Burn/138bpm/Trance

[osiris]

This is where you chop up your original version and discard half the lyrics. I like it, but I think I mixed it too wide.


or
https://open.spotify.com/album/1mZYfozJ ... tB4eSRvXkQ

[wagtunes]

I saw this wasn't getting any love so I thought I would give it a listen. Very well mixed and produced. Love the vocals. I won't ask how you did this because there is no way I could come close. LOL.

Well done. Really enjoyed listening to this.

[osiris]

Fantastic.

[ChameleonMusic]

So, how much of this music is AI-generated?

Just asking as it contains a number of classic AI-algorithm elements, so I ran it through 2 high-level university music dept AI checkers and they said:

AI content: 75% or more
AI content: between 70% and 100%

I then asked YouTube for their opinion and got the following response:

Yes, "Burn" by IllekTrikOne is widely recognized as an AI-generated electronic dance music (EDM) track.

While IllekTrikOne maintains an artist presence on platforms like Spotify, Shazam, and the KVR Audio forum—where the entity engages with the electronic music community—this specific track has been identified in various contexts as a prime example of fully automated AI music production.

[osiris]

Fantastic. More like 50.

[ChameleonMusic]

Fantastic. More like 50.

Not quite sure why you think my question was ''fantastic''?

From my experience when someone admits to 50% in a situation like this it normally means a significant bit more?

I would say from my analysis - at least 70%...possibly up to 100%?

Either way...

You should definitely declare to potential listeners on a forum such as this that you've used AI tech!

Any other approach is simply dishonest!

In simple terms it let's us know upfront that the music has not actually been composed and produced by you, but (at least in significant part) by an algorithm that requires very little / no musical & music tech skill on your part.

[osiris]

It was disclosed on the distribution end. I appreciate the forensic investigation into my obscure trance track.

[ChameleonMusic]

It was disclosed on the distribution end. I appreciate the forensic investigation into my obscure trance track.

First statement above = Irrelevant on a forum such as this as we don't have access to that information!

You have to make a full AI disclosure to your distributor (DistroKid) so I'd be mildly curious about exactly what you declared as AI in this case?

Vocals, lyrics, melody, or instrumentation, certain parts of the track or the entire audio?

The reason for the above disclosure is for transparency on streaming sites - so that they can declare how a particular track was made...can't see that info on your tracks on Spotify?

Second statement above - nothing forensic about it, it just sounded like AI in various ways from the first listen:

Eerily polished at first and then little issues crept in such as - artifacts behind the vocals; loss of some high frequency details at times; issues with the arrangement - particularly lack of any rhythmic tension / variation; micro glitches / dropouts where the algorithm has ''stitched'' together different data-mined ideas / too perfect rhythmic elements with no programmed (or played) human adjustments. etc


...and I then spent about 5 minutes looking a bit deeper.

[Hipster Bales]

...

[osiris]

Please don't talk about me in the third person like I'm your demented grandparent, I did create this.
I wrote the lyrics, I made the demo track, and the AI worked from material I created. It didn’t appear out of thin air. What I said was absolutely true. I did a version of this, wasn't happy with it, so I did an AI assist that made a better track.
I also find it ironic that people are using AI detectors as if they’re definitive proof of something when even the detectors can’t agree. I tested the same track with three different systems and got three wildly different results — 60%, 70%, and 90%. It's probably closer to 40%. I think reactions like this are part of why many DAWs have been cautious about deeply integrating AI features. There’s still a “villagers with pitchforks” mentality anytime someone mentions AI-assisted creativity, even though it's evolving every day.
You wouldn't say: This is a purchased Sample Pack, which I have used before also. I'm just counting the days until some loops someone gets off the Cloud are going to be clocked as AI.
Nobody has to personally like AI-assisted music, but acting like there’s zero human creativity involved just isn’t accurate in my case.

[osiris]

I also think you should temper your judgement. This is a new track I'm working on. It was done ENTIRELY by me. I ran it through 4 AI detectors and only 1 got it correct, with one saying it was 90% AI and one even incorrectly saying the old version of Suno detected. Take of the what you will.

v2.mp3

[ChameleonMusic]

Thought this discussion had run it's course? No idea why anyone had to reopen it...we had a reasonably open discussion...points made...all done?

the AI worked from material I created.

Understood, but my main point above was that you didn't declare this at all!?

so I did an AI assist that made a better track.

Ditto my above point + off course...why didn't you try and make it a better track with your own skills and knowledge?

I also find it ironic that people are using AI detectors as if they’re definitive proof of something when even the detectors can’t agree. I tested the same track with three different systems and got three wildly different results — 60%, 70%, and 90%.

1) 60 / 70 /90% are not wildly different - they point to a clearly significant amount of AI input.
2) Yes, I agree that Free AI music detectors are no proof of anything, but they do err on the side of caution for various reasons.
3) I didn't base my anlaysis on the AI detector at first, but on what I could hear in the track.
4) I didn't use free detectors, but much more sophisticated ones used to detect AI in music student work at uni.

It's probably closer to 40%.

You said 50% last time...hmmm! Even if that was accurate, it's still a helluva lot of the piece ain't you!

I think reactions like this are part of why many DAWs have been cautious about deeply integrating AI features. There’s still a “villagers with pitchforks” mentality anytime someone mentions AI-assisted creativity, even though it's evolving every day.

The reasons for reluctance regarding the integration of AI into DAWs are very complex and very little to do with your silly “villagers with pitchforks” comment...the idea that someone who's been in the industry professionally since 1994 and runs his own music studio is a techno luddite - don't start lashing out...makes you sound ultra-defensive.

Just for the record - one of the many reasons is what you said here: ''it's evolving everyday''...not always in the right ways...there have been some quite bizarre episodes with AI interventions on the mixing front so far...sometimes it learns in a very weird way!

You wouldn't say: This is a purchased Sample Pack

Actually I would (and have on here) if the piece was full of such preset loops/ hits and other musical ideas with not enough input from the person.

but acting like there’s zero human creativity involved just isn’t accurate in my case.

Never said that - just want people to be honest at the start and say they've used AI and how they've used AI so we can listen fully informed.

[Hipster Bales]

Thought this discussion had run it's course? No idea why anyone had to reopen it...we had a reasonably open discussion...points made...all done?

the AI worked from material I created.

Understood, but my main point above was that you didn't declare this at all!?

so I did an AI assist that made a better track.

Ditto my above point + off course...why didn't you try and make it a better track with your own skills and knowledge?

I also find it ironic that people are using AI detectors as if they’re definitive proof of something when even the detectors can’t agree. I tested the same track with three different systems and got three wildly different results — 60%, 70%, and 90%.

1) 60 / 70 /90% are not wildly different - they point to a clearly significant amount of AI input.
2) Yes, I agree that Free AI music detectors are no proof of anything, but they do err on the side of caution for various reasons.
3) I didn't base my anlaysis on the AI detector at first, but on what I could hear in the track.
4) I didn't use free detectors, but much more sophisticated ones used to detect AI in music student work at uni.

It's probably closer to 40%.

You said 50% last time...hmmm! Even if that was accurate, it's still a helluva lot of the piece ain't you!

I think reactions like this are part of why many DAWs have been cautious about deeply integrating AI features. There’s still a “villagers with pitchforks” mentality anytime someone mentions AI-assisted creativity, even though it's evolving every day.

The reasons for reluctance regarding the integration of AI into DAWs are very complex and very little to do with your silly “villagers with pitchforks” comment...the idea that someone who's been in the industry professionally since 1994 and runs his own music studio is a techno luddite - don't start lashing out...makes you sound ultra-defensive.

Just for the record - one of the many reasons is what you said here: ''it's evolving everyday''...not always in the right ways...there have been some quite bizarre episodes with AI interventions on the mixing front so far...sometimes it learns in a very weird way!

You wouldn't say: This is a purchased Sample Pack

Actually I would (and have on here) if the piece was full of such preset loops/ hits and other musical ideas with not enough input from the person.

but acting like there’s zero human creativity involved just isn’t accurate in my case.

Never said that - just want people to be honest at the start and say they've used AI and how they've used AI so we can listen fully informed.

+1