Joshcore - AI generated covers of Null

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
Slightly offtopic, but the slobbermutt in the OP looks very cute contrasted to the slightly unerving one Josh's profile has.
1690402705687.png
 
Gradio still isn't working for me, so here's what resulted from wandering aimlessly in my Silent Hill folder earlier.

Hometown


Shot Down in Flames


Hell Frozen Rain (acapella opening clip)


Your Rain (first verse, it immediately went to shit after)
 
Gradio has stopped working for me. I keep trying to close it out, clear cookies and reopen from the link in the OP, but it keeps automatically prepping the downloads or whatever without me clicking the play button, and it doesn't work whenever that's the case. Sometimes it will download the AI model and not process the files after I upload vocals, but most of the time it just says "There's been an error." where it should say "success" when getting it to download the Google Drive file of the AI model.

Is anyone else getting this issue or knows why it's tarding out?
I think I may have gotten too cocky with my requests
 
  • Feels
Reactions: Big Mommy
Is anyone else getting this issue or knows why it's tarding out?

I think that is what happens when you have hit the limit. Should reset daily. Also sounds like it is variable, so some days you might be able to process more than others.

I can't believe the skoop diddy poop song @WelfareNiggerQueen posted is a real Kanye song, what the fuck lmao
 
Alright tried a few more, finding songs with no backing vocalists is tough

Cheers Theme Song, I think every Kiwi that listens in to MATI can relate. Came out alright except there was backup vocals the AI really can't get rid of, so bit of robot ear rape. Also way ahead of it's time for calling out a husband becoming a tranny, at about 1:14


Smiling Face - James Taylor, this one actually turned out pretty good I think. Not a lot to interfere with the vocal extraction and it's closer to Null's range


Everything will be okay - Helluva Boss, cause why the fuck not


Not a song, but it would be pretty kino for some to do a Josh cover of this


View attachment 5235172
 
Last edited:
Last edited:
I think that is what happens when you have hit the limit. Should reset daily. Also sounds like it is variable, so some days you might be able to process more than others.

I can't believe the skoop diddy poop song @WelfareNiggerQueen posted is a real Kanye song, what the fuck lmao

You can swap google accounts and log in to a new one and use the same google drive links to keep going after the limit.
 

Ken Ashcorp - What PS2?

I had to beat the AI with a stick to get around Ken's use of layering his own voice on top of itself for some sections. At points Null sounds extra autistic, and even mispronounces a word, but it's still good enough for me.

At some point I'm going to get bored of this, but considering it's just ~20 minutes of time to make and I only do it once per couple of hours, it could take a few more days.
 
I should also mention if you're familiar with Melodyne or have it with you, you could either fix the pitch issues before or after converting the audio to AI. I actually use this in tandum with "Mangio-crepe with the hop length set really high" so that if Josh sounds like he's slurring the pitch a bit and doesn't hit the note I could always just fix it in-post. I should've also mentioned I use both an rmvpe and mangio-crepe exported audio file so that if one or the other doesn't sound right, I could always just splice in parts from one another that sound better, though usually I use an rmvpe output as a base and splice in mangio-crepe parts whenever I feel it is necessary.
 
I present Puccini's Nessun dorma from Turandot.


Some kiwi who knows video editing and face swaps (AI I think could?) could do something with the end of Sum of All Fears
Have Ralph and Null be the Russian and US president, with America First as assassination target or something.
 
  • DRINK!
Reactions: Waifu Days
Of course, I'm a dumb lazy piece of shit.
I did it. I was experimenting a bit and I think some of the lisping is from background vocals. The file I made for Boys Be Free, whenever Melissa Fahn's vocals were solo, the lisp was gone. I found it easier to record myself singing. With that said, I might make a few recordings of my own to Jershify.

Song - You Think You're a Man performed by Divine


Song - Boys Be Free (JershxChantal Love Theme :feels:)


I know this is a selfish request but I really want a vocal model of Ethan Ralph.
 
Last edited:
Do you think it would be worth making a thread just for storing the voice models?

I've got the Cyrax one pretty spot on and my attempts with the Keffals and Metokur one are coming along.
I can leave links to the models in the Joshcore OP and just direct any OC to the general AI thread. I'll go ahead and put Ralph and Cyrax in right now.
 
I can leave links to the models in the Joshcore OP and just direct any OC to the general AI thread. I'll go ahead and put Ralph and Cyrax in right now.

Perfect. Once I'm happy with the other voice models I'll upload them to the farms and drop you the thread links.
 
Back