• 0 Posts
  • 180 Comments
Joined 3 years ago
cake
Cake day: July 24th, 2023

help-circle












  • I mean, yes, but also that’s a bit nuclear. Machine learning has real, fully good ethical and responsible uses… The problem is that society has yet to agree on the philosophy of what that is, and most business-first minded people have SUPER shitty, or even completely missing moral compasses.

    So, effectively what you say, yes. But technically with much nuance and many clauses, not entirely.

    We are clearly not ready as a species to handle it. Though, maybe we’ll burn the shit out of our hands in the next coming century enough to learn. But either way, it’s DEFINITELY not an “ignore all risk and run blindly at this shiny new flame” thing like a lot of people seem to think and treat it.






  • I’ve done a handful of those online “tests” where it’s a 320kbps mp3 and flac or wav clips.

    I could almost always tell the difference. The prob was that I would think the mp3 was the higher quality one. In a friends group years ago, another friend of mine had similar results.

    A lot of those “tests” also are strategically designed such that the bitrate of the 320mp3 isn’t saturated enough to run into bitrate aliasing. A lot of people, myself included, sometimes lean on flac because we have heard it make a difference, and it’s so shitty that we just go to the higher quality when we want archival quality versions.

    Also, if you start introducing eq or other processing for various reasons, when you start magnifying sections of lossy, or even lossless audio, you can start hearing missing data or compression artifacts.