I also get that, that’s why I up-voted every reply from you. I actually love seeing such completely different perceptions of the same situation. And I also just want to explain my reasoning and how I got there. Which is why my replies tend to be so long.

I’m aware it has no concept of artistic quality. But I also don’t care about the quality of music, especially if perceived by some journalist. I only care if I like music. Some of it is intricately composed, masterfully performed. Some is pop, or generic/simple house.

I have discovered entire genres with the algorithms you seem to think only give narrowing recommendations. Some people probably listened to those and something I liked.

Let me repeat again: I have discovered many, many artists for me that I literally would have no realistic chance of every hearing about in any other way. Ever!

As I said in my other reply, different people like different things. I don’t want an adventure. I want the passive experience. I do other things while listening to music (work, read, tinker, …). I almost always have some music playing, but rarely do I just listen to music (it does happen though). I’ll pick styles depending on mood or task, it’s like the rails that keep me on track while working (as an example). If I’m not listening to music, I lose focus.

I simply can’t do that with an article or other medium that requires my primary attention. I don’t feel a sense of fulfillment either, but increasingly annoyed that reading this thing about music is taking more and more time. Believe me when I tell you, it’s not for me.

It’s the opposite for me. I don’t want to read about music. I just want to listen to music that I don’t know yet but am likely to like. I don’t want to dig around for it. The algorithms you dislike do something that no article or podcast can: give me personally tailored recommendations. She not in an abstract way but just as a playlist.

I was addressing this part of what you said

Ah ok that part wasn’t clear to me, sorry (maybe quote it if you’re reffering to a small part of a comment?). Yes, it would work for that, but I don’t have that collection. I could sail the high seas, but that kinda defeats the purpose of wanting artists to get paid and rather hypocritical. At least they do get paid (even if poorly) using Spotify. So somehow getting to the point where that would work for discovering new-to-me music and that also doesn’t screw over artists seems hard, unless I’m missing something?

EDIT: also, fwiw, I didn’t downvote you lol.

No worries, I don’t pay attention to votes anyway. Doesn’t matter on Lemmy (esp. on comments) unless you’re talking about visibility, which doesn’t do anything on a comment chain like this one either…

First of all, after recent events I’m not touching anything from “Plex” with a proverbial 10 foot pole.

But even that aside, no it won’t do what I want because it can’t. I can’t discover something outside of my library with it. It’s a music player for a Plex library. It can generate playlists of songs with similar styles, and that’s nice and all, but not what I’m looking for. I’m looking for playlists of things I don’t own, or know, or ever heard of, but that are still likely to be something I like. I don’t want a sophisticated “shuffle”.

See my other reply to tofu. Not the same thing. You just couldn’t do what these services do even 2 decades ago. You could discover things, but at a very different pace and very different reach. You’re limited to discover what friends know from them. Discovering things via “press” isn’t free either, it takes time to read the articles, buy the magazines (do they still exists?) and you’re likely to only hear about popular things. You also need to find publications that suit your own taste, or learn which authors are compatible with it.

As for concerts you can only go to those that are near you, which is either local artists or those big enough to tour away from their home base. There are artists that don’t tour at all (probably a third of my catalog falls into this category).

Friends don’t work for me. I don’t know a single person who listens to even close to the things that I like. Sure there’s some overlap occasionally, and I might hear about one artist once a week or month. I get dozens to hundreds recommended by spotify weekly, and I actually end up liking a handful of those. With friends, it also only works with known artists, and it’s incredibly rare to get reommended something that isn’t well known but happens to fit my taste by them (don’t think that ever happened, actually). As an example just last week I got recommended an artist that has 60-something monthly listeners on Spotify (now 74!). I liked them so much I tried to see what I can find, and they got a youtube channel with 3 (live) videos and like 500-ish views each (38 subscribers). NOBODY is ever gonna recommend me those kinds of things, cause nobody ever heard of them, let alone anyone of my friends (and even if they have, they’d have to know to recommend them to me).

As for the listenbrainz/last.fm that is kind of a solution, but it takes a very long time to train up your profile to actually be useful. I haven’t used it in a VERY long time (decades), but last I did it was kinda “meh”. You can also only start out with what you have, as you’re scrobbling what you’re listenting to. I no longer have most of the music I listen to daily as an actual file/library. So getting that up to date would probably cost thousands of dollars, too. Not to mention it being incredibly tedious to actually gather them on various individual shops and sites like bandcamp or wherever those artists happen to be.

So as much as I wish there was, there isn’t really a (pracical) alternative. Let alone one of the same “competence”.

Yes it’s using “an AI”. But that doesn’t mean anything. You can’t just use any AI and have the same result. Just cause AI got a global hype doesn’t mean this is new either. Neural networks have existed for many decades, which is likely what they’re using. The hard part is to get the training data. That is where the value (or usefulnes) comes from. And that source is all their users, listening to all the music, importantly including newly released music, all the time. It’s the basic idea of “people who liked X also liked Y”. What songs people combine together in a playlist. That sort of thing.

We don’t have that data to train “an AI” so we have a local version of this. They have it for millions of users. That’s why their AI is incredibly good at this task. Sure, they also let labels pay them to rank things higher so they get more listens, and that is anything but transparent when and how that happens. But over all, you can’t just magically do what they are doing locally.

Having your own collection is great. But it doesn’t provide the service Spotify does (or any streaming service). 80% of the time I listen to discovery-type generated playlists. I want to find new music. This is fundamentally impossible with the music I own. This is something you can’t self host. Even if you have a vast collection of music you don’t know (by whatever means your get it), you still need the algorithms to pick the music that you’re likely to like.

I really wish I could. I self host basically everything else. Even tried some local music similarity training for “smart playlists”. It’s kinda neat at best, but no where remotely close to the music discovery of Spotify and other online services. You need the massive amounts of users to derive that data.

If separation is all you’re after, any managed switch will work. Even a “smart managed” one. But you’ll need Wi-Fi access points that can actually have SSIDs assigned to VLans, like the TP-Link, Mikrotik or ubiquity (basically anything aimed at business). At least if some or many of the iot devices are Wi-Fi based.

Just to add to this, if you have periodic snapshots on the server side, this does solve the problem. And it simplifies things a lot.

Yes corrected. But I still can never remember what the correct spelling for this is, I actually thought about it. That was the one my keyboard picked with swyping so I just left it…

Wether or not that’s a plus or minus depends on perspective.

Yea, but at least it’s there and still kinda working. Half the shit might be late, but you’re still getting there. Which is nice.

… Of which it effectively still is one.

Yes. That’s basically the point. They call it a “drop in replacement”, but last I used it manually there were some extra steps for what I wanted to do. To be clear: not for every thing you want to setup, just one if the things I read don’t up required extra steps. But I also hear that those things have changed since then and it’s mostly seamless now.

Especially without any additional context or knowledge about their background, directed at someone clearly only starting out, this is incredibly bad advice.

Edit: typos (italic), sorry that was probably hard to read.

Proxmox and Docker don’t really do the same thing. They live in the same area, but the coverage is very different. You can always use docker when your host is running proxmox: either individually or in groups inside of an lxc, or all in w dedicated VM, or even natively on the same house if you prefer chaos. But you can’t do the opposite: Sometimes you just need a VM. Maybe you only need a couple of devices, and you know they run on or are even designed for docker, then that’s the better option. In all other cases, and when just getting started, proxmox is just the way more universal solution if you’re only planning on having a single host (for now).

The management tools in proxmox are great. The community scripts are a fantastic resource and only work with proxmox. I would suggest you set it up natively, not on top of Debian though, even if that’s already installed. Not the least of the reasons are to be able to use ZFS easily, including on on the boot partition (select that in the installer).

Finally, if you’re gonna stick with docker, like others said: consider podman. That really does the same thing docker does, but it’s fully open source. Arguably it’s better in some areas, but on the flip side might, in occasion, require fiddling with something intended specifically for docker and using advanced setups.

Also there really is no wrong answer, either. And you can always change whatever you choose.

If I remember right, problems of that type are either NP-Complete or even NP-Hard, meaning they are incredibly hard to compute and/or don’t have a known solution. They have been studied for centuries, so there should be ample scientific research (see the links the other commenters posted already), but my point is a “perfect solution” probably isn’t a realistic goal. I would aim for an approximation that’s “good enough” for your use case.