Automatic Speech Recognition System with Output-Gate Projected Gated Recurrent Unit

作     者：Cheng, Gaofeng Zhang, Pengyuan Xu, Ji 

作者机构：Univ Chinese Acad Sci Sch Elect Elect & Commun Engn Beijing Peoples R China Univ Chinese Acad Sci Beijing Peoples R China Chinese Acad Sci Inst Acoust Key Lab Speech Acoust & Content Understanding Beijing Peoples R China 

出 版 物：《IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS》 (电子信息通信学会汇刊：信息与系统)

年 卷 期：2019年第E102D卷第2期

页      面：355-363页

核心收录：

学科分类：0809[工学-电子科学与技术（可授工学、理学学位）] 08[工学] 0835[工学-软件工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：National Key Research and Development Plan [2016YFB0801203, 2016YFB0801200] National Natural Science Foundation of China [11590770-4, U1536117, 11504406, 11461141004] Key Science and Technology Project of the Xinjiang Uygur Autonomous Region [2016A03007-1] Pre-research Project for Equipment of General Information System [JZX2017-0994/Y306] 

主　　题：GRU LSTM neural network language model speech recognition 

摘      要：The long short-term memory recurrent neural network (LSTM) has achieved tremendous success for automatic speech recognition (ASR). However, the complicated gating mechanism of LSTM introduces a massive computational cost and limits the application of LSTM in some scenarios. In this paper, we describe our work on accelerating the decoding speed and improving the decoding accuracy. First, we propose an architecture, which is called Projected Gated Recurrent Unit (PGRU), for ASR tasks, and show that the PGRU can consistently outperform the standard GRU. Second, to improve the PGRU generalization, particularly on large-scale ASR tasks, we propose the Output-gate PGRU (OPGRU). In addition, the time delay neural network (TDNN) and normalization methods are found beneficial for OPGRU. In this paper, we apply the OPGRU for both the acoustic model and recurrent neural network language model (RNN-LM). Finally, we evaluate the PGRU on the total Eval2000 / RT03 test sets, and the proposed OPGRU single ASR system achieves 0.9% / 0.9% absolute (8.2% / 8.6% relative) reduction in word error rate (WER) compared to our previous best LSTM single ASR system. Furthermore, the OPGRU ASR system achieves significant speed-up on both acoustic model and language model rescoring.