n42r / muze-aI

An AI/LLM-driven music discovery and recommendation tool
GNU Lesser General Public License v2.1
0 stars 0 forks source link

Music Discovery Approaches #37

Open n42r opened 10 months ago

n42r commented 10 months ago

A. AI-assisted playlists based on the user profile and their desired subject/theme of the playlist B. Recommended playlists, albums (also tracks but to a lesser degree at the moment) C. Recommended music (sub-)libraries of other users, curators, and sources (see #21, #32) D. Active exploration by sampling tracks from different (sub-)libraries, albums, playlists, curators, etc. (Think about gamifying this process). E. Any others?...

n42r commented 9 months ago

Also

F. Devise an intuitive design for modular/emergent/hierarchical structuring of music libraries

The way most users organize their music in the streaming age is to not organize it and just use search.

However, it might be nice to have an optional way to structure and organize music libraries. This is for several reasons:

  1. Open up the door to a new approach to music discovery: discover music collections (or sub-collections) of others that share similar taste. This does not have to be explicitly shown to the user, the discovery engine can recognize such collections and use them as seeds for music recommendation for the user.
  2. Some music geeks (like myself) find that organizing music (especially for big libraries) helps navigate and (re)discover music from ones own collection. It also places less burden on remembering the names. For some people, organizing their libraries is not something they like to do. This opens the question about if the service one the side, provide an auto-organization of the users library for them to use it. Many benefits include: ability to play only a sub-collection (ex., all my Jazz music).

This is still an early rough idea that needs more thinking