Deep learning for emotional speech recognition

• Main Authors: Alhamada, M. I., Khalifa, Othman Omran, Hassan Abdalla Hashim, Aisha
• Format: Conference or Workshop Item
• Language: English; English
• Published: AIP Publishing 2020
• Subjects: T Technology (General); T10.5 Communication of technical information
• Online Access: http://irep.iium.edu.my/82389/8/Certificate%20ICEDSA%202020%20%20%2328%20Deep%20Learning%20for%20Emotional%20Speech%20Recognition.pdf; http://irep.iium.edu.my/82389/18/82389%20Deep%20learning%20for%20emotional%20speech%20recognition.pdf; http://irep.iium.edu.my/82389/; https://aip.scitation.org/doi/pdf/10.1063/5.0032381

Emotion speech recognition is a developing field in machine learning. The main purpose of this field is to produce a convenient system that is able to effortlessly communicate and interact with humans. Speech signals are loaded with information which is divided into two main categories, linguistic and paralinguistic; emotions belong to the latter tree. Developing systems that can understand paralinguistic information is paramount for better human-machine interactions. The complete reliability of the current speech emotion recognition systems is far from being achieved. To wit, the objective of this project is to review different methods used in speech emotion recognition SER. Different extracted features like MFCC as well as feature classifications methods like HMM, GMM, LTSTM and ANN are also researched. This research will also investigate different speech emotion databases that are commonly used. Finally, this paper implements an architecture of CNN that is used for speech emotion recognition. The proposed CNN model achieved 93.96% accuracy rate in detecting 5 emotions.