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.
Requesting "Don't Stop Me Now" by Queen
"Head-Cha-La" just cause Josh hates anime
"Weather Man" Valley of Wolves
I'm still shocked no one has done "Erika"

Your autism has brought us much joy, I thank you for it.

I wanna hear this nigga sing the "Pokerap", just for the fun of it too ngl


edit: the Reshiram song was a great choice! justinRPG stream was one of the first MATI's I ever saw.

Ask and you shall receive fellow Joshcore enjoyer. I've tried to mix the pokerap to make it somewhat like the original. I'll add in more fades etc later on.
 
Hey, I am trying to run the model on my local RVC install and I am running into an error saying "Please select a model and input audio file" when trying to convert. Here is my log;
Code:
.pth file directory: ./models\nullmodel72\nullmodel_e72_s2160.pth
.index file directory: ./models\nullmodel72\added_IVF2626_Flat_nprobe_1_nullmodel_v2.index
loading ./models\nullmodel72\nullmodel_e72_s2160.pth
gin_channels: 256 self.spk_embed_dim: 109
Exception in Tkinter callback
Traceback (most recent call last):
  File "D:\AI\rvcGUI\RVC-GUI-pkg\runtime\lib\site-packages\tkinter\__init__.py", line 1892, in __call__
    return self.func(*args)
  File "D:\AI\rvcGUI\RVC-GUI-pkg\runtime\lib\site-packages\customtkinter\windows\widgets\core_widget_classes\dropdown_menu.py", line 101, in <lambda>
    command=lambda v=value: self._button_callback(v),
  File "D:\AI\rvcGUI\RVC-GUI-pkg\runtime\lib\site-packages\customtkinter\windows\widgets\core_widget_classes\dropdown_menu.py", line 106, in _button_callback
    self._command(value)
  File "D:\AI\rvcGUI\RVC-GUI-pkg\runtime\lib\site-packages\customtkinter\windows\widgets\ctk_optionmenu.py", line 381, in _dropdown_callback
    self._command(self._current_value)
  File "D:\AI\rvcGUI\RVC-GUI-pkg\rvcgui.py", line 443, in selected_model
    get_vc(pth_file_path, 0)
  File "D:\AI\rvcGUI\RVC-GUI-pkg\rvcgui.py", line 241, in get_vc
    print(net_g.load_state_dict(cpt["weight"], strict=False))  # 不加这一行清不干净, 真奇葩
  File "D:\AI\rvcGUI\RVC-GUI-pkg\runtime\lib\site-packages\torch\nn\modules\module.py", line 2041, in load_state_dict
    raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for SynthesizerTrnMs256NSFsid:
        size mismatch for enc_p.emb_phone.weight: copying a param with shape torch.Size([192, 768]) from checkpoint, the shape in current model is torch.Size([192, 256]).
 
Damn, layered vocals are this thing's kryptonite. The vocals sound mildly muffled until 53 seconds in when it clears up. I have no idea why; I swear I didn't touch any settings buttons. I'm thinking that this probably works better with languages, cadences and accents that aren't his own simply because they're not as comparable to what we're used to hearing on MATI, so I wanted to test this with Japanese rap music.
 
Liking Traps is Gay - Trixie

I think this one turned out nearly perfect :)

Tomorrow Belongs to Me - Ian Stuart & Stigger

I couldn't get this one to come out great but its not as bad as these:
Bombs Over England ("Bomben Auf Engeland")

Erika

I am surprised no one has made a cover of God Save the King and Rule Britannia in the name of dear feeder.
I tried both but the problem is the only covers I can find of the song are sung by a chorus in a very opera-esque way which seems to completely skrew up the AI. It sounds really bad. If anyone can find a single person singing one of those in a less opera-y way I think it could work.
Honestly suggestion for everyone dont waste your time trying to do songs with a chorus of singers, you need to find covers with one singer or two at most.
 
A couple of ideas for people who are better at this than I am:

Aqua - Barbie Girl
Linkin Park - Quite a few of them, take your pick
Kanye West - Power (Ralph recently mentioned this song and they're best fwiends now.)
Carly Rae Jepsen - Call Me Maybe
 
  • Like
Reactions: BadGoy1488
Damn, layered vocals are this thing's kryptonite. The vocals sound mildly muffled until 53 seconds in when it clears up. I have no idea why; I swear I didn't touch any settings buttons. I'm thinking that this probably works better with languages, cadences and accents that aren't his own simply because they're not as comparable to what we're used to hearing on MATI, so I wanted to test this with Japanese rap music.
View attachment 5229986

Yeah, it really struggled with a few songs. When I edited the Pokerap I just had to straight up remove entire sections of vocals that were going insane.
 
This old ass song recently became a meme, and right on because it's a great song, but it would be amazing to hear Jawsh sing it..
 
I had to do one more. Unfortunately Fred Durst likes to use excessive chorus filters on his vocals so these dont't come out as clean as I would like. I'm sure someone more tech savvy could get better results. Still pretty funny though.

Limp Bizkit "Break Stuff" - Joshcore

View attachment 5229369
I'm having Woodstock 99 flashbacks.
Shouldn't have charged so much for water.
 
  • Feels
Reactions: suspended animation

K. Flay - CRAZYtown

I like the way this came out, but I'd be tempted to use the -12 pitch for the singing part and maybe splice in the -10 for the more spoken word part. Also "Awe we rearry this mentarry diseased?" lol

EDIT: Version2

 
Last edited:

Peter Lorre Eyes - Electric Ham Sandwich

I like how it struggles with "cope and seethe", sounds a bit like "cope and sneed". Also "Raaaaatttiaaaaa"

Visualizer:

 
Back