Enhancement of cardiac and respiratory sounds for cellphone reproduction by means of digital sound processing methods.

Telemedicine's rising popularity, driven by its convenience and accessibility, faces a challenge in remote physical auscultation, particularly for assessing lung and heart sounds. We propose a smartphone-based tele-auscultation approach for capturing lung and heart sounds, based on pitch-shifting customized for smartphone listening, overcoming the technical obstacle found in the limited accuracy of smartphone speakers for reproducing low-frequency sounds, such as heart sounds. We created a database of heart and lung sounds captured with a smartphone, and then we conducted two evaluations, one with sounds from open-source databases and one with sounds from an in-house database. Pitch-shifting algorithms from PaulStretch and SoundTouch libraries were applied, and validated against original recordings through a web survey, initially using conventional headphones, as a first step towards delivering them through loudspeakers. In the open-source database experiment, 71.6% and 80% of 40 final-year medical students indicated preserved clinical information in respiratory and heart sounds, respectively. In the in-house database experiment, 14 physicians and final-year medical students validated the processed audio samples, revealing that 76.5% and 71% of respiratory and heart sounds, respectively, maintained clinical information. These results suggest the potential use of pitch-shifted sounds in tele-auscultation devices like smartphones. However, further research is essential to understand smartphones' playback capabilities in a clinical setting. [ABSTRACT FROM AUTHOR]