Accelerated construction of stress relief music datasets using CNN and the Mel-scaled spectrogram

• DOI: 10.1371/journal.pone.0300607
• Publication Date: 2024-05-24T17:22:54Z
• Authors: Suvin Choi; Jong-Ik Park; Cheol-Ho Hong; Sang-Gue Park; Sang-Cheol Park
• Keywords: Spectrogram; Stress Relief

Listening to music is a crucial tool for relieving stress and promoting relaxation. However, the limited options available stress-relief do not cater individual preferences, compromising its effectiveness. Traditional methods of curating rely heavily on measuring biological responses, which time-consuming, expensive, requires specialized measurement devices. In this paper, deep learning approach solve problem introduced that explicitly uses convolutional neural networks provides more efficient economical method generating large datasets music. These are composed Mel-scaled spectrograms include essential sound elements (such as frequency, amplitude, waveform) can be directly extracted from The trained model demonstrated test accuracy 98.7%, clinical study indicated model-selected was effective researcher-verified in terms stress-relieving capacity. This paper underlines transformative potential addressing challenge relief. More importantly, proposed has profound implications therapy because it enables personalized selection, offering enhanced emotional well-being.