Hey all, I just happened upon a trove (~100gb) of MP3s from when I was a sweaty teenager with a DSL line and a 5-disc ripper. Anyone have recommendations on OSS toolsuites to begin indexing/sorting out the files? Everything apparently got flat dumped from a directory structure to a folder. There’s metadata in about 90% of the content - ~10k songs to sort through shouldn’t be too rough with hashing.

I’d be more than willing to seed once everything is in a happy place, there’s probably a few hundred studio cuts of songs that haven’t seen the light of day yet.

Sounds like a good task for MusicBrainz Picard

If you don't have trouble using the terminal, I'd recommend Beets.

Now that's a name I haven't seen in a loooooooong time. Excellent tool last time I used it...about 15 years ago.

Good recommendation.

Picard mentioned in the other comment is great. If you want a more manual approach, on a Linux machine you can use exiftool to extract the metadata and then use that information to do whatever, e.g. organise them by folders.

$ exiftool -j 救世主.ogg 
[{
  "SourceFile": "救世主.ogg",
  "FileName": "救世主.ogg",
  "FileSize": "3.4 MB",
  "FileType": "OPUS",
  "FileTypeExtension": "opus",
  "MIMEType": "audio/ogg",
  "OpusVersion": 1,
  "AudioChannels": 2,
  "SampleRate": 48000,
  "OutputGain": 1,
  "Vendor": "Lavf60.9.100",
  "PictureType": "Front Cover",
  "PictureMIMEType": "image/jpeg",
  "PictureDescription": "",
  "PictureWidth": 0,
  "PictureHeight": 0,
  "PictureBitsPerPixel": 0,
  "PictureIndexedColors": 0,
  "PictureLength": 102126,
  "Picture": "(Binary data 102126 bytes, use -b option to extract)",
  "Totaldiscs": 1,
  "Genre": "J-POP",
  "Encoder": "Lavf60.9.100",
  "Releasetype": "single",
  "Album": "救世主",
  "Language": "jpn",
  "Date": "2023-05-07",
  "Title": "救世主",
  "Totaltracks": 1,
  "Artist": "月詠み",
  "Duration": "00:03:20.868000000",
  "TrackNumber": 1,
  "Tracktotal": 1,
  "Disctotal": 1
}]

You're saying each mp3 is ~10MB? That's probably 320Kbps quality, so should be at least better than topical 128Kbps.

My maths:
10,000,000B ÷ 240 seconds song = 41,666B(yte).

41,666B × 8(bit) = ~333,333b

333,333b ÷ 1,024 = 325bps (thinking kibibit vs kilobit depending on your IS)

(again, rough numbers)

I haven’t said anything 😘

My sushi roll experience was going to be iterating the byte-string headers and grabbing out anything of value, so essentially this. My hope was that there was a solution present to reduce the hand-rolling… tbd but it appears that Picard is a viable option based on yours and others commentary/upvotes

Have you tried strawberry music player? Has some pretty decent tools to get music files organized.

I've for plenty of space to seed indefinitely. Waiting for that AA scrape to finally be available from Spotify lol

There were grand ambitions for a n-petabyte scale u. server in a milk crate, but then LLM happened 😭

Oooooooooo, this is hot af. You’re a real one

I used this very recently to clean up multiple iPod backups (80-120GB) from my high school days. You still need to delete duplicates manually, but the software makes the task immensely easier.

Music isn't torrents in 2026. We're all on Soulseek. You could certainly start there but make sure it's tagged well.

I could make it happen eventually. I currently house a 4U 36 bay and 3U 24 bay in my 42U rack. Not fully populated, but currently 10x 8TB in raid 5 and 8x 16TB in raid 5. Only 1 spare 8TB 😅 a few older 3TB in raid 10 with plenty of spares.

Picard can help you sort through most of it. It can identify most things even without tags. Just be careful and check everything it wants to do before letting it do its thing.

Is VPN enough protection for Soulseek? I think Nicotine+ flatpak is available on my distro

This is why thumb drives and write bits were made

They use a bunch of copilot LLM actions and code updates across their platform. Picards last release is before any major agent rewrites to the main branch, so it’s been pinned.

The real question, will rawdogging it trigger a DCMA?

Sitting on mostly spinny shucked wd reds from whenever they’ve gone on sale and a few 1u servers w u. shenannigans. Nothing insane, still under 100tb in raid 10.

The only upside to my spinning disks are they're SAS enterprise grade. So long uptime and expected lifespan

It really is the best tool that comes to mind if you want to go through a very large library with minimal manual work.

GIMME MAXIMA S.M.A.R.T.!

I’ll see if I can track down my first SSD, The Ancient One, it’s an old 32gb g.skill

As long as each music disc has it's own folder

Practices from a sweaty teen that obviously loved during the time of CD changers, with a reflective self objectifying them as such. Obviously I’ve shown growth since based both on the OP and comments in this thread.

Reading literacy is hard.

???

Well if it indeed is a flat directory structure and you can't restore some kind of subdirectory from e.g. filenames or metadata, then musicbrain picard won't help you because it can't "cluster" and use that find matching releases. Afaik there is no software that can do that without some kind of pre-clustering.

It's safer than torrents, but tbh anytime you're uploading anything, you should be using protection

Sorry, what are you saying? No offence but this response is a mess of non sequitur. The Picard thing makes no sense.

I get that maybe you don't like SMP cause they started using AI tools (fair) but, if you actually look at the codebase none of what would help you is even AI driven and has been a stable function long before the dev started using AI.

We're still developing it! New major version soon!

https://blog.metabrainz.org/2026/09/10/picard-3-release-candidate-2/ it gets closer!