In the age of information and automation, where the robotics sphere is increasing each day, and people interact with automation systems, emotion recognition mechanisms will play an important role for better interactions between humans and machines. The emotion recognition is very important in AI spheres since it will make human-computer interface (HCI) more user-friendly and similar to the real-man behavior. The audio emotion corpus of Kazakh and Russian languages which has more than 16 000 records on 8 type of emotions is collected during the research. One hundred and one participants participated in the assembly of the corps. Extensive amount of work has been done related to sorting and human recognizing of emotion using the majority voting method. The data set was divided into TRAIN (80%), VALIDATION (10%) and TESTING (10%) sets. For this problem, Deep Neural Network has been applied with Stochastic Gradient Descent optimizer with batch normalization. MFCC feature supported by the LIBROSSA library was used as an input.
Digital Object Identifier (DOI)
Kozhakhmet, Kanat; Zhumaliyeva, Rakhima; Shoiynbek, Aisultan; and Sultanova, Nazerke
"Speech Emotion Recognition For Kazakh And Russian Languages,"
Applied Mathematics & Information Sciences: Vol. 14:
1, Article 8.
Available at: https://digitalcommons.aaru.edu.jo/amis/vol14/iss1/8