• CaptainSpaceman@lemmy.world
    link
    fedilink
    English
    arrow-up
    13
    arrow-down
    1
    ·
    6 months ago

    Incorrect. This will better train LLMs since they can detect distinct speakers/sounds more easily and thus applying the proper metadata tags and profile information at a more accurate clip.

    All so they can deliver more ads!

    • KairuByte@lemmy.dbzer0.com
      link
      fedilink
      English
      arrow-up
      1
      ·
      6 months ago

      <.< I’m assuming this is a joke, but feel the need to point out to the ones who don’t realize… LLMs aren’t trained on audio recordings.