Acoustic Feature Extraction of Speech for Intelligent Garbage Classification Based on MFCC

Abstract:
Garbage classification to the society, ecology and economy can create positive benefits. According to a small-scale social survey, about 70% of people subjectively have the intention to sort garbage, 80% of cities have implemented relevant classification policies. However, the development of garbage classification is not smooth, and the main factor causing classification difficulties is the lack of recognition of classification standards. Therefore, it is of great significance to extract speech features from speech intelligent recognition garbage classification. In this paper, voice endpoint detection VAD algorithm is used to select the voice endpoint, and then MFCC algorithm is used to extract the voice features to obtain the voice features. Finally, ResNet neural network is constructed as the acoustic model, which reduces the computation on the premise of ensuring the accuracy of feature extraction.