Exploring the Potential of Sound Leakage for Sharing Music

While music recommendation systems effectively deliver music tailored to individual preferences, they face challenges such as filter bubbles, where users are recommended only music matching their preferences, and a reduction in serendipitous musical encounters. In this research, as an approach to these challenges, we reinterpret the Music Leak system in today’s music streaming environment, which we originally proposed at WISS2007 in 2007, focusing on the phenomenon of “sound leakage” in daily life. Music Leak is a system that automatically shares the music a user is listening to with nearby people using the same system via FM radio waves. In evaluation experiments, 75% of users expressed interest in using the system, and 71% showed interest in music from people they did not know. Additionally, all test subjects were able to identify songs even in environments where different genres of music were playing simultaneously. The distinctive feature of this system is that it provides a music sharing experience with a sense of physical distance, creating qualitatively different musical encounters from algorithmic recommendations by sharing information with strong reality—music that ‘someone’ is listening to ‘now’ and ‘there’. This research proposes a new approach to promote ‘unexpected discoveries’ in music recommendation, demonstrating the potential of musical experiences through ‘unintentional sharing’ through implementation and evaluation.