Reply
  • lucid 🍹
    Dec 21, 2025
    ·
    1 reply

    i doubt it but i wonder if this could rip deleted songs off there bc i have alot of them

  • lucid 🍹
    Dec 21, 2025
    relic

    the torrent is literally just metadata rn not actual music files btw

    makes alot more sense

  • Dec 21, 2025
    BIG STEPPER SINGH

    What’s that?

    Software that allows you to download from other peoples computer. If I couldn’t find an album I’d go to soul seek. Haven’t used it in a long time but it’s probably still amazing for digging through

  • Dec 21, 2025
    ·
    1 reply
    Oblivion X

    This is bad. Just gonna lead to people training AI models on people's songs without permission.

    People can already do this tho

    Yes but there was much more friction to doing this at a large scale. The amount of training data easily available means much more people will make use of it that wouldnt have before. Its pretty much how the field works

  • Dec 21, 2025
    fun guy

    Just gonna lead to people training AI models on people's songs without permission

    Please read my response above and if you disagree then do say why

  • Dec 21, 2025
    ·
    1 reply
    Platinum

    Yes but there was much more friction to doing this at a large scale. The amount of training data easily available means much more people will make use of it that wouldnt have before. Its pretty much how the field works

    What friction are you speaking of ?

  • Dec 21, 2025
    Corporate Mór

    I’ve written most of my best stuff in D, Bb and B major

    C and G major feel like cheating

  • Dec 21, 2025
    ·
    2 replies
    Oblivion X

    What friction are you speaking of ?

    Having a very large amount of audio files of songs in an organised way that you can easily download in one go. No searching of songs/knowing what songs you want is required before hand and no complex data structure required to get the songs.

    As far as I know there is no other publically available dataset that has this many songs in one download. SoulSeek for examples requires you to provide the artists/songs you want and download them one at a time if theres multiple you want - this is friction.

    This means you can pretty easily train one model on nearly all music of any significance in recorded music history.

    This isnt the same as training a model on Drakes songs, or on Cartis songs.

  • bmass 🇵🇭
    Dec 21, 2025
    ·
    1 reply
    lucid

    i doubt it but i wonder if this could rip deleted songs off there bc i have alot of them

    Hi

  • Dec 21, 2025
    ·
    1 reply

    Slsk all u need

  • lucid 🍹
    Dec 21, 2025
    bmass

    Hi

    sup

  • Dec 21, 2025
    ·
    1 reply

    this image is hilarious the most uploaded hip hop subgenre is christian hip hop and two of the most popular rock genres are black metal and rockabilly

  • Dec 21, 2025

    hyperpop the most popular pop genre

  • Dec 21, 2025
    WRU

    this image is hilarious the most uploaded hip hop subgenre is christian hip hop and two of the most popular rock genres are black metal and rockabilly

    Rockabilly elite

  • Dec 21, 2025

    Extremely based

  • Dec 21, 2025
    Bo Ceephus

    Lol there is no reason to pay for anything once you have internet.

    real 1 detected

  • Dec 21, 2025
    ·
    1 reply
    Platinum

    Having a very large amount of audio files of songs in an organised way that you can easily download in one go. No searching of songs/knowing what songs you want is required before hand and no complex data structure required to get the songs.

    As far as I know there is no other publically available dataset that has this many songs in one download. SoulSeek for examples requires you to provide the artists/songs you want and download them one at a time if theres multiple you want - this is friction.

    This means you can pretty easily train one model on nearly all music of any significance in recorded music history.

    This isnt the same as training a model on Drakes songs, or on Cartis songs.

    Ehh to a degree but you have to account for the fact that the file size is 300 TB which is not something that's accessible to just download in one go. And then add in the amount of power it would take to train an model on all that just makes it not something feasible.

    You would have to piece meal it, and at that point you can just use numerous other music scrapper out there to the point it doesn't really change much

  • Dec 21, 2025
    ·
    1 reply
    Oblivion X

    Ehh to a degree but you have to account for the fact that the file size is 300 TB which is not something that's accessible to just download in one go. And then add in the amount of power it would take to train an model on all that just makes it not something feasible.

    You would have to piece meal it, and at that point you can just use numerous other music scrapper out there to the point it doesn't really change much

    As far as I can see you dont need to download all 300TB as its in batches. And then its not really that much data as its just signals or other derived features from the songs youll be training on, not the actual audio file.

    Dont agree that it makes it similar to using the existing tools if you piece meal it as the scale is just a whole different realm even if you piece meal. You can easily train on 100k - 1mill songs which wont be that big in size, and isnt really feasible using current scraping tools

  • Dec 21, 2025
    Bo Ceephus

    Lol there is no reason to pay for anything once you have internet.

    most people pay for convenience really rather than access

  • Dec 21, 2025

    ask someone to listen to or even worse download something not on streaming/youtube and see how they look at you like you just asked them to shoot their dog in the head

  • proper 🔩
    Dec 21, 2025
    ·
    1 reply
    Platinum

    Having a very large amount of audio files of songs in an organised way that you can easily download in one go. No searching of songs/knowing what songs you want is required before hand and no complex data structure required to get the songs.

    As far as I know there is no other publically available dataset that has this many songs in one download. SoulSeek for examples requires you to provide the artists/songs you want and download them one at a time if theres multiple you want - this is friction.

    This means you can pretty easily train one model on nearly all music of any significance in recorded music history.

    This isnt the same as training a model on Drakes songs, or on Cartis songs.

    you can download entire albums at once using soulseek what are you talking about lol

    why would anyone wanna dled 300tb or go through 300tb worth of files to pick and choose which songs/albums they want to listen to

    when they can just search whatever album they actually need/want to listen to in soulseek at any moment without compromising hd space

  • proper 🔩
    Dec 21, 2025
    ·
    1 reply

    avg person/music listener won’t even have 300tb of hd space
    most these ppl don’t even have 4tb man lol

  • Dec 21, 2025
    ·
    1 reply
    proper

    you can download entire albums at once using soulseek what are you talking about lol

    why would anyone wanna dled 300tb or go through 300tb worth of files to pick and choose which songs/albums they want to listen to

    when they can just search whatever album they actually need/want to listen to in soulseek at any moment without compromising hd space

    My point is that you wouldnt want to pick and choose albums - complete opposite of what you are talking about. Feel you arent reading my messages as Im obviously not talking about your average music listener

  • Dec 21, 2025
    ·
    1 reply
    Platinum

    As far as I can see you dont need to download all 300TB as its in batches. And then its not really that much data as its just signals or other derived features from the songs youll be training on, not the actual audio file.

    Dont agree that it makes it similar to using the existing tools if you piece meal it as the scale is just a whole different realm even if you piece meal. You can easily train on 100k - 1mill songs which wont be that big in size, and isnt really feasible using current scraping tools

    Yea you dont need to download the whole thing. But if ur not downloading the whole thing, whats the difference than just using another music scrapper where you can download music in batches already ?

    That scale is definitely feasible with other scraping tools. You can download whole discographies easily via torrents that amounts to that size pretty easily.

  • SABMAN TURNT 🧔🏻
    Dec 21, 2025
    proper

    avg person/music listener won’t even have 300tb of hd space
    most these ppl don’t even have 4tb man lol

    a single TB*

    unless their computers are post-pandemic imo

    (obviously aside from people who focus on specs)