Enhancing radar heart sound signal denoising using generative adversarial network

Radar heart sound signal denoising is an important and challenge technology in wireless sensing for contactless healthcare. This paper explores the feasibility of using generative adversarial networks (GANs) to enhance the denoising performance of radar heart sound signals by expanding the dataset within a deep learning framework. Initially, high-quality heart sound signals are generated using WaveGAN. Both synthesized and experimental data are then fed into a dual-autoencoder network, an end-to-end architecture designed for denoising tasks. The numerical results demonstrate that the synthesized data is effective to improve the denoise performance.