The IOIT English ASR system for IWSLT 2016

The IOIT English ASR system for IWSLT 2016 Van Huy Nguyen1, Trung-Nghia Phung2, Tat Thang Vu3, Chi Mai Luong3 1Thai Nguyen University of Technology, Vietnam 2University of Information and Communication Technology, Thai Nguyen University, Vietnam 3Institute of Information and Technology (IOIT), Vietnamese Academy of Science and Technology, Vietnam huynguyen@tnut.edu.vn, ptnghia@ictu.edu.vn, {vtthang,lcmai}@ioit.ac.vn Abstract This paper describes the speech recognition system of IOIT for IWSLT 2016. Four single DNN-based systems were developed to produce the 1st-pass lattices for the test sets using a baseline language model. The 2nd-pass lattices were further obtained by applying N-best list rescoring on topic adapted language models which were constructed from closed topic sentences by applying a text selection method. The final transcriptions of test sets were finally produced by combining the rescored results. On the 2013 evaluation set, we are able to reduce the word error rate of 1.62% absolute. On the 2014, provided as a development set, the word error rate of our transcription is 11.3%. 1. Introduction The International Workshop on Spoken Language Transla- tion(IWSLT) is a yearly scientific workshop, associated with an open evaluation campaign on spoken language transla- tion. One part of the campaign focuses on the translation of TED, QED Talks, and the conversations conducted via Skype. TED and QED talks are a collection of public lec- tures on a variety of topics, ranging from Technology, Enter- tainment and Education to design. As in the previous years, the evaluation offers specific tracks for all the core technolo- gies involved in spoken language translation, namely auto- matic speech recognition (ASR), machine translation (MT), and spoken language translation (SLT). The goal of the ASR track is the transcription of au- dio coming from unsegmented TED, QED talks, and Mi- crosoft Speech Language Translation (MSLT) Corpus that was drawn from Skype conversations [1], in order to inter- face with the machine translation components in the speech- translation track. The quality of the resulting transcriptions is measured in word error rate (WER). In this paper, we describe our speech recognition system which participated in the ASR track of the IWSLT 2016 eval- uation campaign. The system is a further development of our last year’s evaluation system [2]. There are four single hy- brid acoustic models in our system. These models and an interpolated language model were used to produce lattices which were further applied N-best list rescoring with a topic adapted language model. The final transcriptions of the test sets were combinations of the rescored results. The organization of the paper is as follows. Section 2 describes the data that our system is trained on. This is fol- lowed by Section 3 which provides a description of the way to extract acoustic features. An overview of the techniques, used to build our acoustic models, is given in Section 4. Lan- guage model and dictionary are presented in Section 5. We describe the decoding procedure and results in Section 6 and conclude the paper in Section 7. 2. Training Corpus For training acoustic models, we used …

Content truncated to 500 words