Title: Speech emotion recognition of mobile application based on Chebyshev-PCA optimisation

Authors: Guohua Hu; Xiaoxia Zheng; Qingshan Zhao

Addresses: Department of Computer, Xinzhou Teachers University, Xinzhou, Shanxi, China ' Department of Computer, Xinzhou Teachers University, Xinzhou, Shanxi, China ' Department of Computer, Xinzhou Teachers University, Xinzhou, Shanxi, China

Abstract: In order to solve the problem of speech noise interference in mobile applications, the combination of speech noise reduction and feature dimensionality reduction was proposed to improve the emotional recognition performance of emotional features to noisy speech. Firstly, three kinds of common emotions (sad, happy and neutral) in Berlin speech database and CASIA corpus were selected to add noise and extract the emotional features of noisy speech. Secondly, Chebyshev filter was used to filter the noisy signals, and PCA was used to extract the emotional features of the speech after noise reduction. Finally, in order to analyse the effect of speech features processed by Chebyshev-PCA on emotion recognition, the effect of emotion recognition was compared with that of emotional features after denoising by Chebyshev filter and extracting noisy signal features by PCA. The experimental results show that the emotional features processed by Chebyshev-PCA can represent the emotional information more accurately.

Keywords: speech emotion recognition; Chebyshev filter; principal component analysis.

DOI: 10.1504/IJWMC.2020.109262

International Journal of Wireless and Mobile Computing, 2020 Vol.19 No.1, pp.42 - 47

Received: 07 Jan 2020
Accepted: 20 Mar 2020
Published online: 02 Sep 2020 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article