cross-posted from: https://europe.pub/post/7719730
cross-posted from: https://europe.pub/post/7719728
Here it is: https://annas-archive.org/blog/backing-up-spotify.html
How much of that music is AI generated slob?
Not that I can help out much with my already full measly 16TB of SSDs…
Anyway, noice!
Why so much SSD storage? Just get SSD’s for you OS and games etc and get HDD’s, maybe put them in a NAS. Much cheaper that SSD’s. You don’t need the speed for data storage.
Download all existing literature to build a library for preservation and you’re called a pirate. Download all existing literature from aforementioned library to train an LLM and you’re a tech innovator. What a strange world we live in.
Hey let’s create our own LLM or something that can pass as an LLM😏 maybe then we can get away with the pirating
Are you rich? Otherwise we’ll still be arrested.
If we’re pirates then they’re privateers, and I know which I respect less.
Download all existing literature to build a library for preservation and you’re called a pirate.
Said library contains petabytes of the exact text of each and every piece of literature.
Download all existing literature from aforementioned library to train an LLM and you’re a tech innovator.
Said model contains gigabytes of a bunch of weights that can never go back to the exact words of the book.
What a strange world we live in.
It’s not strange at all. It’s degrees of compression. You compress a JPEG to the point that it’s unrecognizable, and it’s no longer breaking copyright. It’s essentially like trying to write a book you just read based on memory.
so you’re saying degrading quality while getting filthy rich by stealing everyone else’s work is better than archival efforts? not sure what your point is.
His point is basically that if you remove every 5th word of a book it’s legal to hoard as it’s compressed.
Caping for big tech?
Nasty work.
Lol Meta literally torrented 81 TB of data from the site. Stop with this “degrees of compression” bs
Said model contains gigabytes of a bunch of weights that can never go back to the exact words of the book.
And yet, the tech bros do have access to the exact words. The only difference is that they don’t share, instead choosing to extract value from it by training an LLM and (eventually, hypothetically) turn a profit. The product is created by processing the intellectual labor of billions of people into a formless amalgam of human creativity, which is then exploited for their private benefit.
want the link just so i can know how to avoid it i’m a good girl who does’t steal totally.
I like that billboard linked directly to it
You forgot to administer head pats
As an artist I’m very happy to see my work archived in there. Any suggestions where I can submit my music directly to archives.
I don’t know about Anna’s Archive but I suggest uploading it to Internet Archive.
You could provide a torrent of it directly.
This is a good thing honestly, fuck Spotify it ruined music as much as any single company/service could.
Not as much as Ticketmaster. I would love to be able to see shows, but I’d soon chew my tongue off than buy their nonsense, and fuck their affiliates too. When people stop buying this shit we can solve the problem.
Ngl, it pisses me off that number 4 on the Top 10,000 list is “Clean Baby Sleep White Noise (Loopable)”
Most people still don’t know that their phones most likely already has a noise generator build in without any extra app (at least on iOS)
Which app is that? (I have an iPhone)
It’s in the accessibility settings. You can just search your phone for Background Sounds.
You can then also add it as a button to control center. Or create a shortcut if you want to have it on your Lockscreen or Home Screen.
Neat! Thanks!
Doesn’t exist on Android but highly recommend Atmosphere
One time cost and you can make your own “scene”. You want owls hooting near a stream with cars whizzing by, and a vacuum in the other room? You got it!
There was an app called Taomix on Android, where you could add sounds on the screen around a marker and the volume of the sounds would vary according to the distance between sound and the marker.
Like you could place a windchime near the marker and birds chirping or a river stream sound a little further away. You could always have new combinations, so that sounds weren’t repetitive.
You could then swipe the marker with a push and it would bounce around the screen creating a dynamic sound like passing through a stream or birds singing on your walk.
Then it got bought out by a company and they made a new version with sounds as an in-app purchase, while the previous app was a single purchase.
Then I stopped using it.
Cool people made cool thing! -> awful people made cool thing awfulIdk exactly what comes next for us nor when, but for fuck’s sake WHATEVER we do next has gotta get rid of this goddamn shit.
If it makes you feel any better noise is hard to compress so it costs Spotify more I imagine
How long can whitenoise go before it repeats? Or vice versa, how short?
If it’s only 5seconds it can be played alot…
However, these existing efforts have some major issues:
Over-focus on the most popular artists. There is a long tail of music which only gets preserved when a single person cares enough to share it. And such files are often poorly seeded.
Later…
We primarily used Spotify’s “popularity” metric to prioritize tracks. View the top 10,000 most popular songs in this HTML file (13.8MB gzipped).
I must be kinda stupid, but it sounds to me like there’s some double speak. “Only popular music gets preserved, so we preserved music by popularity”
It’d probably be more beneficial to read the article directly from Anna’s Archive where they display plenty of graphs and infographics to make the data understandable. Unfortunately this article has none of that. The “over-focus on popular artists” is quite literally meaning they’re only missing artists who aren’t being listened to, most of which are probably AI anyway.
https://annas-archive.li/blog/backing-up-spotify.html
To be fair, the 10k is just a sample. The true amount is 86 million, about a quarter of all Spotify songs.
Put another way, for any random song a person listens to, there is a 99.6% likelihood that it is part of the archive. We expect this number to be higher if you filter to only human-created songs. Do remember though that the error bar on listens for popularity 0 is large.
For popularity=0, we ordered tracks by a secondary importance metric based on artist followers and album popularity, and fetched in descending order.
We have stopped here due to the long tail end with diminishing returns (700TB+ additional storage for minor benefit), as well as the bad quality of songs with popularity=0 (many AI generated, hard to filter).
Also it sounds like they had difficulty scraping some of the less popular songs and got them from somewhere else.
Sadly there will be no King Gizzard in this archive.
I’m not sure about that. They only recently removed their music from Spotify and this archive certainly took a good while to create.
I bought all of their albums on Bandcamp for $1 when they had a deal going.That was to good to pass up.
Also got a couple albums on vinyl 🤟
Feels good to have sent some money they way after they dared to ditch Spotift 🙏
Same for GY!BE, I guess
trying to locate individual tracks in massive torrent files of presumably 10,000’s of tracks each sounds horrible, Meta data and tracks and located in different areas. Audio is reencoded to OGG Opus.
For this to be useful for me I would have to spend about $6000 on hard drives (20/terabyte X 300 TB), than convert the files to MP3, and somehow rename the files to their original songs and artists and create appropriate directories.
Do not think this is practical.
most of them are vorbis actually, and not reencoded from spotify
Archival and practical use are different goals. This is not about making it easy to use as a music library
Or stop being an idiot and consider using self-hosted media solutions that handle the metadata for you. Like Plex, Jellyfin, or any of the roughly three dozen options here.
The right torrent client will also allow you to pick and choose which files to download, and you could even go a step further and add a new source provider to e.g. Lidarr that would handle these torrent files and pick out the music you want.
Result?
- no need to transcode to MP3 (not sure why you’d want to do that anyway when OPUS files can be played by practically any modern device)
- no need to manually do any namings
- no need to manually get metadata
- no need to get 300TB storage
Hell if you really wanted to, you could even vibe code a solution that includes a torrent client, these music torrents, and a web interface + API that provides all the necessary info for existing clients to be essentially used as a quasi Spotify alternative, only downloading music you actually listen to.
OPUS files can be played by practically any modern device
The radio of my car (bought in 2020) begs to differ.
Then you’re either transcoding when burning the CD or plugging in a modern player via aux, aren’t you?
I understand why people might not want a music library in FLAC, but just pre-transcoding everything to MP3 in 2025 just seems silly
I use, depending on mood or circumstances, a SD cars with a dozen GB of MP3, or use Finamp on my phone via Android Auto.
My collection is still made exclusively of MP3, mainly because it’s a large-ish collection of pretty high quality files (mostly LAME V0) with all the tags just right (Picard+beet and a ton of work).
I curated this over the years, it sounds more than good enough on my hardware, and I don’t feel like throwing the whole thing away because something a little fancier came along, especially if in this day and age it still means taking a loss in terms of compatibility.
Both with the car, and with my Yahama network receiver/amplifier. The car is relatively new (2020) the amplifier is a little more seasoned, but it can direct play mp3, while I’d have to transcode opus.
Someone shoot me the day I change HiFi hardware over codecs.
With this being said, I’m not sure I’d transcode Opus into MP3 on purpose.
Given you can buy a car made in the 1950s in 2020, that statement is worth about as much as the dump I just took
Only if you think I’m here to screw you over.
It was a new car. A Skoda Fabia. Ordered in January, delivered in May after the first lockdown. The autoradio supports AAC, MP3, FLAC, WMA and vorbis.
And I do use the SD slot, with a dozen GB of MP3. Anything fancier does not make much sense in a car.
And I do use the SD slot, with a dozen GB of MP3. Anything fancier does not make much sense in a car.
I guess my American is showing here, but, do you not want a better stereo in what is arguably one of your most expensive purchases? Your Skoda is just a VW Polo under the skin, and lots of aftermarket headunits are available. I’ve replaced the headunit in most of my vehicles over the years. Worth it every time, plus it’s one of the easier ways to modernize an older vehicle (even my 2008 Toyota Sienna got a new headunit).
I don’t think any car can ever have the acoustic qualities needed to tell the difference between FLAC 192/24 and a decent MP3. Assuming that’s possible at all, but that’s a different discussion.
I don’t think I’d care to go through the trouble of replacing the headunit (which already supports Android Auto) to optimize for codec selection. If anything I’d replace the speakers.
But I don’t use the car so much on local movement (german city, plenty of other options) and on the highway I think the noise is bit too loud to be worth it. I’ll probably just wait until the current ones age enough to annoy me, then buy a nicer set.
That’s fair.
