The NIN and Trent Reznor AI Voice thread

[simonn]

Banged and Blown Through with Saul's vocals replaced by Trent, using AI. Didn't know this was a new form of entertainment, but it's astoundingly good!

[Toadflax]

Banged and Blown Through with Saul's vocals replaced by Trent, using AI. Didn't know this was a new form of entertainment, but it's astoundingly good!

Strange choice when that already exists. It would be cool if AI could strip Trent's live vocals, clean them up, and lay them over the studio track, though!

[leftshoe18]

I've tried ripping Trent's vocals from a few different live recordings of the song with the intention of doing just that. The problem is I can't get a clean rip of the vocals that doesn't also have a ton of crowd noise.

[thelastdisciple]

Strange choice when that already exists.

Somehow i totally forgot about this, i always thought he brought Saul out with him whenever something from Tardust came up.

Which.. IMO despite TR's hand in producing a lot of the tracks, having Saul there would always be the more apropos move considering how much of the lyrical subject matter for the album is from the point of view/experience of a black artist.

"Banged and Blown Through" isn't "Trigger" though i suppose.

Regardless, the A.I. what if TR vocal stuff just rubs me the wrong way in general.

[Toadflax]

I've tried ripping Trent's vocals from a few different live recordings of the song with the intention of doing just that. The problem is I can't get a clean rip of the vocals that doesn't also have a ton of crowd noise.

Right. An AI could theoretically do a blend of keeping as much of the original track as possible but then doing AI stuff to smooth out the edges, filter out crowd noise, adjust for any tempo changes, etc. for the rest. AI fully trying to recreate someone's voice still sounds pretty wonky at the moment, but if it could be used to take an existing track and clean it up (thus bypassing the many hours it would take a human to do this), that would be cool.

[nihilmatters]

Strange choice when that already exists.

Well then, perhaps you'd prefer the AI version of a Dua Lipa cover, haha.

[Elrickooo]

Well then, perhaps you'd prefer the AI version of a Dua Lipa cover, haha.

lol these are so much fun. Please keep posting them

[paul_guyet]

Well then, perhaps you'd prefer the AI version of a Dua Lipa cover, haha.

Shit. THIS IS WHAT STROBELIGHT COULD HAVE BEEN.

[tony.parente]

Hell yeah, we'll just make our own NIN record.

[paul_guyet]

Okay...waaaaay back when when Strobelight was "announced", I wrote a few songs. They. Are. Awful. Exactly what I imagined something like the actual lyrics to Strobelight to be. Would anyone actually be interested in using them to create...something?

Let me know.

[leftshoe18]

Hopping on the AI Trent bandwagon. Skin of a Drum by Saul but with AI Trent AND I put it over The March from Deviations. The verses are kind of buried in the mix but that's by design. I was going for something like I Do Not Want This where the spoken word portions are pretty drowned out by the music around them.

[Jackryder2025]

Really shocked at how good these sound. AI is going wild these days. You really could make a whole album out of this and it would be fun to listen to. Sure its not official but, The Ai voice is honestly really good

[shagg_187]

So uhh... how are y'all making this? Interested in the tech and apps being used here. Would love to see the process flow.

[Wretchedest]

Well this is equal parts impressive and depressing

[leftshoe18]

So uhh... how are y'all making this? Interested in the tech and apps being used here. Would love to see the process flow.

RVC2 and So-Vits-SVC are the two AI voice conversion tools I use. RVC is way better in my opinion but there are some models I was only able to find with sovits. It takes a bit to set up but I think it's a super fun tool to play around with.

[henryeatscereal]

Has anyone posted this? Unironically i loved it

[shagg_187]

Has anyone posted this? Unironically i loved it

LOL! He/She made an entire album of this.

This is getting out of hand (that feeds). So hilarious.



(VOMIT)

[Jackryder2025]

When i read those titles i was hoping it would be The Post malone song rockstar lol Someone should make that happen!

[DVYDRNS]

I hate this.

[Zur En Arrh]

I hate this.

Strap in, shits about to get super weird....and bad....

[Frozen Beach]

Not Trent, but this is one of the funniest things I've ever heard.

[Jackryder2025]

I found a pre made trent Model, and learned how to use RVC, i am going wild with this lol
So far i have done Deftones digital bath and post malone rockstar!

[AVirtualMemoryOfficial]

This is all extremely disheartening.

[benoïde]

I concur...
This is digital impersonation at its finest/worst. Having fun is one thing - I think of the mashup/shredds etc scene - but taking the dna of a voice seems wrong to me, and I have a hard time understanding how this is more fun or interesting or providing any kind of elevation than creating something of your own.

[ghostaustin]

This shit freaks me out.

[fillow]

I refuse to even touch with a 10-foot pole the play button on any of these videos.

I heard some examples of this stuff at the pub quiz that I often play in my city (we were supposed to guess whose voice is being faked) and boy it was ugly.

[nihilmatters]

I'm all for the cautious path of neo-luddism for the most part, but this AI tech will be quite saturated in our daily culture within 10 years times.

Deep fakes will abound on TV, the internet, the news, movies, the radio, etc. Even the next iPhone software release has been announced to contain tech that can simulate the users voice with just 15 minutes of audio training. We can decide not to partake now but it will definitely not slow down the oncoming cultural saturation of this tech.

It will be disheartening when AI can drop a "new" NIN album in 5 minutes that will be almost indistinguishable from The Source.

I hope in the future there will be an intense desire for the next generation of kids, in an act of social rebellion, to pick up physical instruments again to form a REAL band.

Until then...mentally prepare yourself for AI generated 3D hologram concerts performed by dead celebrities.

[Jackryder2025]

Ai is not that good yet, you can still tell its being done by a computer, and i really pushed the AI, Depending on the song it will struggle to sing it properly and other times it sounds spot on. It dosent do Screaming and some things very well, But yeah i can see it getting to the point where they can just make full blown movies or music videos just using AI, but it still has a long way to go, On top of that alot of programs you have to pay for. So copyright and having permission to use someones voice still comes into play, thats why i have not uploaded mine

[shagg_187]

The future of music will be artists licensing their voice to the industry, and raking in checks.

I can already see someone like Britney Spears who is somewhat tired of it all to lend her voice to a recording label, who will just produce the music and use her model to enhance the tracks. Using Britney as an example because her latest Elton John remix track doesn't even sound like her anymore. So using a bulletproof AI to make it sound like her is going to be super easy.

Imagine folks like The Beatles giving their voice for future tracks. They will all be dead soon, so they label don't give a fuck, and the family who wants to see a couple millions will happily sign the deeds.

Many may revolt, but once the checks start coming, and the artists start getting older, this will become a hideous norm.

People used to joke about 2Pac being alive. He will be once they start producing albums with his voice.

What's interesting, you will start seeing bedroom artists with great production skills but zero vocal skills to use AI to blend voices and not reveal that it's AI, and that they stole the voice. God knows the record label would love to hear a new rocking tracks with Chris Cornell AI being featured.

Y'all see the shitty vision? Incomplete albums and demos will now be completed. Two artists who never got along will now get along.

Art.

[tony.parente]

The future of music will be artists licensing their voice to the industry, and raking in checks.

I can already see someone like Britney Spears who is somewhat tired of it all to lend her voice to a recording label, who will just produce the music and use her model to enhance the tracks. Using Britney as an example because her latest Elton John remix track doesn't even sound like her anymore. So using a bulletproof AI to make it sound like her is going to be super easy.

Imagine folks like The Beatles giving their voice for future tracks. They will all be dead soon, so they label don't give a fuck, and the family who wants to see a couple millions will happily sign the deeds.

Many may revolt, but once the checks start coming, and the artists start getting older, this will become a hideous norm.

People used to joke about 2Pac being alive. He will be once they start producing albums with his voice.

What's interesting, you will start seeing bedroom artists with great production skills but zero vocal skills to use AI to blend voices and not reveal that it's AI, and that they stole the voice. God knows the record label would love to hear a new rocking tracks with Chris Cornell AI being featured.

Y'all see the shitty vision? Incomplete albums and demos will now be completed. Two artists who never got along will now get along.

Art.

Super, super, super unpopular opinion but that all seems absolutely tight. I know i'm wrong about this, but hearing new linkin park songs with Chester, or a new NIN record from the broken era, or ACSS part 2 would be neat as fuck even if it's void of a soul and completely fake.