• AngryRedHerring@lemmy.world
      link
      fedilink
      English
      arrow-up
      10
      ·
      20 hours ago

      It was fairly basic, just analyzing audio tracks and extracting the dialogue from them to text. Simple stuff for AI now.

        • Bloefz@lemmy.world
          link
          fedilink
          English
          arrow-up
          1
          ·
          edit-2
          1 hour ago

          Well yeah it has but it really sucked at it till things like Whisper came along. You really had to train it to specific speakers and even then you’d have to manually go over everything.

          Now running the heaviest whisper model and then an LLM to fix obvious mistakes and you have the same quality a human contractor would deliver. Except it can do a day’s work in 5 minutes.

          • lastlybutfirstly@lemmy.world
            link
            fedilink
            English
            arrow-up
            1
            ·
            3 hours ago

            I’m sorry if you lost your job, but if you actually helped improve talking to these damn things then I thank you for your service.