Research on the mechanism of english stress pronunciation based on DIVA model

Author

Zhang Shaobai ; Ji Yanchun

Author_Institution

Comput. Sci., Nanjing Univ. of Posts & Telecommun., Nanjing, China

fYear

2014

fDate

28-30 July 2014

Firstpage

8381

Lastpage

8385

Abstract

English is a accented beats language. Based on DIVA model, this paper develops a neural computing model which can analyze English word stress. This study modifies the DIVA model´ prosody parameters to simulate English word stress generation and acquisition through the Time-domain pitch synchronous overlap-add (TD-PSOLA) algorithm. The experimental results show that the improved model better express the stressed syllable in English words. In addition, this study also increases pitch perturbation and detects the compensation of pitch and duration, to verify the correctness of the improved DIVA model.

Keywords

natural language processing; neural nets; speech processing; DIVA model; English stress pronunciation; English word stress acquisition simulation; English word stress analysis; English word stress generation simulation; TD-PSOLA algorithm; accented beats language; directions-into-velocities-of-articulators; duration compensation detection; neural computing model; pitch compensation detection; pitch perturbation; time-domain pitch synchronous overlap-add algorithm; Analytical models; Computational modeling; Computer science; Educational institutions; Electronic mail; Speech; Stress; DIVA model; English word stress; TD-PSOLA; perturbation compensation;

fLanguage

English

Publisher

ieee

Conference_Titel

Control Conference (CCC), 2014 33rd Chinese

Conference_Location

Nanjing

Type

conf

DOI

10.1109/ChiCC.2014.6896406

Filename

6896406