所属单位：信息化处（信息化技术中心）

发表刊物：2017 SECOND INTERNATIONAL CONFERENCE ON MECHANICAL, CONTROL AND COMPUTER ENGINEERING (ICMCCE)

关键字：air traffic control LVCSR transfer learning crosslingual deep neural network

摘要：In order to reduce the accidents due to errors of Air Traffic Control (ATC) directives and do responsibility investigation, it's necessary to recognize the audio of the ATC directives into texts. However, the existing Automatic Speech Recognition (ASR) systems are aimed for isolated word recognition, which can't apply to LVCSR. Thus, we analyze the characteristics of ATC directives and develop Large Vocabulary Continuous Speech Recognition (LVCSR) for it. In addition, to solve the issue, that the data of ATC directives are scarce, we proposed a new crosslingual knowledge transfer learning method, i.e. semi-shared-hidden layers crosslingual (Semi-SHL-CDNN). We demonstrate that the Semi-SHL-CDNN can reduce errors by 16.76%, relatively, over monolingual DNNs. Compared with SHLMDNN, the WER is reduced by 1.38% extra.

是否译文：否

合写作者：Wang, Jiawen,杨群

通讯作者：刘绍翰

发表时间：2017-01-01