Environmental sound classification of western black-crowned gibbon habitat based on spectral subtraction and VGG16
Xiaotao Zhou, Kunrong Hu, Zhenhua Guan · 2022 IEEE 5th Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC) · 2022
Deep learning-based approaches have been widely used for environmental sound classification tasks in recent years due to the growth and development of this field, creating the technical groundwork for the research of environmental sound classification. However, the presence of noise in the ecological environment leads to a dramatic decrease in its recognition rate. The major goal of this study is to improve the classification accuracy of the model in the presence of noise interference. It does this by proposing an ecological environmental sound classification approach in a noisy background. The input audio is first noise-reduced by spectral subtraction, and then the Mel frequency cepstral parameter features are extracted and fed into the fine-tuned VGG16 model for classification training. The classification of five environmental sounds in the western black-crowned gibbon habitat of the Mourning Mountain was identified with an accuracy of 98.76%.