Emotion Recognition Results using Deep Learning Neural Networks for the Romanian and German Language

摘要

We used the SRoL database for the Romanian language and Emo-DB database for the German language. For the emotion recognition we applied a Deep Learning Neural Network with convolutional layers (DL-CNN). This study includes four emotions: joy, sadness, fury and neutral tone. A 3layer deep learning neural network is used, and the best accuracy rate for Romanian language is 84.5% with fine tunning and for German language 86.8%.