I pull music from a wide array of sources (Bandcamp, yt-dlp, etc.), drop each album or release in its own folder in ~/Music, run beets on those files to clean the metadata and consolidate into V0 MP3, then rsync the beets library over to a VPS running Navidrome.

Took a fair bit of scripting and config to get beets working as desired, but the process is pretty close to perfect, and scales well into the terabyte range.