This is interesting in and of itself. For your use case, AI actually does help you getting what you want faster and better than you yourself could do without that brush. But I may ask, why would you need to learn "more tricks" on Suno or Udio if an AI could do that for you? Where do we stop that neverending cycle or why should we? If an AI can help you make better lyrics faster, why would you not use it?
Most free AI song checkers, maybe. You can yourself try and investigate. In my experience and doing lil' research, only certain operations that act non-linearly over the audio actually confuse the checker. Suno is in fact the easiest one to spot nowadays, both humanly and with computer assistance. Other lesser known software gives more uncertainty to the blocks.
This is from your latest version. The other one was also "AI Generated". Usually with other sources I get a more mixed percentage. You can try the operation of re-recording or other stuff to check if there's something that tricks it. But it'd be a waste of time imo. https://www.submithub.com/ai-song-checker
Returning to the initial topic: my solution to these questions is simple -> Composing is not a tool for me like it is for you. Composing for me it's the goal, not the route towards it. I don't have any other aim with the music I do but to do it, perform it if possible, share it sometimes. Whatever else happens, so be it. Your objective has nothing to do with that, so it makes sense for you to cut time and effort wherever possible in steps that you don't control. But mind that this is the key difference, you cannot get too nitpicky with Suno, Udio or others. You can run the slot machine as long as you want but if you wanted to make a series of minimal changes it would become increasingly difficult and tedious to do so. Advancing in your knowledge of music may become a pain more than an aid because you will begin to detect things that are off in AI songs that are not off even in mainstream songs.
That leads us to your next I would say weird statement:
But AI does not learn the way we do. It cannot become a composer who developed his own voice for half a century. You can ask AI itself about that, if my word is not convincing. AIs like Suno can imitate some styles up to a certain degree by prediction. It may be unfair but what you describe does not match with the state-of-the-art knowledge of AI models we possess. Feel free to send that video for reference.
Also:
I'm sure there will be. Right now they're not very good as some members here can assess. Suno itself has the capacity of extracting tracks and converting them to MIDI if I'm not mistaken so you wouldn't even need another AI as long as you handle some software that reads MIDI as sheet music. You can ask AI to generate a sheet music , it's code after all, but more often than not it will fail because there's a fundamental barrier for pieces that are heavily polyphonic. In other words, feel free to try with a symphony movement any of the tools (paid or free) available that use AI to convert audio to PDF sheet music directly. Sure, you may get a single-instrument or a piano transcription, but mind that AI is not magic, it's math, and math used to craft AI cannot solve certain problems that easily. Mind that you do NOT need that either.
That they sound and set me off kinda in a similar way that your piece does. Different styles, yes, but very similar brush, similar way of crafting that can nowadays be perceived. That makes them sound not very good to me, but probably they are acceptable for masses. Mind your target, your target ain't people like me.
These people have told you a half-truth. You can indeed export a midi or get a midi, edit stuff and then re-Suno-ise it. Problem is that even the best workflow is far from perfect and that you do need some musical knowledge to change the "note". More often than not you will end up just "rerolling" till you get something you convince yourself it's the best output because you don't remind what the first iteration was. Imo, this is a waste of time specially with Suno because it will anyways Suno-ise everything you throw at it. Voices will still be weird no matter the note, this characteristic noise will still be present, and more often than not you would not do a dramatical change.
But once again, let me repeat. From what I extract from your words, I would say you do NOT need to be a perfectionist in any way or form. The more you approach that with AI, the less sense it makes to go through it cause the time you consume trying to get exactly what you want grows non-linearly. My recommendation for you would be: don't try to paint a hyperrealistic portrait with a single big shiny brush. If you're already satiisfied
🤔 Man, what are you talking about? Paper does not need internet either. Sheet music is the main medium of music transmission for a reason. Engraving things on stone on a mountain 😁? Pretty expensive probably, I would prefer other more traditional stuff. I have my stuff secured offline just in case.
Not sure if there was anything that I did not reply, I just went out and came back leaving this message as a draft. Attaching 3 things: the audio you requested and what happens when I throw it to some AI that promises to get a piano transcription or a score of my audio:
Kind regards!
Third eye AI transcription.mp3
Third Eye (Arr.).mp3
Third Eye (Arr.).pdf