SpectraLayers 12

[Robin Lobel]

Unmix Song is only half-GPU accelerated now - full acceleration will likely come with the upcoming patches.

[koalaboy]

I guess in fairness, there is a 20% discount on updates until the 16th July (explains the differences being seen).

I think my hesitance to upgrade is because it seems mostly aimed at the 'unmix', which I can appreciate is really popular right now, rather than improving restoration and separation around reverb/ambience which I would use more (being able to remove/replace/replicate reverb and environment is much more my path).

I will demo it I guess, and see what it brings.

[muzicxs]

Unmix Song is only half-GPU accelerated now - full acceleration will likely come with the upcoming patches.

This is my main gripe with Steinberg. Many times they release half-baked new versions just to get their software out on set target dates. If you're lucky they fix it within the upcoming year and otherwise you know you'll have to pay for their next release.

[koalaboy]

This is my main gripe with Steinberg. Many times they release half-baked new versions just to get their software out on set target dates. If you're lucky they fix it within the upcoming year and otherwise you know you'll have to pay for their next release.

As someone who works in the Software Industry (Games and previously 'big tech', not music), this is common in most 'big' businesses these days, as the releases are aligned to show growth in EBITDA and features either have to go out 'as they are' or will rarely be approved for any long term.

As an engineer, I would rather take the time, but consumers are constantly craving 'new updates' otherwise they move on and whilst you can have a solid core audience who don't mind, that doesn't 'grow' the business which is what Investors all want.

It's unfortunate, but it's the state of the tech industry these days. Everyone wants everything faster and better... but that's not how tech works. Of course, AI is going to solve all of that

I'm just happy, despite it not really being an update for me, that they continue to improve the product and it is still possible to buy a perpetual license and hasn't switched to subscription, which is the death of a product company.

I would agree that a roadmap would be nice, but that then doesn't allow the 'pivots' that many exec teams like to make. I have no insight into Steinberg (I would love to, but they don't do remote work AFAIK), but I've seen most companies follow similar patterns.

[Caine123]

strange over at the Steinberg forums someone named jarredou or something claims Steinberg stole his code and/didnt credit him or something. his posts were deleted it seems...
would be interesting what kind of modules/codes Spectralayers uses.

[DrGonzo]

User definable unmix - that sounds like an absolute killer feature.

[Shabdahbriah]

^^^ it IS.

[Calagan]

strange over at the Steinberg forums someone named jarredou or something claims Steinberg stole his code and/didnt credit him or something. his posts were deleted it seems...
would be interesting what kind of modules/codes Spectralayers uses.

Jarredou seems to be french (like me) and posted the same accusation in the dedicated Spectralayers thread on Audiofanzine.
https://fr.audiofanzine.com/editeur-aud ... e,p.3.html
I actually can't verify his accusation (I didn't update to SL12) but he says that if you use his own code (published in open source on github one year ago) and the Spectralayers one on the same source, the results are absolutely identical. And he even says that his code is not that great because it was not trained sufficiently, so it means Steinberg didn't even use any ressources to train the algo...
If someone updated to SL12 and can use these strange github stuff, it should be easy to verify...
https://github.com/jarredou/models/rele ... model_v0.1

If it's true, it's a bit crazy...

[Caine123]

strange over at the Steinberg forums someone named jarredou or something claims Steinberg stole his code and/didnt credit him or something. his posts were deleted it seems...
would be interesting what kind of modules/codes Spectralayers uses.

Jarredou seems to be french (like me) and posted the same accusation in the dedicated Spectralayers thread on Audiofanzine.
https://fr.audiofanzine.com/editeur-aud ... e,p.3.html
I actually can't verify his accusation (I didn't update to SL12) but he says that if you use his own code (published in open source on github one year ago) and the Spectralayers one on the same source, the results are absolutely identical. And he even says that his code is not that great because it was not trained sufficiently, so it means Steinberg didn't even use any ressources to train the algo...
If someone updated to SL12 and can use these strange github stuff, it should be easy to verify...
https://github.com/jarredou/models/rele ... model_v0.1

If it's true, it's a bit crazy...

thanks for the infos! yes thats him!
also obvious how his posts and account was deleted over at Steinberg. I did some screenshots.
you never know of course but just deleting posts smells alot of time not cool.

would be interesting if he could do a youtube video of his programm and showing it off, this should be provable if true by him.

[dionenoid]

strange over at the Steinberg forums someone named jarredou or something claims Steinberg stole his code and/didnt credit him or something. his posts were deleted it seems...
would be interesting what kind of modules/codes Spectralayers uses.

Jarredou seems to be french (like me) and posted the same accusation in the dedicated Spectralayers thread on Audiofanzine.
https://fr.audiofanzine.com/editeur-aud ... e,p.3.html
I actually can't verify his accusation (I didn't update to SL12) but he says that if you use his own code (published in open source on github one year ago) and the Spectralayers one on the same source, the results are absolutely identical. And he even says that his code is not that great because it was not trained sufficiently, so it means Steinberg didn't even use any ressources to train the algo...
If someone updated to SL12 and can use these strange github stuff, it should be easy to verify...
https://github.com/jarredou/models/rele ... model_v0.1

If it's true, it's a bit crazy...

The Spectralayers dev is not Steinberg but Robin Lobel. And he's posting here, as well as in that topic. And as you can see he's telling that Jarredou character that he sent him a private message (my French is bad but that's what i understood).

But perhaps Robin can tell us a bit more about it. Altho in all honesty he owns nobody here an explanation.

[Shabdahbriah]

Really excited about the unmix improvements.

Workflow Improvements New Features in SpectraLayers 12.png

https://www.soundonsound.com/news/stein ... alayers-12

[Calagan]

strange over at the Steinberg forums someone named jarredou or something claims Steinberg stole his code and/didnt credit him or something. his posts were deleted it seems...
would be interesting what kind of modules/codes Spectralayers uses.

Jarredou seems to be french (like me) and posted the same accusation in the dedicated Spectralayers thread on Audiofanzine.
https://fr.audiofanzine.com/editeur-aud ... e,p.3.html
I actually can't verify his accusation (I didn't update to SL12) but he says that if you use his own code (published in open source on github one year ago) and the Spectralayers one on the same source, the results are absolutely identical. And he even says that his code is not that great because it was not trained sufficiently, so it means Steinberg didn't even use any ressources to train the algo...
If someone updated to SL12 and can use these strange github stuff, it should be easy to verify...
https://github.com/jarredou/models/rele ... model_v0.1

If it's true, it's a bit crazy...

The Spectralayers dev is not Steinberg but Robin Lobel. And he's posting here, as well as in that topic. And as you can see he's telling that Jarredou character that he sent him a private message (my French is bad but that's what i understood).

But perhaps Robin can tell us a bit more about it. Altho in all honesty he owns nobody here an explanation.

Robin didn't explain anything. It seems he doesn't have nothing concrete to say regarding the accusations of Jarredou. And these accusations seem verifiable : just unmix the same source with SL12 and with the open source stuff Jarredou did upload one year ago, it should null perfectly...

I think Steinberg (or Robin, whatever) owes any customer an explanation : why do they charge 70 bucks for a new version with bugs and for which the main new features are just copy-paste stuff from a free open source code ?
It doesn't look like fair business, does it ?

[Caine123]

strange over at the Steinberg forums someone named jarredou or something claims Steinberg stole his code and/didnt credit him or something. his posts were deleted it seems...
would be interesting what kind of modules/codes Spectralayers uses.

Jarredou seems to be french (like me) and posted the same accusation in the dedicated Spectralayers thread on Audiofanzine.
https://fr.audiofanzine.com/editeur-aud ... e,p.3.html
I actually can't verify his accusation (I didn't update to SL12) but he says that if you use his own code (published in open source on github one year ago) and the Spectralayers one on the same source, the results are absolutely identical. And he even says that his code is not that great because it was not trained sufficiently, so it means Steinberg didn't even use any ressources to train the algo...
If someone updated to SL12 and can use these strange github stuff, it should be easy to verify...
https://github.com/jarredou/models/rele ... model_v0.1

If it's true, it's a bit crazy...

The Spectralayers dev is not Steinberg but Robin Lobel. And he's posting here, as well as in that topic. And as you can see he's telling that Jarredou character that he sent him a private message (my French is bad but that's what i understood).

But perhaps Robin can tell us a bit more about it. Altho in all honesty he owns nobody here an explanation.

Robin didn't explain anything. It seems he doesn't have nothing concrete to say regarding the accusations of Jarredou. And these accusations seem verifiable : just unmix the same source with SL12 and with the open source stuff Jarredou did upload one year ago, it should null perfectly...

I think Steinberg (or Robin, whatever) owes any customer an explanation : why do they charge 70 bucks for a new version with bugs and for which the main new features are just copy-paste stuff from a free open source code ?
It doesn't look like fair business, does it ?

that's what i found a bit fishy? why do his posts get deleted over at Steinberg Forums, here on KVR Steinberg cannot delete his comments so easily. so i hope this Jarredou can maybe also verify it or make some youtube video or something. who knows if it is true but i found the deletion of his posts asap kinda strange...

[Shabdahbriah]

User definable unmix - that sounds like an absolute killer feature.

[qtheerearranger]

I bet he’s deleting posts bc he’s hired a lawyer. It’s probably no more difficult than that. Calling out a big company won’t do anything. Filing a lawsuit will.