Classification of Audio Signals Based on Deep Learning Methods for Google Net and Alex Net

Audio signals have a significant impact on every element of human life. Audio classification systems can be developed to solve the efficiency problems of conventional methods using the learning capabilities of cloud computing and deep learning architectures. This research proposes deep learning techniques to classify environmental audio signals based on the spectrograms generated for these sounds while using deep learning techniques. Spectrogram images of environmental audio signals train deep-learning networks such as Google Net and Alex Net on cloud computing. Both networks are trained on a large number of datasets. The achieved accuracy can reach 92.5% on Google Net, and 87.5% on Alex Net. This paper provides that the proposed technique for audio signal classification can be efficient for sound classification and recognition systems.