Joshcore - AI generated covers of Null

  • Happy Easter!
Listen I love all of you but going through this thread I am reminded how tone deaf the average person is.

1. If you shift Josh’s voice in either direction before processing it, please for the love of God shift the instrumental track the same number of semitones when combining the two in Audacity. This is dead simple to do.
2. If you have a song with backing vocals and are using Ultimate Vocal remover, go to settings and download Karaoke 5 under the VR options or really any of the other Karaoke models. These will attempt to put backing vocals and harmonies into the instrument track. Read the guide from the OP for more specifics. The Karaoke models are not perfect, but if you isolate the lead singer, you will avoid most of the random pitch shifting where the AI processing tries to hunt down the melody. Plus you give Josh some cool buddies to sing with him in the split instrumental track. I am going to do this with one or two of the ones I’ve already posted.
3. If anyone has tips for improving enunciation, please share. I’ve dicked around in the Collab settings and basically nothing gives obviously better results. So far the clarity of Nulls words is just a function of how clearly the original singer enunciates, and nasally singers make it worse.

Thank you for coming to my rant. Rock on Joshcore friends.
 
I hope someone plays this at their wedding, very beautiful:

Sarah MacLachlan - Arms of an Angel

EDIT: Two more tracks, I had to sing the vocals myself as the AI struggled.

ABBA - Dancing Queen

ABBA - Fernando
 
Last edited:
...dear god.


An extended version without the recorder accompaniment

Gene Wilder - Pure Imagination


Oompa Loompas - Oompa Loompa Song

Gloria Gaynor - I Will Survive

I've been working on a list of song links throughout the threads but real life got in the way then the archive was banned. Next order of business is an index, I'm at page 6 so might take me a while.
 
Felt someone had to do it. I'm a fan of how this sounds like his ralph voice lmao.

Null - Rich Men North of Richmond


and my go-to favourite edgy song, though he come out a bit faggy
Null - 88 Rock and Roll


Null - WUHAN GANG


Null -Everybody Knows Shit's FUCKED
 
Last edited:
The Band Played Waltzing Matilda


Tomorrow


99 LuftBalloons (Joshcore Metal Mix)


People So Stupid


Bizarre Love Triangle (Acoustic - pushing Null's vocal range)


Fast Car (Acoustic)


Dirty Old Town
 
Last edited:
josh_moon_pointing_up.png

Here is Josh rapping in Finnish, performing the song "Jalat Ylös" ("Legs Up" in English) by Skandaali. Enjoy.
 
Bush - Letting The Cables Sleep


3. If anyone has tips for improving enunciation, please share. I’ve dicked around in the Collab settings and basically nothing gives obviously better results. So far the clarity of Nulls words is just a function of how clearly the original singer enunciates, and nasally singers make it worse.
I was looking up stuff about imitating elevenlabs and I saw something about a script that's able to sing to an audio sample and follow a prompt at the same time. I didn't look too deeply into it, but it might be worth checking into.

Even when the vocals are clean, if the source speaks too quickly or hits the right note, the bot will freak out. It also depends on the bot too--some will catch everything perfectly in a specific sample and others won't. Thankfully the music covers most of it and anything too egregious can be modified. When I was working on My Oh My, I was able to chop up Ralph's backing vocals to remove bad pronunciations or fill in the blanks and it didn't bleed through.
 
Last edited:
Felt someone had to do it. I'm a fan of how this sounds like his ralph voice lmao.

Null - Rich Men North of Richmond

View attachment 5291476

M8, this is so out of key it physically hurts to listen to. You can't just raise the voice a few steps, it puts the entire thing out of tune, you need to either do full octaves (+/- 12) or, if you're going to do quarter of half octaves, raise the instrumental by that as well.

Here's one where I pitched Josh down a full octave, -12. I'm not saying it's perfect (strong accents lead to poor AI replication) but it doesn't make me want to recreate it on the spot for being so terrible.

Rich Men North of Richmond - Josh (SiV Version)

 
1. If you shift Josh’s voice in either direction before processing it, please for the love of God shift the instrumental track the same number of semitones when combining the two in Audacity. This is dead simple to do.
multiple times in the thread people have said 'this one sounds way better' and the instrumental and josh are singing in different fucking keys.
 
Has anyone made Colt 45 by Afroman yet? Because I might just finally try making one of these.

Someone with talent, please beat me to the punch. I just want to hear it.
 
  • Feels
Reactions: TVB
Joshua & Travis Scott - 90210

I had to use some parts of the original vocals where the AI messed up (and use them as background vocals). But damn Josh really chopped his balls off on the high notes
 
M8, this is so out of key it physically hurts to listen to. You can't just raise the voice a few steps, it puts the entire thing out of tune, you need to either do full octaves (+/- 12) or, if you're going to do quarter of half octaves, raise the instrumental by that as well.

Here's one where I pitched Josh down a full octave, -12. I'm not saying it's perfect (strong accents lead to poor AI replication) but it doesn't make me want to recreate it on the spot for being so terrible.

Rich Men North of Richmond - Josh (SiV Version)

View attachment 5294544
im a functional retard with zero sense of pitch, tone or music so ¯\_(ツ)_/¯ I just made it sound like jersh and mixed it.
 
multiple times in the thread people have said 'this one sounds way better' and the instrumental and josh are singing in different fucking keys.
90 percent of people are fucking tone deaf. If you don't play an instrument then chances are low that you pay attention to these kinds of things.
 
  • Agree
  • Informative
Reactions: Puff and MrB
Back