Sooo.... I used the text from the mailout for this episode, ran it through NotebookLM and created an AI 'podcast' of a podcast discussing AI generated slop. I highly recommend the section, about 8 min in, where the 'hosts' discuss Mr Blobby... 😂
The only tell I have found so far w newer ai slop is that the drums still sounds kinda shit. (Thankfully it also is easy to spot whats AI when I check out a songs spectogram, like any normal person do.)
and some more thoughts from me (as someone working in film and tv) is that film and tv has been using machine learning for probably ten years or more in stuff like upscaling, generating information in pictures where the exposure is blown out, or noise reduction in audio (and picture). AI use in movies might be a bit more granular than what Suno is, since Suno is what's called end to end, and what people in film and tv need often has to be more granular and open for input. theres also the legal questions that is actually quite relevant since big studios might just blanket ban AI until stuff is resolved because compliancy and liability.
and afaik, with the stochastic models like Suno, Veo 3, Sora etc the big caveat is that you can never do takes the way you do in film and tv, i.e. one more take of the same thing and all that. its all inherently random and you never get another take of the same scene. it will always be different and the characters will look different.
so, as a summary; as long as AI is what they call end-to-end, and it basically removes the whole process of making anything besides prompting, I do not think it holds any cultural value, but I also think it severely flawed as a product as well. The process is such an important part of music and the community of music/musicians, and a subreddit about Suno doesn't hold up to that type of real community at all. I also agree w Chal that it is something worse than Splice-slop because it eliminates most of the process, even when the process already is lazy.
Sooo.... I used the text from the mailout for this episode, ran it through NotebookLM and created an AI 'podcast' of a podcast discussing AI generated slop. I highly recommend the section, about 8 min in, where the 'hosts' discuss Mr Blobby... 😂
I give you podslop...
https://www.youtube.com/watch?v=360jHZanmXE
very into how deep my voice is here.
In regards to the chat about DSPs and AI music, Deezer tag AI content and exclude it from recommendations - https://support.deezer.com/hc/en-gb/articles/28222522835101-AI-Content-Tagging-on-Deezer / https://wegetartists.com/deezer-becomes-first-streamer-to-tag-ai-music-18-of-all-uploaded-music-is-now-ai/
I use Qobuz and don't think I've ever come across AI music, but it has a very different approach to music and business model to Spotify
One way to combat AI music slop:
https://www.youtube.com/watch?v=xMYm2d9bmEA
The only tell I have found so far w newer ai slop is that the drums still sounds kinda shit. (Thankfully it also is easy to spot whats AI when I check out a songs spectogram, like any normal person do.)
and some more thoughts from me (as someone working in film and tv) is that film and tv has been using machine learning for probably ten years or more in stuff like upscaling, generating information in pictures where the exposure is blown out, or noise reduction in audio (and picture). AI use in movies might be a bit more granular than what Suno is, since Suno is what's called end to end, and what people in film and tv need often has to be more granular and open for input. theres also the legal questions that is actually quite relevant since big studios might just blanket ban AI until stuff is resolved because compliancy and liability.
and afaik, with the stochastic models like Suno, Veo 3, Sora etc the big caveat is that you can never do takes the way you do in film and tv, i.e. one more take of the same thing and all that. its all inherently random and you never get another take of the same scene. it will always be different and the characters will look different.
so, as a summary; as long as AI is what they call end-to-end, and it basically removes the whole process of making anything besides prompting, I do not think it holds any cultural value, but I also think it severely flawed as a product as well. The process is such an important part of music and the community of music/musicians, and a subreddit about Suno doesn't hold up to that type of real community at all. I also agree w Chal that it is something worse than Splice-slop because it eliminates most of the process, even when the process already is lazy.
The VCs and tech anuses pushing all this stuff want music to be fully commoditised like domestic energy services.
I'm cold > put the heating on.
It's dark in here > switch on the lights.
It's quiet and weird > put "the music" on.
They want you to not care about who made your music the same way you don't care about who processed your tap water as long as you don't get dysentery.