New album Foregone out February 2023

Eh, I think Colony has more weak tracks than TJR does personally. I’d take Wayfaerer over Pallar Anders Visa and Dead God In Me over Resin any day. And I’d certainly take anything on TJR over songs like Insipid 2000. Don’t get me wrong, I like Colony. It has some great songs — Zombie Inc., Embody, etc. but I will always pick TJR and Whoracle over it. Something about the production and mix of Colony bothers me as well.

Embody is probably my favourite opening to any album I've heard. Just a melodic punch to the face which doesn't let up at all. Anders' vocals on Colony are my favourite overall the IF discography. I just think he absolutely nails that balance between the deeper growl and higher scream. I can understand why he couldn't keep it up though. It's what I tried to emulate back in the day, and whilst I had achieved the sound I wanted by the end, it was murder on my throat. That was just with me doing a handful of vocals across a single song every so often, I can't imagine doing that repeatedly live as well. I couldn't have done it. You can hear the toll it's taken on him by the time Clayman comes out, and even on Reroute. It sounds amazing on Colony though, and in the live performances around the 1999 period. As he wasn't utilising any kind of safe technique, though, it was inevitable that he wouldn't be able to keep it going. I'm glad we got at least one album with him doing that style of vocal almost flawlessly though.
 
  • Like
Reactions: mindd and Midas
Yeah, I mean, the songs on Colony that I like, I really like. The first five songs are among my favorite IF songs ever. But in terms of petering out, I think Colony suffers from that more so than TJR. If it wasn’t for Behind Space, which shouldn’t even count because it’s a re-record, I wouldn’t listen to the last 1/3 of the album. I do agree that it’s Anders’ best vocal performance. I’ve always wished that he had kept the style of cleans he used on songs like Colony and Coerced Coexistence, but as you said, it wasn’t sustainable. Who knows, as AI gets better and better maybe we can see what albums like RTR and STYE would have sounded like with that vocal style :D
 
  • Like
Reactions: Midas
New track. Not done as it lacks bass. I like the calm parts, but I figure the intro could need some more spice. I don't think I like it. It is too repetitive.

Nero <- Click

Nero, take 2 <- Click

For take 2 I lowered the kick drum as they caused clipping, and were too "present". I also removed the clean guitars near the end of the track.

Nero has a few brutal parts, and I kind of like what I came up with. I still do think that it is too repetitive. - Again though, ideas and suggestions are very welcomed.

As for Rafe, I tried to do what you said, @ThePhobiac and distorted the chorus and I liked it. I'll post it later. Thank you guys for showing interest. You too @DE4life
 
Last edited:
Keep us posted for sure!

In terms of creativity, I've mainly been playing with Ultimate Vocal Remover to see if it's possible to extract the vocals from my old In Flames covers from 10+ years ago and mix them into the real IF tracks without vocals, as UVR can produce instrumental tracks pretty much flawlessly. Back in the day I just had to use YouTube covers, and at the time Shehf was the go to. Dude did some really great instrumental covers.

Most of the attempts I've made so far haven't come out great, as you'd expect. Minus and Sober and Irrelevent were alright, but I was recording on a cheap mic and using Cool Edit 2 for editing back then, lmao, so it was never going to be studio quality stuff. The covers also tended to be recorded at slightly different timings to the actual tracks, so my vocals often don't have an exact match with the real track. Sometimes that can be fixed with editing, but a lot of the time it just messes with the flow and sounds weird.

I tried Satellites and Astronauts though, and wow, Schehf must have absolutely nailed his timings because my vocal patterns were almost perfect, and I was surprised to hear that for the most part it sounded solid against the music. Not perfect, obviously, but better than pretty much anything else I'd tried. Considering S&A is one of my absolute favourite IF songs, that was awesome to hear.

I won't have a bad word said against AI, it's the king :cool:
 
  • Like
Reactions: Midas
@DE4life
So you're also into that? (Removing vocals and so on.) That's awesome. I used to mess around with vocal removers before but I've never heard of
Ultimate Voice Remover before. I have to check that one out.

I'll keep you guys posted with my shitty music if that's what you guys want. :3
I actually removed a few parts on Nero (see previous post.) and changed a part where I messed up. It now sounds a bit better.
I do still think it sounds repetitive. Maybe I have to come up with a Bridge. I'm however out of ideas as of what it should sound like, and I won't force anything.
Usually when you try to force a part it just ends up sounding like shit.
 
Last edited:
@DE4life
So you're also into that? (Removing vocals and so on.) That's awesome. I used to mess around with it before. It is very interesting.

I'll keep you guys posted. I actually removed a few parts on Nero (see previous post.) and changed a part where I messed up. It nbow sounds a bit better.
I do still think it sounds repetitive. Maybe I have to come up with a Bridge. I'm however out of ideas as of what it should sound like, and I won't force anything.
Usually when you try to force a part it just ends up sounding like shit.

Never heard of ultimate vocal remover, I have to check that one out.

Yeah, I've always found it interesting, but up until recently there's never really been a very good way to remove vocals and leave the instrumental side of the track still sounding 100%. With the rise of AI that's changed though, and UVR is a good tool when it comes to doing this kind of thing. It separates the instrumental and vocal stems so that you can have both separately, which obviously leads to a lot of interesting possibilities. When messing around, I:

- Removed vocal lines that I didn't like from certain songs, including "Joey was right" from TGD
- Reduced vocal volume on certain tracks where I felt vocals were too high in the mix
- Switched the studio vocals from Black & White with the awful live Hultsfred vocals, just for a laugh (although I actually prefer how Anders does the final part of the chorus at the end live)
- Put the vocals from the demo version of Episode 666 onto the studio version
- Put the vocals from Blinded By Fear onto State of Slow Decay (yes, as expected, they fit pretty well)
- Put the vocals from the original Behind Space onto Behind Space '99

I've also created instrumentals from a bunch of IF albums. There are some albums it does better than others. UVR created instrumental versions of TJR - Clayman almost flawlessly, minus a few issues on a handful of songs, but it struggles more with albums like Come Clarity, ASOP and SOAPF. Presumably because of the production and the change in vocal style on ASOP and SOAPF to the mixed yell/scream.

Probably the most complex thing I've done so far was combining UVR with a voice cloning tool called RVC-beta. I trained a voice model on Anders' vocals from Clayman, then cloned that model over the vocals from some of The Halo Effect songs. On some tracks it doesn't work, but on many of them it worked surprisingly well. Conditional in particular sounded very good.

So yeah, I've been having fun with it!
 
  • Like
Reactions: Midas
You have to share your edits with me, I am extremely interested in hearing what you've came up with.

I have to thank you for letting me know about this mindblowing tool. It is beyond anything I've heard.

Edit: I just tried out UVR-5 and consider me surprised. I went with Clayman as test track and HOLY FUCK does it sound great.
Here are my settings:

Process:
Ensemble Mode
-

Main Stem Pair:
Vocals/Instrumental
-
Ensemble Algorithms:
VR Arc: 7_HP2-UVR
MDX-NET: UVR-MDX-NET Main
MDX-NET: UVR-MDX-NET Inst Main
Demucs v4 | htdemucs_ft

GPU conversion: Checked
Instrumental Only: Checked

Render time was around 3 minutes.

I agree with you when it comes to AI. Especially when it comes to this. AI is King, for sure.
 
Last edited:
Yeah, I was pretty blown away by how well it achieves the separation. At first I thought it was fucking up the vocal stem as it sounded kinda bad, but when I put the instrumental and vocal stem back together the song sounded exactly the same, so that is actually how the vocals sound. Amazing how production choices hide stuff like that.

In terms of my edits, I think they've all been posted here at some point, but as a recap:

Episode 666 Studio Version chorus w/demo vocals:


Black and White w/Hultsfred vocals:


State of Slow Decay w/Blinded By Fear vocals (I only did the first 2:18 of the track):


Behind Space '99 w/Stanne's original vocals:


Last of Our Kind w/Anders vocals:


Conditional w/Anders vocals:

 
  • Like
Reactions: Midas
Hmm... Doesn't appear to be working. Only SOSD worked. I like what you did there. :3
 
Google drive can be a bit weird like that. If it gives you the option to download then you can just download to your device to listen locally, otherwise I'm not sure!
 
This (UVR) proves that the things that were "hidden" when we did the "Hidden messages in their songs"-thread, was full of shit.
I now think that schizophrenia plays a part to the voices we "heard" and actually do not exist at all.
 
Last edited:
That is for sure. The bare truth is that In Flames are far too lazy to have gone through the effort of putting hidden stuff into their songs.
 
Currently putting my GPU to work training a Colony vocal model, will then try to combine the Colony/Clayman models to see if that improves the sound or makes it worse.
 
  • Like
Reactions: Midas
So, the Anders Colony/Clayman model is definitely an improvement on the individual Colony and Clayman training models.

I'll do some experimental stuff with this on proper IF tracks and other shortly, but the first thing I attempted was a little more complex. I basically took my vocals from the Satellites and Astronauts cover I did, then cloned Anders' vocals over it from the Colony/Clayman model. The results are actually pretty interesting. The cadence and vocal timing is me, but the vocal sound is from the Colony/Clayman model I trained.



Excuse the production on the vocals - it's using the production based off my original recordings, which are obviously not studio quality, so it doesn't sound perfect by any means.

Firstly, it hasn't typically handled clean/spoken parts well with my previous trained models. As it's trained solely on growls/screams, it hasn't really got a reference for clean vocals. However, it does a fairly good job here of transferring the clean to growl affect. Better than it's done with the earlier models anyway.

Personally I think that first screamed verse sounds pretty cool. The only part it struggles with is "moon", possibly because it's a lower growl on my vocals, and from Anders Colony/Clayman vocals it doesn't really have any proper deep vocals to reference. I supplemented this with the original vocals and it seems to fill it in OK. You also get to hear Anders pronouncing astronauts correctly.

One of the things I really liked about the cover I did was that, for the second chorus leading into the solo, I elongated the 'ahead' scream. I really think it sounds awesome like that, and wish that this is how it sounded on the original track.

Anyway, I just thought this was a cool little experiment. I'll be experimenting with some other IF tracks shortly, and maybe some non-IF tracks based on the new Colony/Clayman vocal model.
 
I tried to train a vocal model using clean vocals and it turned out terribly. Trying to get it to emulate power metal style vocals was a complete disaster. It seems to work much better with growls/screams. I used Matt Barlow as the model and the vocals that it reproduced sounded like a cat being raped. My thought was changing the vocals on Iced Earth’s ‘The Glorious Burden’ to Barlow’s because I’ve wanted to hear that for a long time and it was just brutal. Maybe if you really know what you’re doing, it could be done.
 
  • Like
Reactions: Midas
I'm going to try putting Anders' clean vocals from ITM/Battles on Oops I Did It Again. Will report back on success or failure.
 
@DE4life Do you think it is possible to convert ones own voice into someone else?
I don't quite understand what you mean by "training a Colony vocal model". Is that kind of what I am asking for?

Dang, shame it failed. Would've been hilarious.
 
If I understand what you're asking correctly, then the Satellites audio I posted is doing exactly that. They are my vocals converted to a different voice based on a voice model trained on Anders' screams/growls from both Colony and Clayman. How much you think it actually sounds like Anders is open to debate, but it doesn't sound like my original vocals either, so it has effectively transformed my voice into someone else's.