Multimodal Music Emotion Recognition based on WLDNN_GAN

Abstract

In order to solve the current concern of music emotion recognition, this paper proposes the WLDNN_GAN algorithm, the abstract obtained music features are MFCC features, GTF features, midi music information features, through these three features for music emotion recognition and classification. Using the same dataset, the MSE, RMSE and R2 of some currently popular model models are compared horizontally for evaluation, and the experimental results show that the model proposed in this paper can achieve excellent performance in analysing music emotion information.