USEN – U MUSIC / USEN

Through collaborative research with USEN CORPORATION, a music selection system using AI called “U MUSIC” was officially released as a product in September 2020.

For the development of this music selection system, approximately 500,000 songs from USEN’s music dataset were used, which were manually tagged with labels such as “comfortable” and "refreshing," genres, and others. These were used for training, and an AI model was developed that can automatically tag new songs. In addition to the characteristics of the songs, the system also incorporates real-time information such as season, weather, and time of day, to select music that matches the current environment. The goal was to create a system that can automatically select BGM that always matches the store’s situation and owner’s preferences.

Previously, it was common for each user or store to choose their favorite songs from a limited set of programs (channels). Tagging with AI enables us to explore areas that are difficult for humans to quantify, such as “both light and sentimental,” creating unexpected encounters with new songs.

Recommendation functions based on past selection trends make it difficult to encounter new music. “U MUSIC,” which delivers music with a continuous flow that transcends genres and eras, will provide novel encounters with unknown songs by breaking away from simple recommendation functions based on past selection trends.

Links

- U MUSIC (Japanese only) - U MUSIC for HANEDA INNOVATION CITY (Japanese only)

Client

USEN CORPORATION

Credits

Techical Direction: Nao Tokui (Qosmo, Inc.) Machine Learning: Max Frenzel (Qosmo, Inc.), Bogdan Teleaga (Qosmo, Inc.) Server Development: Bogdan Teleaga (Qosmo, Inc.) Background Visual: Shoya Dozono (Qosmo, Inc.) Project Management: Hiroshi Yamato (Signal Compose, Inc.), Yumi Takahashi (Qosmo, Inc.)

Product

This project uses Qosmo Music & Sound AI. Playlist Generation