| 汉语语音识别中语言模型的并行优化 |
| |
| 引用本文： | 金玮,孙甲松.汉语语音识别中语言模型的并行优化[J].电声技术,2010,34(8):49-52. |
| |
| 作者姓名： | 金玮 孙甲松 |
| |
| 作者单位： | 清华大学,电子工程系,北京,100084 |
| |
| 摘 要： | 研究了汉语语音识别中语言模型的并行优化问题，在已有汉语语音识别有调拼音模型和多核体系结构的基础上，对其算法进行了并行优化，并优化了查询机制和前后向算法。试验表明，该方法取得了很好的效果，识别速度有了很大改进。
|
| 关 键 词： | 关键词语音识别 N元有调拼音文法模型 并行优化 |
Parallel Optimization Method in Language Model for Mandarin Speech Recognition |
| |
| JIN Wei,SUN Jiasong.Parallel Optimization Method in Language Model for Mandarin Speech Recognition[J].Audio Engineering,2010,34(8):49-52. |
| |
| Authors: | JIN Wei SUN Jiasong |
| |
| Institution: | JIN Wei,SUN Jiasong(Dept.of Electronic Engineering,Tsinghua University,Beijing 100084,China) |
| |
| Abstract: | In order to improve performance and throughput of speech recognition system,parallel optimization method in language model is proposed based on mandarin N-gram model and multi-core systems.The searching algorithm is optimized,and the forward-algorithm and backward-algorithm are parallelized on different CPU-cores.The ideal properties of the optimized speech recognition system provide users with high speed and throughput.It makes real-time recognition much easier. |
| |
| Keywords: | speech recognition N-gram model for Pinyin with tone parallel optimization |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
