Voice Cloning is the Game Changer
I'm sure most here that use Suno AI are aware of a modest voice cloning feature they introduced (as kind of a 'disguised' feature imo). Essentially, you can create "voices" on the app that you can select to be emulated as the artist voice for whatever track you're creating. This is only accessible via desktop (at the time of writing), so if you've been using Suno from your phone, you're not gonna see it.
Even though they call the feature 'voices', it basically mandates that you provide your own voice (or at least that individual must be present) because there is a "verification" process that you must complete after your submission where you're required to say a randomly generated phrase they give you. So, in essence, Suno is asking for you the creator to lend your own voice.
Through experimentation, I figured the way to generate the best results is for someone to take a track they've recorded (personally), then isolate the vocal stems (their part) and upload that to Suno as the 'sample' (make sure its the raw vocals). Then, complete the verification process. Afterward, upload the full song (stems & all) and select the new voice you created to accompany it. Then paste all of the actual lyrics to that song in the lyrics box. Then tune your settings for the "weird", "style" and "audio" accordingly. I'd recommend tuning weird between 20-30; closer to 30 for more creativity. Then leaving "style" at 50 and pushing "audio" up to 50 to ensure the final output track doesn't deviate too much from your original.
When you complete this process, you should have the Suno version of the song with your voice.
Now - let me temper expectations for a second. Will it sound exactly like you? No. You'll notice the differences. But it will sound enough like you to where you should be able to send it to someone you know and expect them to respond back asking, "Is this you??". Is that good enough for most of ya'll? Probably not. That's not the point of why I'm writing this though.
My point is that this is where that feature is right NOW. It won't remain stagnant. I know for a fact it won't. The pace at which AI research is receiving funding and accelerating is face melting. There are constant improvements in the "speech to text", "speech to speech" and "text to speech" sector of NLP. Look at how many different model generations Suno has released in just the last year alone. I'm saying all that to say...if things head toward where I think they're gonna head...I believe the voice cloning feature will get good enough to where it legitimately does sound like you got in the booth and recorded it yourself. Or at least the average ear won't be able to discern the difference.
And that's gonna be the inflection point. Why? Because pushback gets a hell of a lot harder. At that point, its (if you choose) your own lyrics, your own prompt alongside your own reference track accompanied by your actual voice. The only part the AI is really playing is polishing it up, mastering & engineering it on a level previously only available to engineers. Pay attention to that last sentence. This is a benefit professional musicians get everyday. The only thing AI is doing at that point is democratizing that access.
That now becomes music you could credibly perform. Why not? Its your lyrics and your voice. It doesn't need to sound the exact same way live as it does when you're in the studio. You guys have been to concerts. Does your favorite artist's live performances EVER sound exactly like it does in the studio? You don't give a fuck. You're there to have fun and sing along and watch them perform it. Why wouldn't it be the same for you? Unless you're someone with zero singing ability whatsoever but you've managed to transform a track that has you hitting Adele-esque runs, I doubt your AI track has anything in it you can't solidly perform yourself in a way where it basically hits the same.
Once this transition happens - AI music has to be accepted. People will look at AI as just another DAW plug-in that artists use. Its a computerized enhancement. EVERY fucking artist has that. That's what those big ass mixing and mastering boards in professional studios are there for. To make that artist sound like a million bucks. Once the music reflects the creator, AI music will be accepted.