Emotion recognition from speech using source, system, and prosodic features

Abstract

In this work, source, system, and prosodic features of speech are explored for characterizing and classifying the underlying emotions. Different speech features contribute in different ways to express the emotions, due to their complementary nature. Linear prediction residual samples chosen around glottal closure regions, and glottal pulse parameters are used to represent excitation source information. Linear prediction cepstral coefficients extracted through simple block processing and pitch synchronous analysis represent the vocal tract information. Global and local prosodic features extracted from gross statistics and temporal dynamics of the sequence of duration, pitch, and energy values represent the prosodic information. Emotion recognition models are developed using above mentioned features separately, and in combination. Simulated Telugu emotion database (IITKGP-SESC) is used to evaluate the proposed features. The emotion recognition results obtained using IITKGP-SESC are compared with the results of internationally known Berlin emotion speech database (Emo-DB). Autoassociative neural networks, Gaussian mixture models, and support vector machines are used to develop emotion recognition systems with source, system, and prosodic features, respectively. Weighted combination of evidence has been used while combining the performance of systems developed using different features. From the results, it is observed that, each of the proposed speech features has contributed toward emotion recognition. The combination of features improved the emotion recognition performance, indicating the complementary nature of the features.

Learning Salient Features for Speech Emotion <newline/>Recognition Using Convolutional <newline/>Neural Networks

“…For example, prosodic speech features such as pitch and energy, can be extracted from each interval and are considered as local features [26], [30]. On the other hand, global features such as statistics, can be obtained from the whole speech utterance [22], [23], [16], which typically have a lower dimension than the local ones, leading to less computing time [13], [30].…”

Deep Visual Attributes vs. Hand-Crafted Audio Features on Multidomain Speech Emotion Recognition

“…Moreover, their results indicated that spectral features from consonants may outperform the ones of vowels. Other approaches investigated combinations of features, such as the one of Koolagudi and Rao [14] who used global and local prosodic features and developed emotion recognition models using each feature separately, or combinations of features. Their experiments indicated that features may work in a complementary fashion towards higher recognition performance.…”

SVM based speaker emotion recognition in continuous scale

“…In [8] the authors have performed experiments with selected features, extracted from speech, that represent excitation source information and vocal tract information. The speech features have served as inputs to classifier based on either Gaussian Mixture Model (GMM), Support Vector Machine (SVM), or Auto-Associative Neural Network (AANN).…”