The AI hysteria never gets less entertaining to read. It's not the greatest song I've ever heard, but it's a nice little gesture I feel like.
The AI hysteria never gets less entertaining to read. It's not the greatest song I've ever heard, but it's a nice little gesture I feel like.
Remove Balmung already.
I especially love all the "AI proves that there is such a thing as a SOUL, because low quality AI generated content clearly has none."
Seeing 99% of fanarts and fanfics written by actual people makes me think there is no such thing as a SOUL, because it's all absurdly derivative slop containing literally nothing creative or unique. No wonder that all those "creators" are so terrified of AI taking over. They're on the same level if not worse.
And newsflash, things of low quality are low quality. While things of high quality are high quality. If one ignores all the practical AI applications that are already happening and have been for years, ignores all the potential for collaboration, ignores all the incredibly well created content utilizing AI (because nobody even recognizes AI was involved in the first place), and only judges based on the nonsense read on Twitter from some furry artists who made poor career decisions... Then sure, AI is real terrible, guys.
Finally a nice comment among all this hate and unjust bashing !
Sorry, it's just a bad song even without taking AI into account. But go off about how generative AI is super totally the future and isn't already boring to most people because it's beginning to hit the limits of what the models are capable of producing.
There may very well come a day where we have true AI. What we have right now with things like Chat GPT and DALL-E isn't nearly as revolutionary as people like to hype it up to be. It's a highly sophisticated algorithm that predicts things fairly accurately based on the data its trained in, but it cannot understand what it is creating and stay consistent across a long period of time. It has real, practical uses. It depends on the field. Things like art and books and music? Probably not one of them. It makes an initial splash because it's new and exciting, then quickly becomes boring as people find the limits of what it can do and learn to recognize shoveled out slop that all sounds the same (this is true of low quality human made art too, there's just more AI stuff now).
I say this respectfully because the gesture is nice, but.. okay, you have learned how to write a generative AI prompt and basic markup for verse, chorus, and bridge sections, and how to extend a song around a previously generated fragment. That is just a starting point though. I would encourage you to consider giving further thought to the balance between how many lines of actual lyrics you provide, with the length of the song, because chances are quite high that most listeners will have already stopped playback before they have heard any of it since it only starts after a full 215 seconds of fairly uninspired bagpipes. The reason the bagpipes sound bland is because you did not seemingly provide the AI any direction about what movement that whole section should embody like other than the single word "bagpipes". You might have more success if you clarified that as e.g. "slowly swelling bagpipes, marching, triumphant" -- and that is just to start. Next, look into learning how to do music editing. A link to the song still on the AI service that generated it should never be your final version. You are seeing too much meandering because as you expand the song with additional generated portions to reach that 7+ minute length, all the AI looks at is a last bit of the part it needs to connect to, and then it takes it somewhere based on the prompt but without regard for the character of all the parts that came before. AI generated songs are REALLY bad at stylistic consistency over a multipart song, and forget a leitmotif or getting a reprise of a theme.
There is a right way to do this.
If you can not otherwise compose or play an instrument, you CAN use AI as a tool for valid artistic expression, but only by using the AI to generate tons of bits and pieces from which you selectively clip parts to edit together, like those artists that create entirely new images using thousands of different color postage stamps. Only through doing that by hand in an actual editor can you get a song that can carry a theme throughout, reprise a previous progression shifted to a different key, etc -- and the human effort you put in will redeem it from being just another soulless AI piece.
Better yet -- don't use an AI that generates a wav audio file, but one that gives you melodies and progressions as a midi file, which will really let you go wild on chaining sections together, build up a song with any number of instrumental tracks, completely change up what instrument a track is played as, and more -- because at that point you are basically working with notes; it's just that instead of writing entirely from scratch you would be sequencing together melodies and chord progressions that sound good to you. Then when you have your instrumental tracks down, you can use something like autotune to make voice lines fit over it.
Last edited by Eyrilona; 09-05-2024 at 05:05 PM.
|
![]() |
![]() |
![]() |
|
Cookie Policy
This website uses cookies. If you do not wish us to set cookies on your device, please do not use the website. Please read the Square Enix cookies policy for more information. Your use of the website is also subject to the terms in the Square Enix website terms of use and privacy policy and by using the website you are accepting those terms. The Square Enix terms of use, privacy policy and cookies policy can also be found through links at the bottom of the page.