Hey, that’s almost what I made Gire for. It doesn’t support completely anonymous reads though (requires SSH keys for auth).

Oh, a shameless self-plug opportunity. TLDR: I’m making a Subsonic server trying to replicate a somewhat passable self-hosted Spotify/… experience.

My current setup is a Subsonic stack: gonic as a server, Lidarr as a library organizer/downloader (it works natively with Prowlarr BTW) and Sonixd/Tempo as desktop/mobile clients. Frankly, it sucks - no discovery and almost nothing gets auto-downloaded. Even ignoring the crucial discovery part, I got tired of manually downloading albums from Bandcamp - manually tagging, manually feeding files into Lidarr, manually adding metadata to MusicBrainz so Lidarr actually works, …

So, I’ve been working on a Subsonic-compatible server for quite a while. The basic idea is:

• regular user libraries are already covered by various Subsonic (and other) servers, no need to reinvent this
• we can, uuuuh, cache stuff from YouTube/Bandcamp/… to stream it via Subsonic API to leverage existing clients
• we can fetch recommendation lists from somewhere else (ListenBrainz/…) and use the existing library or even auto-search YouTube/Bandcamp/… if the track’s missing

Well then, what can Tapesonic already do?

• given a YouTube/Bandcamp/maybe-more URL, download it and then two-click import it as a playlist or an album, making it listenable via most Subsonic clients (single-video “mixtapes” with timecoded chapters and playlists are supported)
• combine your “main” library (from probably most other Subsonic servers like gonic/Navidrome/…) with it’s own so you don’t have to switch servers back and forth

What’s broken/not implemented yet?

• acceptable server UI; features are far more important especially considering that the goal is to use the existing clients for almost everything
• seeking in clients; working on a fix for the past few days
• transcoding; not a priority for now
• auto-searching YouTube/Bandcamp/…; not a priority for now
• ListenBrainz playlists; I have “already listened” playlists kinda working in a stash, but those are paused until seeking works; “discovery” will just work after auto-searching gets implemented

No user documentation, half-broken, breaking changes one after another, slow progress - but hey, it doesn’t mess with your existing library and already made my life quite easier with Bandcamp albums. A docker image is available - so give it a spin if you want to and then ragequit because it sucks