| 大词表孤立词语音识别的快速搜索算法 |
| |
| 引用本文： | 梁维谦,原道德,丁玉国.大词表孤立词语音识别的快速搜索算法[J].清华大学学报(自然科学版),2011(1):101-104,110. |
| |
| 作者姓名： | 梁维谦 原道德 丁玉国 |
| |
| 作者单位： | 清华大学电子工程系;清华大学微电子学研究所;北京凌声芯语音科技有限公司; |
| |
| 基金项目： | 国家“八六三”高技术项目(2008AA010700) |
| |
| 摘 要： | 在大词表孤立词语音识别中,Viterbi搜索是时间消耗的主要因素。为改善基线系统性能,根据汉语孤立词识别的特点,提出了一种基于音节切分的束搜索算法,在音节层和词条层进行剪枝。该算法不增加内存开销。实验结果表明:在词表规模为10 000时,该算法以0.23%的识别率下降率为代价,将Viterbi搜索的时间消耗降低为基线系统的26.73%;相对于小词表,该算法在大词表情况下对系统性能的改善尤为明显。
|
| 关 键 词： | 语音识别 音节切分 束搜索 |
Fast search algorithm for large vocabulary isolated-word speech recognition |
| |
| LIANG Weiqian,YUAN Daode,DING Yuguo.Fast search algorithm for large vocabulary isolated-word speech recognition[J].Journal of Tsinghua University(Science and Technology),2011(1):101-104,110. |
| |
| Authors: | LIANG Weiqian YUAN Daode DING Yuguo |
| |
| Institution: | LIANG Weiqian1,YUAN Daode2,DING Yuguo3(1.Department of Electronic Engineering,Tsinghua University,Beijing 100084,China,2.Institute of Microelectronics of Tsinghua University,3.Beijing VoiceOn Speech Corporation,Beijing 100085,China) |
| |
| Abstract: | In large vocabulary isolated word speech recognition,the Viterbi search is the major time consumes.The performance of the baseline system is improved using the characteristics of isolated Mandarin words in a syllable detection based beam search algorithm with pruning of syllable and word levels.This algorithm has no additional memory cost.Tests show that for vocabulary with 10 000 words,the algorithm reduces the time consumption to 26.73% of the baseline system,while the recognition rate is reduced by only ... |
| |
| Keywords: | speech recognition syllable detection beam search |
| 本文献已被 CNKI 等数据库收录！ |
|
