Speech Emotion Classification and Recognition with different methods for Turkish Language

Cigdem Bakir,Mecit Yuzkat

doi:10.17694/bajece.419557

Abstract

In several application, emotion&nbsp; recognition from the speech signal has been research topic since many years. To determine the emotions from the speech signal, many systems have been developed. To solve the speaker emotion recognition problem, hybrid model is proposed to classify five speech emotions, including&nbsp; anger, sadness, fear, happiness and neutral. The aim this study of was to actualize automatic voice and speech emotion recognition system using hybrid model taking Turkish sound forms and properties into consideration.&nbsp; Approximately 3000 Turkish voice samples of words and clauses with differing lengths have been collected from 25 males and&nbsp; 25 females. In this study, an authentic and unique&nbsp; Turkish&nbsp; database has been used. Features of these voice samples have been obtained using Mel Frequency Cepstral Coefficients (MFCC) and Mel Frequency Discrete Wavelet Coefficients (MFDWC). Moreover, spectral features of these voice samples have been obtained&nbsp; using Support Vector Machine (SVM). Feature vectors of the voice samples obtained have been trained with such methods as Gauss Mixture Model( GMM), Artifical Neural Network (ANN), Dynamic Time Warping (DTW), Hidden Markov Model (HMM) and hybrid model(GMM with combined SVM).&nbsp; This hybrid model has been carried out by combining with SVM and GMM.&nbsp; In first stage of this model, with SVM has been performed&nbsp; subsets obtained vector of&nbsp; spectral features. In the second&nbsp; phase, a set of training and tests have been formed from these spectral features. In the test phase, owner of a given voice sample has been identified taking the trained voice samples into consideration. Results and performances of the algorithms employed in the study for classification have been also demonstrated in a comparative manner.&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;

Highlights

Today, with the development of technology, security problems have come to light
Speech recognition and speech emotion recognition plays an important role in our day due to security and many other reasons
Speech emotion recognition of systems have been developed, being based on an unique database obtained by utilizing Turkish language in this study

Summary

INTRODUCTION

With the development of technology, security problems have come to light. The success of feature extraction techniques, AIBO and segmentation based approach, SBA on different databases and different classification techniques are compared. Since the frequency of speech changes abruptly in some emotions, more study is required to be performed in this respect [2] He et al proposed two different methods of feature extraction for emotion classification from speech data. They studied with a method that calculates the average spectral energy in the lower bands of the speech spectrogram He et al calculated the success of these two methods by using GMM and kNN classification algorithms on two different databases. Lee et al Carried out and emotion study by using the Recurrent Neural Network (RNN) algorithm In this study, they applied the Bidirectional long short-termmemory (BLST) algorithm to determine the time-varying emotions. Classification techniques used of the study performed, experimental study of results and conclusion were given respectively in the section 2, section 3, section 4 and section 5

FEATURE EXTRACTION METHODS

METHOD

Hidden Markov Model

EXPERIMENTAL STUDY

CONCLUSION