Abstract
Speech emotion recognition (SER) plays an important role in real-time applications of human-machine interaction. The Attention Mechanism is widely used to improve the performance of SER. However, the applicable rules of attention mechanism are not deeply discussed. This paper discussed the difference between Global-Attention and Self-Attention and explored their applicable rules to SER classification construction. The experimental results show that the Global-Attention can improve the accuracy of the sequential model, while the Self-Attention can improve the accuracy of the parallel model when conducting the model with the CNN and the LSTM. With this knowledge, a classifier (CNN-LSTM×2+Global-Attention model) for SER is proposed. The experiments result show that it could achieve an accuracy of 85.427% on the EMO-DB dataset.
Highlights
Speech Emotion Recognition (SER) has a wide range of potential applications in areas, such as human–robot interactions [1], computer-aided instruction, e-commerce, and medical assistance [2,3].Traditional, effective analysis methods can be roughly divided into dictionary-based methods and machine-learning-based methods
To propose a Convolutional Neural Networks (CNNs)-LSTM×2+Global-Attention model: By comparing the training convergence speed, accuracy, and generalization ability of different models, we proposed a CNN-LSTM×2+global-attention model and conducted experiments on the EMO-DB dataset, which achieved an accuracy of 85.427%
CNN+ LSTM alone, and combined with the comparison of the confusion matrix as shown in Figures 12–14, we found that the prediction accuracy of the CNN+LSTM×2+GlobalAttention model is significantly better than that of the CNN network alone and the CNN+
Summary
Speech Emotion Recognition (SER) has a wide range of potential applications in areas, such as human–robot interactions [1], computer-aided instruction, e-commerce, and medical assistance [2,3]. Traditional, effective analysis methods can be roughly divided into dictionary-based methods and machine-learning-based methods. The methods based on an emotion dictionary need to use an emotion dictionary which has been labeled manually. These methods rely heavily on the quality of the emotional dictionary, and the maintenance of the dictionary needs a lot of work and material resources. With the continuous emergence of new words, the dictionary cannot meet the application’s needs [4]. The machine-learning method has attracted great attention from researchers. Convolutional Neural Networks (CNNs) [3,5,6] and LSTM networks [7] and their combination [7,8,9]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
