提高了17.0%。
|
| 关 键 词： | 语音识别 关键词检测 置信度 后验概率 网格 |
| 文章编号： | 1009-5896（2007）07-2063-04 |
| 收稿时间： | 2006-2-22 |
| 修稿时间： | 2006-02-22 |
Phoneme Lattice Based Confidence Measures in Keyword Spotting |
| |
| Zhang Peng-yuan,Han Jiang,Yan Yong-hong.Phoneme Lattice Based Confidence Measures in Keyword Spotting[J].Journal of Electronics & Information Technology,2007,29(9):2063-2066. |
| |
| Authors: | Zhang Peng-yuan Han Jiang Yan Yong-hong |
| |
| Affiliation: | Zhongke Xinli Speech Lab, Institute of Acoustics, Chinese Academy of Sciences, Beijing 100080, China |
| |
| Abstract: | Phoneme lattice based Confidence Measure (CM) is proposed in this paper. It makes use of phoneme lattices generated by a phoneme recognizer. Acoustic Model (AM) based CM is also introduced. For a decoded speech frame aligned to an HMM state, the CM based on AM is calculated. These two confidence measures are combined using a weighting factor to obtain a hybrid CM as they had different dynamic scales. On spontaneous conversational telephone database, the Figure Of Merit (FOM) achieves 17.0% relative improvement comparing to AM based CM. |
| |
| Keywords: | Speech recognition Keyword spotting Confidence Measure (CM) Posterior probability Lattice |
| 本文献已被 维普 万方数据 等数据库收录！ |
| 点击此处可从《电子与信息学报》浏览原始摘要信息 |
|
点击此处可从《电子与信息学报》下载全文 |
