Improving acoustic model for English ASR System using deep neural network

Author

Quoc Bao Nguyen ; Tat Thang Vu ; Chi Mai Luong

Author_Institution

Univ. of Inf. & Commun. Technol., Thai Nguyen Univ., Thai Nguyen, Vietnam

fYear

2015

fDate

25-28 Jan. 2015

Firstpage

25

Lastpage

29

Abstract

In this paper, a method based on deep learning is applied to improve acoustic model for English Automatic Speech Recognition (ASR) system using two main approaches of deep neural network (Hybrid and bottleneck feature). Deep neural networks systems are able to achieve significant improvements over a number of last year system. The experiments are carried out on the dataset containing speeches on Technology, Entertainment, and Design (TED) Talks. The results show that applying Deep neural network decrease the relative error rate by 33% compared to the MFCC baseline system.

Keywords

Gaussian processes; hidden Markov models; learning (artificial intelligence); mixture models; natural language processing; neural nets; speech recognition; English ASR System; English automatic speech recognition system; GMM; Gaussian mixture model; HMM; MFCC baseline system; TED Talks; Technology Entertainment and Design Talks; acoustic model; bottleneck feature; deep learning; deep neural network; hidden Markov models; hybrid feature; Acoustics; Decoding; Hidden Markov models; Neural networks; Speech; Speech recognition; Training; Automatic Speech Recognition; Deep Bottleneck Features; Hybrid HMM/GMM;

fLanguage

English

Publisher

ieee

Conference_Titel

Computing & Communication Technologies - Research, Innovation, and Vision for the Future (RIVF), 2015 IEEE RIVF International Conference on

Conference_Location

Can Tho

Print_ISBN

978-1-4799-8043-7

Type

conf

DOI

10.1109/RIVF.2015.7049869

Filename

7049869