Five Former Members Of In Flames Unite In The Halo Effect
- Thread starter CladInShadows971
- Start date
Lmao, I just fr converted Mikael's Conditional vocals based on Anders' Clayman vocals using AI 
it's not perfect, but I'm shocked it actually worked at all. Will try to merge vocals and upload a sample.
it's not perfect, but I'm shocked it actually worked at all. Will try to merge vocals and upload a sample.An AI model based on Anders' Clayman vocals, on the first verse on Conditional...
It's not perfect and I didn't even train it fully (or do any proper mixing), but holy moly... that's not bad for a first attempt.
It's not perfect and I didn't even train it fully (or do any proper mixing), but holy moly... that's not bad for a first attempt.
DOTL chorus with the same Anders model:
With a better trained model and better quality vocal audio you could genuinely put Anders' old school vocals onto the album if you so wished. What a crazy time to be alive.
With a better trained model and better quality vocal audio you could genuinely put Anders' old school vocals onto the album if you so wished. What a crazy time to be alive.
I think Whoracle would be quite possible, using the same training technique as I used for the Clayman vocals.
Foregone I'm not sure about. Basically from what I've read you need at least 10 mins of voice data for the model to work properly. To make the model I have to rip the vocals stems from the songs, then remove all of the clean vocals to just leave the growls. Basically any IF album up until Come Clarity should have enough growls to get to at least 10 min, but after that it could get a bit dicey. SC would likely not have enough, for example.
As far as Foregone is concerned, it'll be a case of whether there's enough growls or not. Obviously there are songs like Pure Light of Mind which basically have none. I can't remember a lot of the others, so will have to see how much material we have to work with.
Obviously there are a lot of really cool possibilities - putting Anders' vocals on other MDM tracks from Soilwork, DT, Arch Enemy, etc (or other MDM vocalists on IF tracks), or putting vocals from one era of In Flames into another. Whoracle vocals to Clayman or vice-versa, swapping TJR/Come Clarity vocals... a lot of interesting choices.
However with that said the actual training process is pretty intensive. I checked and my CPU was running at 96C and the GPU at 85C for the 30 min during of the training I did on the Anders Clayman model yesterday.... goes without saying I don't want to put either under that kind of stress very often. So I may transfer the program to my desktop which has a weaker GPU but will have better airflow than the laptop. Training will take longer but hopefully not straining the components quite as much as the laptop.
Anyway, it's very cool to have a version of Conditional with Clayman-era Anders on vocals, even if the sound isn't perfect. Probably the most noticeable thing is cadence and inflection - the AI itself is using Mikael's voice to interpret these, which leads to subtle differences in how Anders historically pronounces certain words and phrases compared to Mikael. Overall though I would say it's done a good job considering this was my first attempt on a somewhat small set of sample data. I could probably combine the Colony/Clayman vocals to get a larger dataset and that possibly would produce better results.
Foregone I'm not sure about. Basically from what I've read you need at least 10 mins of voice data for the model to work properly. To make the model I have to rip the vocals stems from the songs, then remove all of the clean vocals to just leave the growls. Basically any IF album up until Come Clarity should have enough growls to get to at least 10 min, but after that it could get a bit dicey. SC would likely not have enough, for example.
As far as Foregone is concerned, it'll be a case of whether there's enough growls or not. Obviously there are songs like Pure Light of Mind which basically have none. I can't remember a lot of the others, so will have to see how much material we have to work with.
Obviously there are a lot of really cool possibilities - putting Anders' vocals on other MDM tracks from Soilwork, DT, Arch Enemy, etc (or other MDM vocalists on IF tracks), or putting vocals from one era of In Flames into another. Whoracle vocals to Clayman or vice-versa, swapping TJR/Come Clarity vocals... a lot of interesting choices.
However with that said the actual training process is pretty intensive. I checked and my CPU was running at 96C and the GPU at 85C for the 30 min during of the training I did on the Anders Clayman model yesterday.... goes without saying I don't want to put either under that kind of stress very often. So I may transfer the program to my desktop which has a weaker GPU but will have better airflow than the laptop. Training will take longer but hopefully not straining the components quite as much as the laptop.
Anyway, it's very cool to have a version of Conditional with Clayman-era Anders on vocals, even if the sound isn't perfect. Probably the most noticeable thing is cadence and inflection - the AI itself is using Mikael's voice to interpret these, which leads to subtle differences in how Anders historically pronounces certain words and phrases compared to Mikael. Overall though I would say it's done a good job considering this was my first attempt on a somewhat small set of sample data. I could probably combine the Colony/Clayman vocals to get a larger dataset and that possibly would produce better results.
ThePhobiac
be at ease
Sounds great, looking forward to Days of the Lost: The Friden Sessions
I'm a big fan of Mikael and everything, but I can't deny that I would have enjoyed Days of the Lost even more if mid-era Anders was on vocals. There's just a certain untrained energy that he added to Colony/Clayman (and Reroute, imo) that made the whole experience complete and more fun than it would have been otherwise.
I'm a big fan of Mikael and everything, but I can't deny that I would have enjoyed Days of the Lost even more if mid-era Anders was on vocals. There's just a certain untrained energy that he added to Colony/Clayman (and Reroute, imo) that made the whole experience complete and more fun than it would have been otherwise.
also now have Anders AI A Truth Worth Lying For, keeping Stanne on cleans, which technically marks their first duet since 1993 
Moonshield_
Member
@DE4life Man this is insane, in a way you managed to artificially create the band we’ve all dreamt about. It sounds like what a good evolution of IF could have been if the guys stayed together and I must say this short Anders excerpt of Conditional (which is my favourite from THE) really moved me.
The question is : do you think you’ll be able to provide us with a full Anders Conditional & Path of Fierce Resistance version?
Anyway, looking forward witnessing your next experimentations
The question is : do you think you’ll be able to provide us with a full Anders Conditional & Path of Fierce Resistance version?
Anyway, looking forward witnessing your next experimentations
@Moonshield_ check your inbox 
as far as Path of Fierce Resistance is concerned, haven't done it yet but I'll get to it. So far I've tried A Truth Worth Lying For, Days of the Lost, Feel What I Believe, In Broken Trust, Last of Our Kind and Shadowminds. Conditional definitely the best suited, FWIB sadly the least. Gateways vocals also sounded pretty bad for this model, unfortunately, so I haven't even attempted that one and probably won't.
Edit: Actually, In Broken Trust is probably the least suitable, I actually had to use some of Mikael's vocals as backing on the verses because the model struggled with the general vocal stem.
as far as Path of Fierce Resistance is concerned, haven't done it yet but I'll get to it. So far I've tried A Truth Worth Lying For, Days of the Lost, Feel What I Believe, In Broken Trust, Last of Our Kind and Shadowminds. Conditional definitely the best suited, FWIB sadly the least. Gateways vocals also sounded pretty bad for this model, unfortunately, so I haven't even attempted that one and probably won't.Edit: Actually, In Broken Trust is probably the least suitable, I actually had to use some of Mikael's vocals as backing on the verses because the model struggled with the general vocal stem.
Last edited:
Some of the better sounding parts from Shadowminds, Last of Our Kind and Feel What I Believe:
Wow, Last of Our Kind and Feel What I Believe are much improved, more than I would have expected.
For a long time I thought Anders was a/the problem, but this goes to show that he does elevate the music quite a bit. So now maybe we blame Bjorn?
For a long time I thought Anders was a/the problem, but this goes to show that he does elevate the music quite a bit. So now maybe we blame Bjorn?
Well, to be fair this is 1999/2000 Anders, and he hasn't put in a performance of that style since Come Clarity, so still fair to target him when it comes to the likes of ASOP or Battles. The model at times sounds more like a mashup of Anders and Tomas Lindberg, but I suppose they had similar styles back in the day so that makes some sense.
Although in fairness to Stanne, if I had a model based on his vocals from The Gallery/Mind's I period it would probably have more energy too, as he was a lot more full throated back then compared to now. Time is the real culprit sadly
Although in fairness to Stanne, if I had a model based on his vocals from The Gallery/Mind's I period it would probably have more energy too, as he was a lot more full throated back then compared to now. Time is the real culprit sadly
Well. If you like what Anders has been doing since two decades ago, and you still don't like IF, then Bjorn is the problem.
For people like me, when listening to these "remixed" songs, is think that it's more about nostalgia. Of course, we're fans of "old" IF and we're listening to what could have been. But we have to remember that Anders voice is not the same and that he's not willing to do that kind of music anymore. He's more interested into fakery and bland melodies.
Then, I'm pretty sure that fans of DT will say the opposite. That Stanne's versions are far superior.
For people like me, when listening to these "remixed" songs, is think that it's more about nostalgia. Of course, we're fans of "old" IF and we're listening to what could have been. But we have to remember that Anders voice is not the same and that he's not willing to do that kind of music anymore. He's more interested into fakery and bland melodies.
Then, I'm pretty sure that fans of DT will say the opposite. That Stanne's versions are far superior.
Last edited:
The Grayfox
#MusicTheoryBro
Wow, I never thought I'd actually want to hear this album with Anders on vocals -- and I wouldn't if it was today's Anders -- but this is actually fucking insane. I'd love to hear the whole thing. What are you using to do this?
Well, the entire process basically goes something like
1. Identify where to get the 10 mins + worth of vocal material from to train the model (in this case it was Clayman)
2. Use Ultimate Vocal Remover to create the vocal stems from all of the Clayman songs
3. Use Adobe Audition 3 to remove any clean/mumbling vocals and just leave the screams/growls in the vocal stems
4. Use RVC-beta (AI voice cloning tool) to train a model based on the vocal stems. This took about 30 mins with an I7-8750H processor and an RTX 2070 graphics card. You can train the model up to 200 epochs, but I limited it to 40 as it was killing both my CPU and GPU on temperature.
5. Use Ultimate Vocal Remover to create the vocal stems from the DotL tracks, as well as the instrumental stem.
6. Use RVC-beta to clone Anders' voice to the selected DotL track, this only takes a minute or two. The only setting I really changed from default was to take into account Mikael's vocals are of a lower register compared to Anders' Clayman vocals, and there's a setting for that.
7. Import the selected DotL instrumental and vocal into Adobe Audition 3, alongside the Anders' cloned vocal track.
8. Edit the Anders cloned vocal track in Adobe Audition 3 to account for any inconsistencies in volume, and mute any sound artefacts from the non-vocal sections of the vocal track. There was also the occasional need to rearrange parts of the vocal track where the cloning wasn't so good, or using Mikael's vocal track as backing vocal where there was a particularly weak section of the cloned track.
9. Save the mix to an mp3.
And that's pretty much the whole process. Steps 1-5 obviously only need to be done once, so the process becomes a lot quicker once you have the model trained and all of the instrumental/vocal stems created. Sometimes step 8 can take a while depending on how well the cloning process worked. The majority of the cloned vocals came out 90% fine, so only some minor adjustmenets needed, but there were a couple of tracks (Gateways and IBT) where it just doesn't work very well. The tracks that worked best are the ones with standard growling across the board. Overall Anders and Mikael seem like a pretty good match for cloning vocals (I assume it'd be quite good in reverse, too). I also left Mikael's cleans in a couple of songs as the model doesn'twork when putting growls over cleans, but in theory you could create a model for Anders' clean vocals to clone over Mikael's cleans... but really, why do that?
Overall results might be even better if the model was trained to higher epochs, but 30 mins at nonstop 96c for my CPU was as far as I was willing to go.Also possible that you don't really see any improvement past 40 epochs. To improve the model I'd be more inclined to take the vocal stems from Colony and use those to supplement the training model alongside the Clayman vocals. Maybe I'll do that at some point.
If you have a beefy enough CPU/GPU and want to give it a go, I used the below YouTube video to install and navigate RVC-beta:
You can find UVR via google if you don't already have it, it's open source & free. RVC-beta actually does have a vocal stem separation tool you can use, although I'm not sure how good it is. I used Adobe Audition 3 for editing but you can use any music editing software for that job.
It's not really a very complicated process at all overall, the first five steps are just a bit time-consuming. Once you've done those, though, then the rest is fairly straightforward depending on how well the vocals cloned.
1. Identify where to get the 10 mins + worth of vocal material from to train the model (in this case it was Clayman)
2. Use Ultimate Vocal Remover to create the vocal stems from all of the Clayman songs
3. Use Adobe Audition 3 to remove any clean/mumbling vocals and just leave the screams/growls in the vocal stems
4. Use RVC-beta (AI voice cloning tool) to train a model based on the vocal stems. This took about 30 mins with an I7-8750H processor and an RTX 2070 graphics card. You can train the model up to 200 epochs, but I limited it to 40 as it was killing both my CPU and GPU on temperature.
5. Use Ultimate Vocal Remover to create the vocal stems from the DotL tracks, as well as the instrumental stem.
6. Use RVC-beta to clone Anders' voice to the selected DotL track, this only takes a minute or two. The only setting I really changed from default was to take into account Mikael's vocals are of a lower register compared to Anders' Clayman vocals, and there's a setting for that.
7. Import the selected DotL instrumental and vocal into Adobe Audition 3, alongside the Anders' cloned vocal track.
8. Edit the Anders cloned vocal track in Adobe Audition 3 to account for any inconsistencies in volume, and mute any sound artefacts from the non-vocal sections of the vocal track. There was also the occasional need to rearrange parts of the vocal track where the cloning wasn't so good, or using Mikael's vocal track as backing vocal where there was a particularly weak section of the cloned track.
9. Save the mix to an mp3.
And that's pretty much the whole process. Steps 1-5 obviously only need to be done once, so the process becomes a lot quicker once you have the model trained and all of the instrumental/vocal stems created. Sometimes step 8 can take a while depending on how well the cloning process worked. The majority of the cloned vocals came out 90% fine, so only some minor adjustmenets needed, but there were a couple of tracks (Gateways and IBT) where it just doesn't work very well. The tracks that worked best are the ones with standard growling across the board. Overall Anders and Mikael seem like a pretty good match for cloning vocals (I assume it'd be quite good in reverse, too). I also left Mikael's cleans in a couple of songs as the model doesn'twork when putting growls over cleans, but in theory you could create a model for Anders' clean vocals to clone over Mikael's cleans... but really, why do that?
Overall results might be even better if the model was trained to higher epochs, but 30 mins at nonstop 96c for my CPU was as far as I was willing to go.Also possible that you don't really see any improvement past 40 epochs. To improve the model I'd be more inclined to take the vocal stems from Colony and use those to supplement the training model alongside the Clayman vocals. Maybe I'll do that at some point.
If you have a beefy enough CPU/GPU and want to give it a go, I used the below YouTube video to install and navigate RVC-beta:
You can find UVR via google if you don't already have it, it's open source & free. RVC-beta actually does have a vocal stem separation tool you can use, although I'm not sure how good it is. I used Adobe Audition 3 for editing but you can use any music editing software for that job.
It's not really a very complicated process at all overall, the first five steps are just a bit time-consuming. Once you've done those, though, then the rest is fairly straightforward depending on how well the vocals cloned.
Last edited:
As metalheads I don't think any of us need to be too concerned about pop music being created artificially... studios have been doing that for years already without AI At the end of the day AI (at the moment at least) can't play live, and seeing a band in concert is an important part of most people's metal experience. People who can play stuff live will want to create their own stuff, it's human instinct to want to put your own work out there, not just play something generated by AI.
That's not to say it won't happen, or that AI won't be used to help artists when they're having a bit of writers block, but I don't think - for metal at least - it'll become an issue any time soon. Stuff like Pop, Country, even modern Hip Hop already has a ton of manufactured product out there, so AI or not it doesn't make much difference. Listen to Anders on Pure Light of Mind, which might as well be an AI rendition of an actual singer, and ask if it's not already happening?
At the end of the day AI (at the moment at least) can't play live, and seeing a band in concert is an important part of most people's metal experience. People who can play stuff live will want to create their own stuff, it's human instinct to want to put your own work out there, not just play something generated by AI.That's not to say it won't happen, or that AI won't be used to help artists when they're having a bit of writers block, but I don't think - for metal at least - it'll become an issue any time soon. Stuff like Pop, Country, even modern Hip Hop already has a ton of manufactured product out there, so AI or not it doesn't make much difference. Listen to Anders on Pure Light of Mind, which might as well be an AI rendition of an actual singer, and ask if it's not already happening?
If we have in mind that IF need artificial tools to make Anders sing at the studio and they rely a lot on backing tracks live, then I'm not sure that metal is not going to walk that path too. I said IF but there are a lot of bands doing the same. Metal is now, closer to pop than it was ever before.
Similar threads
