TY - JOUR
T1 - Body sound denoising technologies
T2 - A survey and validation
AU - Li, Enze
AU - Zhang, Haojie
AU - Qian, Kun
AU - Tian, Fuze
AU - Hu, Bin
AU - Schuller, Björn W.
AU - Yamamoto, Yoshiharu
N1 - Publisher Copyright:
© 2025 Elsevier Ltd
PY - 2026/4/1
Y1 - 2026/4/1
N2 - Body sounds are closely related to physiological and mental health, offering valuable insights into both health monitoring and disease diagnosis. Recent studies have increasingly highlighted their potential in these tasks. However, the performance of these applications is significantly impacted by the inherent noise present in body sounds. This noise complicates diagnosis and monitoring, making effective denoising crucial. Effective noise removal from body sounds has been shown to enhance feature extraction and improve the performance of machine learning (ML) and/or deep learning (DL) models, thereby increasing the reliability of digital medical diagnostics. Despite the progress, there is currently no comprehensive summary of denoising techniques and their effectiveness. This survey aims to fill that gap by reviewing both classical and state-of-the-art denoising methods. We will first provide an overview of the theoretical foundations and recent advances in body sound denoising technologies. Next, we will detail the specific denoising techniques applied to different types of body sounds. Our analysis will include a comparison of various methods based on their performance using mainstream body sound databases. Finally, we validate and analyze some denoising algorithms, discuss current challenges and limitations, and suggest future research directions. This work promises to contribute to more accurate and effective body-sound based healthcare solutions.
AB - Body sounds are closely related to physiological and mental health, offering valuable insights into both health monitoring and disease diagnosis. Recent studies have increasingly highlighted their potential in these tasks. However, the performance of these applications is significantly impacted by the inherent noise present in body sounds. This noise complicates diagnosis and monitoring, making effective denoising crucial. Effective noise removal from body sounds has been shown to enhance feature extraction and improve the performance of machine learning (ML) and/or deep learning (DL) models, thereby increasing the reliability of digital medical diagnostics. Despite the progress, there is currently no comprehensive summary of denoising techniques and their effectiveness. This survey aims to fill that gap by reviewing both classical and state-of-the-art denoising methods. We will first provide an overview of the theoretical foundations and recent advances in body sound denoising technologies. Next, we will detail the specific denoising techniques applied to different types of body sounds. Our analysis will include a comparison of various methods based on their performance using mainstream body sound databases. Finally, we validate and analyze some denoising algorithms, discuss current challenges and limitations, and suggest future research directions. This work promises to contribute to more accurate and effective body-sound based healthcare solutions.
KW - Artificial intelligence
KW - Body sound
KW - Computer audition
KW - Noise reduction
KW - Non-contact healthcare
UR - https://www.scopus.com/pages/publications/105024197609
U2 - 10.1016/j.bspc.2025.109302
DO - 10.1016/j.bspc.2025.109302
M3 - Review article
AN - SCOPUS:105024197609
SN - 1746-8094
VL - 114
JO - Biomedical Signal Processing and Control
JF - Biomedical Signal Processing and Control
M1 - 109302
ER -