Improved Frame Level Features and SVM Supervectors Approach for the Recogniton of Emotional States from Speech: Application to categorical and dimensional states

TL;DR

This paper explores the use of frame-level features and SVM supervectors to improve speech emotion recognition, focusing on categorical and dimensional emotional states using the Berlin database.

Contribution

It introduces a novel approach combining frame-level features with SVM supervectors for enhanced emotion classification in speech recognition.

Findings

Improved accuracy in emotion classification.

Effective use of frame-level features over global features.

Demonstrated applicability to both categorical and dimensional states.

Abstract

The purpose of speech emotion recognition system is to classify speakers utterances into different emotional states such as disgust, boredom, sadness, neutral and happiness. Speech features that are commonly used in speech emotion recognition rely on global utterance level prosodic features. In our work, we evaluate the impact of frame level feature extraction. The speech samples are from Berlin emotional database and the features extracted from these utterances are energy, different variant of mel frequency cepstrum coefficients, velocity and acceleration features.