future deepfakes

sufi

lala
Apparently:
https://www.ft.com/content/6f022306-2f83-4da7-8066-51386e8fe63b / https://archive.is/20230809134951/https://www.ft.com/content/6f022306-2f83-4da7-8066-51386e8fe63b#selection-2297.58-2313.146 said:
Frank Sinatra’s voice has been used on a version of the hip-hop song “Gangsta’s Paradise” while Johnny Cash’s has been deployed on the pop single “Barbie Girl”. A YouTube user called PluggingAI(opens a new window) offers songs imitating the voices of the deceased rappers Tupac and Notorious B.I.G.

and
https://www.ft.com/content/6f022306-2f83-4da7-8066-51386e8fe63b / https://archive.is/20230809134951/https://www.ft.com/content/6f022306-2f83-4da7-8066-51386e8fe63b#selection-2297.58-2313.146 said:
when an AI-produced song that mimicked the voices of Drake and The Weeknd went viral online. Universal Music, home to Drake, Taylor Swift and other popular musicians, had the song removed from streaming platforms over copyright infringement. Drake in April slammed(opens a new window) another song that used AI to mimic his voice, calling it “the final straw”, while rapper Ice Cube has described such cloned tracks as “demonic”.

Is it surprising that Universal are so slow on the uptake that they only just thought of trying to monopolise this now? Any artist who has ever used autotune is surely ethically bound submit to the deepfakery - their dilemma is similar but different to the one we discussed back when dissensus started when mp3s threatening the biz's bottom line - the tech is out of the box and on the loose so inevitably some compromise will be reached.

Potential for creativity (after the initial wave of gimmicks and provocations) is colossal? already beyond what we can imagine at this moment,

(& at the same time the tech is advancing so that voice tracks can be isolated from the rest of a song, which if i understand right was not possible previously unless you had the original tapes) all those fantasy remixes that never were will soon exist

this thread for requests as well as entertaining gimmicks
 

wild greens

Well-known member
(& at the same time the tech is advancing so that voice tracks can be isolated from the rest of a song, which if i understand right was not possible previously unless you had the original tapes) all those fantasy remixes that never were will soon exist

This can be done pretty easily with web-based apps already, i.e lalal.ai or plenty of others. You do get some legacy audio in the background of some but most are relatively clean.
You could actually do this already in logic or protools but it was a massively pain-staking process for the most part, slightly less if you had the instrumental

*

Regards AI using other people's voices it is more difficult than they're letting on- at present with commercially available tech you're only really able to mask your own voice with an ai skin as opposed to getting the ai to sing convincingly

I.e the "ai drake songs" are someone using his flow over a ripoff of a 40 beat then re-skinning their own voice through a drake filter

The idea of an AI version of drake- "computer make me a drake song about the river lea"- is nowhere close yet, at least with commercially available tech. The google ai "music maker" beta was comically shit once you took it out of the 4 or 5 overtrained comfort zones

There was an article in the FT yesterday about google & universal licensing artist melodies and voices for AI songs, though. Which is probably the first step into the real mire
 

sufi

lala
There was an article in the FT yesterday about google & universal licensing artist melodies and voices for AI songs, though. Which is probably the first step into the real mire
yeah that's the article i linked :ROFLMAO: they are a bit slow if they didnt predict this already - maybe the tech is not there yet, but it will get there
 

mixed_biscuits

_________________________
The year is 2024. I park my hoverboard outside HMV and bound in on my Bounce-o-nikes, throwing a quick 'Hail President Sufi' to the staffer by the door who responds in kind. The shop is void of all product other than a glittering booth in its centre. I bounce to the booth like tigger on coffee and hurriedly key in my choices; knowing I have to take my German literature exam tomorrow, I ask for Goethe's Faust rapped by Eminem featuring Scooter. I then ask for the train timetable to the exam hall in the neighbouring town in the style of Opeth before plumping for an ambient album by Dissensus favourite PJ Proby singing the Arctic Monkeys with Lemon Jelly producing. The booth cogitates noisily for a couple of moments before spitting out a PlexiVinyl for my portable player. I leave the shop without paying, knowing that facial recognition will allow me to be remotely tortured later in the day by a fellow internet user, paying the cost of the music for the privilege of hearing my whimpering.
 

sufi

lala
ingenious, but none of them add much to the sum of their parts tbh?

maybe the more transgressive combinations are getting squished by the owners,

but maybe soon there might be tunes that are better than an irl cover version by the real artist
 

linebaugh

Well-known member
ingenious, but none of them add much to the sum of their parts tbh?

maybe the more transgressive combinations are getting squished by the owners,

but maybe soon there might be tunes that are better than an irl cover version by the real artist
the kanye one is interesting because it sounds just like him but is also singing better than he ever could. I imagine being able to hear yourself sing perfectly would be incredibly useful for improving your skills.
 
Top