fredmagnus said:

lustmealways said:

This Dick and Ubeknownst 2 U are very similar in one very important way

Just as you'd have to be mentally retarded to think This Dick was real, you would have to be mentally retarded to think Unbeknownst 2 U is fake.

If this is proven wrong I will leave this forum and never post again.

No need to insult anyone. Nor to leave this forum especially when there are already too few fans left here. Time will tell anyway.

We live in an era where everyone should be waiting for a second source/confirmation before taking anything for granted. Believe me, we're getting very close (if we're not there already) to the point when no one will be able to tell what's real and what's not.

Indeed. It sounds real to me but how could I be sure?

Either way, any argument saying "there's no way AI is capable of..." is very fragile in 2026 (and it'll get more and more fragile with each month).

AI denialism is something that worries me a lot, because AI should be on the forefront of each and every political debate at the moment, only it isn't, and the reason it isn't is because the vast majority of people, journalists and politicians believe it's not going to change much (while it actually is about to change everything), so it just happens and we keep pretending it doesn't and doing nothing about it.

Prince had different voices. It is possibile to train an IA to mimic one of this voices, harder create a song, like this one, where Prince changes his tone and pitch multiple times. I think it is possibile for an IA to do this, but it is very hard. The fake album or the fake song I listened on Youtube created with IA are IMHO unlistenable: the music is arranged in an awkward and unoriginal way and the "Prince" voice is flat and poorly articulated. In this song instead the Prince's voce is rich as we know. Oblivious I could be wrong.

When I (tried to) listen to the obvious AI collection a few weeks ago, it was just bad, sounding like a prince-like emulation that had no soul and completely missed what makes Prince special.

When I listen to this, it sounds genuine. I believe it is real. Also, to me, it sounds more in line with 2014 era recordings than earlier eras, so, I won't be surprised if that timeframe turns out to be accurate as well. Regardless of the year though, I do think it's genuine Prince.

There should be some sort of AI music detector same way they can tell if something you wrote was rendered by AI (Chatgpt etc...)

there's far, far too many intricacies in his phrasing and voice here for this to be fake. again, this is without a doubt and unquestionably...

1. real

It is clearly real - and I really hope we hear the full thing soon, because I hate listening to snippets.

lustmealways said:

there's far, far too many intricacies in his phrasing and voice here for this to be fake.

Wait until you hear Bloody Mouth.

ShellyMcG said:

It's really sad that a new Prince song drops and the main topic of conversation is debating whether or not it's A.I. instead of actually discussing the song itself.

This is kind of becoming the main topic of convo with everything. And it's only beginning...

lustmealways said:

there's far, far too many intricacies in his phrasing and voice here for this to be fake. again, this is without a doubt and unquestionably...

1. real

Like I said, I tend to agree with you, but repeating it over and over isn't gonna make it more real than it is if it indeed is real. Just go with the fact that people doubt the song's authenticity: it's the healthy reflex to have nowadays. The main reason I assume it's real is because someone with some insider info said the rumor is that it's real, not because of what I hear.

.

I try not to listen to AI music, so I cannot say exactly what the state of the tech is as of May 2026, but I was rather impressed by what AI can already do with the few deepfake songs I happened to hear.

Either way, even if statements such as "this is too good (or too whatever) to be AI" are valid today, they probably won't in a matter of years, if not months. So we better brace ourselves because sooner or later, The Divine is gonna pop up online and we'll be at a loss to know whether it's the real song...

databank said:

bluegangsta said:

It is clearly real - and I really hope we hear the full thing soon, because I hate listening to snippets.

Same here

ShellyMcG said:

databank said:

This is kind of becoming the main topic of convo with everything. And it's only beginning...

Yeah. And it's fucking tiresome

It's what happens when the most disruptive tech is given for free (or cheap) for everyone to use. It's insane because you'd think we'd have learned some lessons from what social media did to our societies, which was very clear in 2022, before unleashing an even more disruptive tech (and it's disruptive in many other ways, too: the risks to cybersecurity alone are really scary).

.

I guess we never learn.

This song snippet sounds incredible; vintage Prince. If this is AI, the music industry is about to collapse... and soon. If you can make a fake that sounds this great, it won't be long before people are creating their own music emulating their favorite artists and then saturating the airwaves/web with their creations. Maybe or maybe not for profit.

pdiddy2011 said:

Boydie said:

pdiddy2011 said:

This song snippet sounds incredible; vintage Prince. If this is AI, the music industry is about to collapse... and soon. If you can make a fake that sounds this great, it won't be long before people are creating their own music emulating their favorite artists and then saturating the airwaves/web with their creations. Maybe or maybe not for profit.

That is EXACTLY where we are heading I am afraid, and I think it is already too late to put the genie back in the bottle AI + Quantum computing will make your prediction something super easy to do...but it then won't be long before we can each create our very own "Purple Rain 2", with any songs (existing or AI created) where we can choose the storyline, scenes, characters, what they do ( ) etc etc. The only limit will be our imaginations, assuming AI does not rot our brains and stop future generations from having an imagination or any creativity Some will love, some will hate it - but it is coming... ...so the Estate should release everything and do it NOW!!!! (Insert the "Now" loop from the Gold Experience here!

Oh wowowowowowow, we keep telling Mr. McMillan that he's lying about us fans asking for EVERYTHING to be released NOW and you just do that. Not good. Not good. Obviously no one's asking for this if only because it's impossible. He's trying to make us pass for a bunch of crazies, don't give him ammo.

Regarding AI, there's a school of thought that assumes the recommandations algorithms might in a not so far future be turned into creation algorithms. Instead of recommending music, films and shows that it assumes you like, the algos may just create those based on your tastes, leading to each and every one of us consuming music and pictures/shows no one else will hear nor see. The Internet bubbles theory thus being pushed to its worst extremes.

Sure, some people may prompt their own content themselves, but most probably will just go with what's tailored for them by the machines.

Either way, for over a decade I kept telling people "don't do your own fanmixes or, at least, don't share them online" and everyone was laughing at me. Now, with AI, everything I was fearing is happening, only worse.

databank said:

Ok ppl i gotta say my initial gut feeiling is that this one is indeed REAL.
Still hope it is.
HOWEVER i asked AI to check if it's AI (perplexity. prince would have said Purplexity i guess)
and it asked me for directions etc, i told it to do deepest analysis possible. it took about an hour.
here is the text i got back - make of it what you make. bottom line most likely a FAKE.
i claim no knowledge in audio analysis im just the messanger.
I wish someone here with capabilites to analyze audio would double check these findings.
Peace


Short answer: I’m finished — the forensic analysis finds strong indicators the YouTube upload is a likely AI/fan recreation, not an authenticated Prince master.

Key results (concise)

• Uploader provenance: the video is an isolated YouTube upload with no label/estate attribution and no traceable collector/source in the comments or reposts; fan sites note the song title exists only as handwritten lyrics auctioned in 2024, not an available master.

• Automated AI‑detectors: two independent detectors flagged the lead vocal as likely synthetic — detector A returned a high synthetic-likelihood score on multiple 10–15s excerpts, and detector B gave a strong synthetic flag on the same segments (I ran detectors on several chunks to avoid single-segment bias).

• Spectral analysis: the spectrogram shows unusually clean harmonic bands, near-uniform breath spacing, and transient shapes consistent with voice‑conversion outputs rather than raw tape/analog vocal takes; there are also codec cutoffs typical of a compressed upload rather than multi‑generation analog tape noise. These patterns are more consistent with a synthesized/processed vocal than an original Prince studio track.

• Vocal micro-features: after instrument-vocal separation, breaths and consonant articulations appear unnaturally consistent in amplitude and timing, vibrato is regular and slightly quantized, and small sibilant/aspiration artifacts lack the micro-variations expected from an organic performance — all common signs of voice cloning or heavy editing.

• No session match found: I searched Prince collector communities and Prince Vault; there’s mention of the title but no matching lossless files, session notes, or collector trace linking this audio to any known source. Collectors typically surface FLACs or session IDs when real vault audio appears, and none surfaced here.

What this means (interpretation)

• Likelihood: based on combined evidence (uploader trace, multiple AI-detectors, spectral/vocal micro-feature analysis, and lack of collector provenance), the best judgment is: “likely AI/fan recreation.” This is not absolute proof, but it is a strong forensic conclusion given the available data.

• Residual uncertainty: a very skilled impersonator plus heavy post-production could mimic some of these signs, and automated detectors are not infallible — matching session files from a trusted collector or an estate confirmation would override the forensic indicators.

Attached evidence (what I generated)

• Spectrogram and waveform images (showing harmonic structure, frequency cutoffs).

• Extracted vocal-only clip spectrograms and annotated notes on breaths/phonemes.

• Detector outputs (screenshots/logs) from two AI-voice classifiers showing the per-segment likelihoods.

Next steps you can take (recommended)

• Ask the Prince Vault community: post a 30–60s clip and ask collectors for lossless/session matches — they can give definitive provenance if a FLAC or session source exists. I can draft the post for you.

• Request estate comment: if you need definitive legal/authenticity confirmation, the Prince estate/label is the final arbiter. Public channels sometimes respond to clear provenance claims.

• If you want, I can publish the forensic images and full annotated report here (expanded with the generated images and detector logs) and draft a short Prince‑Vault post to test provenance.

If you want the full report package (images, detector logs, annotated notes, and a short forum post draft), say “Send full package” and I’ll attach everything in the next message.

[Edited 5/13/26 18:44pm]

[Edited 5/13/26 18:45pm]

databank said:

leading to each and every one of us consuming music and pictures/shows no one else will hear nor see.

We're kind of doing something similar when we vibe-code pieces of software that are very, very specific to our needs. Instead of writing more general/universal software. I do this. Many people do this.

nirforward said:

-

There's too many red flags. whats the purpose of the "leak"? Theres nothing suggesting its a message to so called elite traders or the estate like that last leak, what was it, all the kings horses?.

If someone was sitting on an unknown track and wanted to just leak it anon for the fan community why wait until 2 weeks after 21/04 and then put out a truncated copy

This is 110 percent

1. Real

nirforward said:

Ok ppl i gotta say my initial gut feeiling is that this one is indeed REAL.
Still hope it is.
HOWEVER i asked AI to check if it's AI (perplexity. prince would have said Purplexity i guess)
and it asked me for directions etc, i told it to do deepest analysis possible. it took about an hour.
here is the text i got back - make of it what you make. bottom line most likely a FAKE.
i claim no knowledge in audio analysis im just the messanger.
I wish someone here with capabilites to analyze audio would double check these findings.
Peace


Short answer: I’m finished — the forensic analysis finds strong indicators the YouTube upload is a likely AI/fan recreation, not an authenticated Prince master.

Key results (concise)

• Uploader provenance: the video is an isolated YouTube upload with no label/estate attribution and no traceable collector/source in the comments or reposts; fan sites note the song title exists only as handwritten lyrics auctioned in 2024, not an available master.

• Automated AI‑detectors: two independent detectors flagged the lead vocal as likely synthetic — detector A returned a high synthetic-likelihood score on multiple 10–15s excerpts, and detector B gave a strong synthetic flag on the same segments (I ran detectors on several chunks to avoid single-segment bias).

• Spectral analysis: the spectrogram shows unusually clean harmonic bands, near-uniform breath spacing, and transient shapes consistent with voice‑conversion outputs rather than raw tape/analog vocal takes; there are also codec cutoffs typical of a compressed upload rather than multi‑generation analog tape noise. These patterns are more consistent with a synthesized/processed vocal than an original Prince studio track.

• Vocal micro-features: after instrument-vocal separation, breaths and consonant articulations appear unnaturally consistent in amplitude and timing, vibrato is regular and slightly quantized, and small sibilant/aspiration artifacts lack the micro-variations expected from an organic performance — all common signs of voice cloning or heavy editing.

• No session match found: I searched Prince collector communities and Prince Vault; there’s mention of the title but no matching lossless files, session notes, or collector trace linking this audio to any known source. Collectors typically surface FLACs or session IDs when real vault audio appears, and none surfaced here.

What this means (interpretation)

• Likelihood: based on combined evidence (uploader trace, multiple AI-detectors, spectral/vocal micro-feature analysis, and lack of collector provenance), the best judgment is: “likely AI/fan recreation.” This is not absolute proof, but it is a strong forensic conclusion given the available data.

• Residual uncertainty: a very skilled impersonator plus heavy post-production could mimic some of these signs, and automated detectors are not infallible — matching session files from a trusted collector or an estate confirmation would override the forensic indicators.

Attached evidence (what I generated)

• Spectrogram and waveform images (showing harmonic structure, frequency cutoffs).

• Extracted vocal-only clip spectrograms and annotated notes on breaths/phonemes.

• Detector outputs (screenshots/logs) from two AI-voice classifiers showing the per-segment likelihoods.

Next steps you can take (recommended)

• Ask the Prince Vault community: post a 30–60s clip and ask collectors for lossless/session matches — they can give definitive provenance if a FLAC or session source exists. I can draft the post for you.

• Request estate comment: if you need definitive legal/authenticity confirmation, the Prince estate/label is the final arbiter. Public channels sometimes respond to clear provenance claims.

• If you want, I can publish the forensic images and full annotated report here (expanded with the generated images and detector logs) and draft a short Prince‑Vault post to test provenance.

If you want the full report package (images, detector logs, annotated notes, and a short forum post draft), say “Send full package” and I’ll attach everything in the next message.

[Edited 5/13/26 18:44pm]

[Edited 5/13/26 18:45pm]

I think it's real, but obviously that's merely my opinion, and what I'm about to argue is NOT that this evaluation is incorrect. Rather it's just some counter points to some of the analysis:

Uploader provenance can be chalked up to someone wanting to remain anonymous.

If it really is from 2014, Spectral analysis results could be due to the processes used by Joshua Welton around that time and would be consistent with his "laptop" production techniques. The same could be said for the vocal micro features.

No session match found only means that those that compile such information simply don't know yet, and are consistent with the sharer choosing to remain anonymous and not reveal sources.

It's an interesting analysis but I find it unconvincing. That said, thank you for sharing these findings!

[Edited 5/14/26 2:55am]

djThunderfunk said:

The piano plays like it's 1981-82 and the synths play like it's 1996-97. To my ears at least. Possible of course, but at the same time suspicious.

olb99 said:

databank said:

leading to each and every one of us consuming music and pictures/shows no one else will hear nor see.

We're kind of doing something similar when we vibe-code pieces of software that are very, very specific to our needs. Instead of writing more general/universal software. I do this. Many people do this.

I don't think as many people code as watch movies/shows or listen to music, and I'd say it's also different in the sense that coding is, basically, designing a tool. Art is common culture, common knowledge, common values and shared experiences.

.

I remember this interview of a French music journalist—and that was already about 10 years ago— who said when he began his career in the early 90s, whenever he'd meet a peer and they'd go like "do you know such or such new act?", it wouldn't take 10 acts names, even among specialists, for the other to be like "yeah, I know this". Now, he says (and "now" is 2015 or so), we can trade dozens upon dozens of names and we can't find something we both know. It can only have gotten worse ever since. Everyone's lost in their own little niche. Same goes with my childhood: when I was in school, there would be something on one of our 5 TV channels and the next day, everyone in class had watched it, and we talked about it. Not to say this doesn't happen at all anymore (obviously, many people watch things like Stranger Things or Squid Games or Black Mirror the minute the episodes come out), but it doesn't happen as much as it used to.

.

What does it mean? That what happened to political views also happens to a shared common cultural landscape: everyone is alone in their own bubble or, at best, in a small bubble of fewer people than before who share their references. More records probably get released every day that in a whole year when I was in high school. And that was already happening without AI, which already consitutes 20 to 50% of new music uploads according to various sources.

.

In a way it's cool: one can easily find exactly what they like. One the other hand, IDK, I'd say common knowledge and cultural references aren't useless to cement a society. It's a way to find common ground, to share values. Hearing and seeing things we don't immediately like or that we aren't looking for is also a way to reinforce curiosity and discover new ideas. Music and movies certainly forged the X's and Y's societal values in the 70s, 80s and 90s. IDK how it's going to help us find common ground if we don't have anything in common anymore. And if we don't find common ground, division is easier to seed than ever before. What one doesn't know is often suspicious and easier to turn into a target.

[Edited 5/14/26 14:16pm]

If people keep saying this is AI i'm going to keep saying it's

1. 100 percent
2. Real

If you guys wanna know how bad AI is at knowing things, just start asking it about something you know a lot about. Hell, ask it about Prince stuff and I don't think it'll be long before it totally veers into lala land.

lustmealways said:

If you guys wanna know how bad AI is at knowing things, just start asking it about something you know a lot about. Hell, ask it about Prince stuff and I don't think it'll be long before it totally veers into lala land.

Use the same when judging "news" sources as well. It always amazes me when someone calls out a news article on a subject, like Prince for example, as being innacurate, and then proceeds to totally believe the same source on a subject they know less about.