Based on the advantages of Convolutional Neural Networks,such as strong learning ability and high portability,and combined with the characteristic that data enhancement can improve the model's generalization ability,a Deep Learning acoustic scene classification method based on data enhancement is proposed.Then,this paper constructs an acoustic scene classification model based on VGG16 and Mixup.Finally,extensive tests are conducted on the experimental model using the ESC-50 dataset.The experimental results indicate that the use of the Mixup data enhancement method can improve the model's accuracy by 6.44%,and the model achieves a classification accuracy of 81.56%on this dataset,which is higher than the accuracy of the baseline system by 37.26%.This confirms the reliability and effectiveness of this method and can effectively improve the model's classification performance.
关键词
卷积神经网络/深度学习/声学场景分类/数据增强
Key words
Convolutional Neural Networks/Deep Learning/acoustic scene classification/data enhancement
维普
万方数据