Multi-Layer Hybrid Fuzzy Classification Based on SVM and Improved PSO for Speech Emotion Recognition

• DOI: 10.3390/electronics10232891
• Publication Date: 2021-11-24T07:46:08Z
• Authors: Shihan Huang; Hua Dang; Rongkun Jiang; Yue Hao; Chengbo Xue; Wei Gu
• Keywords: Maxima and minima; Feature (linguistics)

Speech Emotion Recognition (SER) plays a significant role in the field of Human–Computer Interaction (HCI) with wide range applications. However, there are still some issues practical application. One is difference between emotional expression amongst various individuals, and another that indistinguishable emotions may reduce stability SER system. In this paper, we propose multi-layer hybrid fuzzy support vector machine (MLHF-SVM) model, which includes three layers: feature extraction layer, pre-classification classification layer. The MLHF-SVM model solves above-mentioned by c-means (FCM) based on identification information human SVM classifiers, respectively. addition, to overcome weakness FCM tends fall into local minima, an improved natural exponential inertia weight particle swarm optimization (IEPSO) algorithm proposed integrated for optimization. Moreover, non-personalized features personalized combined improve accuracy. order verify effectiveness all popular datasets used simulation. results show can effectively success rate maximum value single emotion recognition 97.67% EmoDB dataset.