摘要
陆空通话标准用语(英语)的训练是空管模拟训练中的重要内容。对空管模拟训练中的语音指令自动识别问题进行了分析研究,包括:陆空通话标准用语基本特征的分析、语言模型的文法描述、指令特殊发音的识别处理、识别后处理方法以及声学模型训练方法等,并基于Sphinx-4语音识别引擎,设计实现了一个语音指令识别系统AIRS(ATC Instruction Recognizer System)。系统实验数据分析表明,声学模型训练后的语音识别正确率可以达到空管模拟训练的需求。
Training of air traffic control（ATC） communication in English is the main content of ATC simulator trai- ning. This paper focused on the voice instruction recognition in ATC simulator training which includes the analysis of the basic characteristics of ATC communication, the grammar description for language model, the recognition processing in special pronunciation of instruction, the processing method after recognition and the way of acoustic model adapting. Based on the excellent speech recognition engine Sphinx-4, we designed and implemented a voice instruction recognition system, AIRS （ATC Instruction Recognition System）. The experimental results show that the accuracy rate of voice in- struction recognition after the acoustic model adapting can reach the demand for ATC simulator training.
出处
《计算机科学》
CSCD
北大核心
2013年第7期131-137,共7页
Computer Science
关键词
语音识别
ATC指令识别
Sphinx-4
陆空通话
Speech recognition, ATC instruction recognition, Sphinx-4, Air traffic control communication
