Automatic Speech Recognition of accented Hindi data

Author

Kumari, Prapti ; Shakina Deiv, D. ; Bhattacharya, Mahua

Author_Institution

Comput. Sci. & Eng. Dept., ABV-Indian Inst. Inf. Technol. & Manage., Gwalior, India

fYear

2014

fDate

16-17 April 2014

Firstpage

68

Lastpage

76

Abstract

Inter-speaker variability resulting from factors such as gender, emotions, accent and age lead to the decrease in recognition accuracy of speaker-independent Automatic Speech Recognition systems. Accent variation in Hindi speech and its effect on the performance of Continuous Speech Hindi ASR was studied with the objective to design and develop a compensation technique for accent variation. This paper presents the results of the preliminary experiments conducted in the process.

Keywords

natural language processing; speech recognition; Hindi speech; accent variation; accented Hindi data; compensation technique; continuous speech Hindi ASR; inter-speaker variability; speaker-independent automatic speech recognition systems; Adaptation models; Computational modeling; Digital divide; Feature extraction; Hidden Markov models; Training; Transforms; Accent variation; Automatic Speech Recognition; Recognition Accuracy; accent modeling;

fLanguage

English

Publisher

ieee

Conference_Titel

Computation of Power, Energy, Information and Communication (ICCPEIC), 2014 International Conference on

Conference_Location

Chennai

Print_ISBN

978-1-4799-3826-1

Type

conf

DOI

10.1109/ICCPEIC.2014.6915342

Filename

6915342