智能飞机牵引机器人语音控制研究  

Research on voice control of intelligent aircraft towing robot

作　　者：张攀 房体会[1] 张威 韩明明[3] 解润海 Zhang Pan;Fang Tihui;Zhang Wei;Han Mingming;Xie Runhai

机构地区：[1]中国民航大学航空工程学院,天津300300 [2]天津大学机械工程学院,天津300072 [3]青岛新南山航空培训有限公司,青岛266070 [4]天津市天大精益科技有限公司,天津300070

出　　处：《起重运输机械》2019年第7期104-108,共5页Hoisting and Conveying Machinery

基　　金：国家自然科学民航联合基金项目(U1533103);天津市企业特派员项目(18JCTPJC68800);中央高校基本科研业务费项目(3122017025)

摘　　要：为研究智能飞机牵引机器人语音控制,实现机场环境下牵引机器人对飞行员语音指令的精确、高效识别,同时针对飞行员语音指令民航针对性差的问题,提出一种基于民航针对性特色语音库的智能飞机牵引机器人语音控制方法。首先,进行了智能飞机牵引机器人语音控制的总体功能设计;然后,在陆空通话标准的基础上分析了机载语音指令的声学特性,并结合机场环境和语料库建立原则分析建立了牵引机器人指令特色语音库。基于孤立词识别的实验结果显示,所建立的牵引车指令特色语音库与特色测试集的语音识别结果相对于特色语音库与普通测试集的语音识别结果,识别错误率降低21.2%,识别效率提高33.5%。实验结果表明:所述牵引车指令特色语音库具有民航针对性,有助于提高智能飞机牵引机器人语音控制系统的识别性能。In order to study the voice control of intelligent aircraft towing robot and realize the accurate and efficient recognition of pilots’ voice commands by towing robot in airport environment, and to solve the problem of poor pertinence of pilots’ voice commands to civil aviation, a voice control method of intelligent aircraft towing robot based on civil aviation pertinence characteristic voice library is proposed. Firstly, the overall function design of voice control for intelligent aircraft towing robot is carried out. Then, the acoustic characteristics of airborne voice commands are analyzed on the basis of the radiotelephony communication standards, and the towing robot command characteristic voice database is established based on the analysis of airport environment and corpus building principles. The experimental results based on isolated word recognition show that the voice recognition results of the tractor instruction characteristic voice database and characteristic test set are 21.2% lower in recognition error rate and 33.5% higher in recognition efficiency than the voice recognition results of the characteristic voice database and common test set. The experimental results show that the tractor command characteristic voice database has civil aviation pertinence and is helpful to improve the recognition performance of the intelligent aircraft towing robot voice control system.

关 键 词：飞机牵引机器人 陆空通话 特色语音库 语音识别 孤立词 

分 类 号：TP391.42[自动化与计算机技术—计算机应用技术]