From: Neural DSP, re: s o o n... (Archetype Tom Morello)

[MirrorProfiles]

I’m still not seeing anything in what you’ve said that shows anything NDSP has done that is “redefining” or innovating. They didn’t do anything that wasn’t already being done by several other companies.

What other companies are doing machine learned amp sims, with both standalone software and plugins, and an interface thats as clear and concise? Am I missing a product that did this first?

I agree on the marketing being a success being a massive factor for them - it helps that not only were the products good (and deliver what they promise), they made ALL of their products either officially licenced, or a signature product of a famous artist/band. I'm not saying they're the first to do some licensed products, but for amp sims, I'm not sure of anyone else who was doing only licensed stuff, that could be marketed in this way. Mixwave seems to have followed suit with that approach although I don't think their marketing, brand tie ins, technology or overall user experience are really on the same level. The UI's look nice though.

They jumped in the market with a solid offering that sounded good and looked pretty and they did a good job with the marketing

This is precisely the point I'm making - if the product wasn't good, they wouldn't have much to market with, They got the UI, price, features, sound etc right. I'd love other companies to match them on this stuff - I think I own most stuff from Nembrini/IK/L6 Metallurgy/Audio Assault/STL/Mercuriall/Softube/Audiority and for the most part, their products still lag behind NDSP's plugins overall IMO. That's not to say that each of those don't have great (and sometimes unique) qualities, but I feel like the overall experience is not as good. I'll be as happy as anyone if they are able to overtake them.

 

[Lysander]

I think I own most stuff from Nembrini/IK/L6 Metallurgy/Audio Assault/STL/Mercuriall/Softube/Audiority and for the most part, their products still lag behind NDSP's plugins overall IMO.

Hard disagree. I've used quite a few plugin suites/brands and, while NDSP's stuff is excellent, there's plenty of good (if not better) competition out there. Nembrini is a brand which comes to mind, and i don't see mentioned often.

Personally i'm using Helix Native these days, which offers a ridiculous amount of features (and great sounds!) for the price of ~2 single NDSP amp plugin suites. Presets are crap though.

 

[AZG]

The main criticism I can give the NDSP plugins is that there is no "suite" option that would let you bundle all your plugins together to use as one. Sure, you can load them in a DAW or VST host but that's definitely not the same as "one thing to control them" type deal like you have in e.g Amplitube 5 or how something like Helix Native works as one software.

This is definitely a reason I won’t buy anymore of their plugins. I tend to use the complete packages like Helix Native and TH-U more. ToneX is the exception since it sounds great and I don’t have Amplitube.

 

[MirrorProfiles]

Hard disagree. I've used quite a few plugin suites/brands and, while NDSP's stuff is excellent, there's plenty of good (if not better) competition out there. Nembrini is a brand which comes to mind, and i don't see mentioned often.

Own lots of Nembrini here - IMO they are too hit and miss. I like them but nowhere near NDSP for me. No standalone versions (they have a host plugin but its not nice to use at all). Iffy poweramp modelling, IR's are rarely great. I feel like almost all of their amp models have a similar gain characteristic to them that isn't the most authentic. Their manual's are written like they're (barely) an afterthought, the included FX are even more similar than NDSP's and sound much worse. I don't use presets but theirs sound dreadful to me. The redeeming factor is they have constant sales, and for $20-29 they are reasonable value. For a "one man band" type company, they support their products well.

I like them but I wouldn't really say they're leading the way for anything as far as amp sims, in my opinion. Definitely curious where you think Nembrini are beating NDSP, and what may make them better for you to use.

I'm a huge fan of Helix Native (one of my favourite plugins ever), but its a totally different style of product and approach to NDSP. For a start, there is no standalone, nothing prerouted, no graphics. Its amazing software and stomps all over the other products in that category (Amplitube, Amp Room, S-Gear, Revalver).

Metallurgy is a much fairer comparison. I really like Metallurgy, and they're probably what I'd rank closest to NDSP as far as the overall user experience for guitar amp plugins in 2023. some occasional bugs, UI is more cluttered and not as nice to use overall IMO (for instance, resizeable UI's would be nice). They scratch an itch that Helix Native doesn't.

STL make some excellent sims and for people who want an entire ecosystem in one plugin (as well as some good Archetype style plugins with JM and LL), I think they do it quite well. But again, I wouldn't say they are on NDSP's level.

 

[metropolis_4]

What other companies are doing machine learned amp sims, with both standalone software and plugins, and an interface thats as clear and concise? Am I missing a product that did this first?

Several offer both standalone and plugin

“An interface that is clear and concise” is completely subjective personal opinion. I could easily say “yes”, you could say “no” and neither of us would be wrong.

So that just leaves ML. Maybe they were the first, and maybe they are the only? I really have no idea. Personally I don’t care if they use the latest ML and AI or if it’s Cobalt.

But assuming they are the first and only one to use ML under the hood. Has that really “redefined” the whole amp sim plug-in industry? Was that so ground breaking that all other companies are rushing to get rid of their current modeling processes and follow suite?

 

[MirrorProfiles]

Several offer both standalone and plugin

Right - and which ones are as nice to use? I like Metallurgy and STL, but besides the fact that there is a lot of similarity to how NDSP lay out their software, they're both less clear and concise to use (as I mentioned in the post above). Mercuriall Ampbox is sort of half way there but again, less clear UI's, unlicensed models, worse presets etc. You're perfectly entitled to think that these UI's are more slick and modern, but I disagree. Not sure of any standalone amp sim's that were around before NDSP AND are nicer to use. What are the options of standalone amp sims pre NDSP? Amplitube? TH-U? Revalver? Audio Assault?

The approach to modelling was more to do with calling them innovative - the user may not care, but it certainly brought something new to the table that hadn't been done before. Some people don't like how it sounds (ML is more prone to aliasing and has other drawbacks). To offer my own opinons/experiences, I've found them to generally always match my real amps very closely, much moreso than I've been able to achieve with most other brands (which I also like, but I think NDSP pushed things forward in terms of accuracy). It may not be something users care about, but IMO NDSP plugins and Fractal (no plugin options) are consistently the closest to my real amps, with Helix a c*nt hair behind. STL and Mercuriall a fraction behind them. Just like with what people think with the UI, others can disagree but thats my opinion, and I've spent a good amount of time testing.

But assuming they are the first and only one to use ML under the hood. Has that really “redefined” the whole amp sim plug-in industry? Was that so ground breaking that all other companies are rushing to follow suite?

People keep grabbing one thing I say and pin it to the overall success of NDSP. The success is down to the combination of getting various important things right. The plugins sound good, they have the associations with famous artists/influencers, they look good, are easy and intuitive to use, they work on many platforms, offer the features people want, good presets and are priced where people are willing to open their wallets.

Having good sounding ML plugins on its own isnt going to be enough to create the hysteria NDSP gets. Neither are plugins that simply look good, or have good features. They landed on something that meets what customers want, and that wasn't really being offered in the same way by anyone else - hence why they had such a rapid expansion and volume of success. If they weren't bringing anything to the table besides clever marketing, why isn't it another brand in their position? I like lots of other amp sim plugins, but they often have their own drawbacks, or clear reasons why they aren't grabbing all the attention. IK are sat on so much good tech, but how its delivered to users is absolute dogshit.

Of course none of that applies to the QC side of things, but we're discussing the software here.

 

[metropolis_4]

Right - and which ones are as nice to use? I like Metallurgy and STL, but besides the fact that there is a lot of similarity to how NDSP lay out their software, they're both less clear and concise to use (as I mentioned in the post above). Mercuriall Ampbox is sort of half way there but again, less clear UI's, unlicensed models, worse presets etc. You're perfectly entitled to think that these UI's are more slick and modern, but I disagree. Not sure of any standalone amp sim's that were around before NDSP AND are nicer to use. What are the options of standalone amp sims pre NDSP? Amplitube? TH-U? Revalver? Audio Assault?

The approach to modelling was more to do with calling them innovative - the user may not care, but it certainly brought something new to the table that hadn't been done before. Some people don't like how it sounds (ML is more prone to aliasing and has other drawbacks). To offer my own opinons/experiences, I've found them to generally always match my real amps very closely, much moreso than I've been able to achieve with most other brands (which I also like, but I think NDSP pushed things forward in terms of accuracy). It may not be something users care about, but IMO NDSP plugins and Fractal (no plugin options) are consistently the closest to my real amps, with Helix a c**t hair behind. STL and Mercuriall a fraction behind them. Just like with what people think with the UI, others can disagree but thats my opinion, and I've spent a good amount of time testing.



People keep grabbing one thing I say and pin it to the overall success of NDSP. The success is down to the combination of getting various important things right. The plugins sound good, they have the associations with famous artists/influencers, they look good, are easy and intuitive to use, they work on many platforms, offer the features people want, good presets and are priced where people are willing to open their wallets.

Having good sounding ML plugins on its own isnt going to be enough to create the hysteria NDSP gets. Neither are plugins that simply look good, or have good features. They landed on something that meets what customers want, and that wasn't really being offered in the same way by anyone else - hence why they had such a rapid expansion and volume of success. If they weren't bringing anything to the table besides clever marketing, why isn't it another brand in their position? I like lots of other amp sim plugins, but they often have their own drawbacks, or clear reasons why they aren't grabbing all the attention. IK are sat on so much good tech, but how its delivered to users is absolute dogshit.

Of course none of that applies to the QC side of things, but we're discussing the software here.

“I personally prefer them” =/= “redefined the industry”

 

[northern_fox]

Neither NAM, nor ToneX, nor the QC, can "model" an amp with all the controls.

NDSP also have a quite a few original amp "models" in their plug-ins, so I'm wondering what they would be able to train a neural net against if there isn't a real amp involved, or at least a component level model, in which case why would you need the ML model?

So, I'm sceptical of their claims, but I'm I do like their plug-ins, so I don't really care, and as such I'm prepared to give them the benefit of the doubt!

NAM is capable of this. It's just not a "user friendly" process. Steve did an example model way on back before any of this became bigger. Not suggesting that it's easy or straightforward, but the tech is real and it is possible to do. People who debate about it are largely concerned with the "number of permutations" required to cover an amp space, but that doesn't necessarily mean it's impossible to generate an accurate model with a sparser data set that doesn't take an infinite amount of time to collect and process.

Most people have no actual experience in building that kind of model, so the "infinite parameter impossibility" issue is largely hypothetical, even if there's truth behind it from a theoretical perspective. There's no reason NDSP couldn't have done the same.

 

[Alex Kenivel]

They did essentially redefine guitar amp sim plugins for the most part

I see it like they made them "more popular" through clever marketing and big names

 

[James Freeman]

there are very obvious clues that the QC captures are doing some kind of Kemper style matching.

think the open cortex project has worked out that QC’s capture tech is no different (in concept) to how Kemper works,

OpenCortex/docs/dev/Captures.md at main · VanIseghemThomas/OpenCortex

A project that opens your Quad Cortex for homebrew software - VanIseghemThomas/OpenCortex

github.com

^

After cracking the encryption to the Neural Captures, and writing some code to decode the captures to JSON format. I was shocked to see that there is very little "Neural" about it. Recently it was discovered that the training process involves some kind of genetic algorithm and until now we haven't found a conrete reference to any neural network training. Looking at one of the captures you can see that the network consists of a 13 parameter network.

 

[MirrorProfiles]

“I personally prefer them” =/= “redefined the industry”

Just going in circles here, because I believe I've made the points that back this up. Has another company had that kind of instant impact with amp sim plugins? I mean, I dont think its even an opinion that the plugins have been very successful and generally highly regarded (as well as making the right product at the right price and targetted to the right audience). Regardless of ANY opinion of them, they've sold shitloads and the industry has responded. Its definitely influenced plugins that have come since (and how they're marketed etc). Could well have happened anyway if someone else did it, but NDSP definitely played a part there. Like them or not, their plugins shot to the top tier of amp sim plugins, so I think its fair to say the industry has changed since they came on the scene.

I see it like they made them "more popular" through clever marketing and big names

Yep, totally agree. But if the products weren't good (and delivering what the market wants) then I don't think they would have got very far. Its an important factor, but far from the only one.

 

[Elf]

OpenCortex/docs/dev/Captures.md at main · VanIseghemThomas/OpenCortex

A project that opens your Quad Cortex for homebrew software - VanIseghemThomas/OpenCortex

github.com

^

WOW!!! I did not have Open Source NDSP QC in my bingo card! Did you guys?

*Edit Forked!

 

[Lysander]

WOW!!! I did not have Open Source NDSP QC in my bingo card! Did you guys?

Honestly, i wonder why it took this long.

The QC is a glorified embedded system running Linux... out of a SD card which can be accessed directly from the mainboard, making it (by far) the easiest modeler to hack out there.

 

[pipelineaudio]

The only "revolutionary impact" I saw from NDSP plugins was with Youtube marketing shills. I don't see so many real tutorials out there using them, they tend to go for the "suite" style stuff like Overloud, Helix native and Amplitube

Also, before there was NDSP, but after Kemper, Audio Assault was doing the multiple incompatible plugins thing "with ai modeling". I put that in quotes for legal reasons and I'm not a lawyer so don't take legal advice from me

 

[Orvillain]

I’m still not seeing anything in what you’ve said that shows anything NDSP has done that is “redefining” or innovating. They didn’t do anything that wasn’t already being done by several other companies.

Yeah, with the anti-NDSP shades off and trying to be objective, this is how I see it too. The signal routing of your typical NDSP plugin is straight from Amplitube 2, as is the GUI and the microphones on the cab (AT3 for that one) and I'm fairly sure Mercuriall had some tabbed plugins the same way NDSP do as well, and all in all... I just don't see them redefining anything. They're riding coat tails of well established design concepts.

 

[Orvillain]

Has another company had that kind of instant impact with amp sim plugins

Well yeah, IMO. For the time, Podfarm did that. As did the original Amplitube. Loaaaaadddsss of studios and engineers and consumers jumped on board with those. Hell, as much as I'm not into it, even Guitar Rig had a well established presence in the market in the mid 2000's.

The SansAmp plugin that came with Pro Tools, does that count???

 

[Gearzilla]

Looking like an Archetype.

 

[[Nathan]]

When I got back into amp sims a couple years ago I looked around at the amplitubes of the world and it was borderline the same dinosaur style setups from 10 years earlier. NDSP has an "Apple" like quality to their graphics and UI/UX. They also tend to have a great default sound so from the millisecond you open it up you have something usable and you can go from there. That seems like small fries but its a huge psychological difference to other amp sims.

I sit here with 20 amp sims, some real amps, the FM3 and a bunch of pedals... but somehow firing up a NDSP sim always feels like "quality". The sum of all its parts make them extremely marketable... Add to this their marketing strategies, who they choose to partner with to hit certain markets and it makes them a premium brand imo. To me its easy to see how they swept in and dominated the plugin world.

Some people dont care about all that stuff and just want an authentic amp sound and to move on, other people appreciate a visually nice easy to use plugin. I think they pretty much tick all the boxes except for keeping things upto date... and the price point. If they made things 40% cheaper and their dev was in full swing for their back catalogue I'd imagine they'd shift a bunch more units.... and while we're at it make a full suite loader like kontakt/tonehub so multiple plugins are easy to manage.

 

[James Freeman]

A clean car drives better.
A nice UI sounds better.

 

[the swede]

Marketing is the big key also. It’s one thing making a plugin that sounds good. All NDSP plugins sounds good. But it’s the wrap/package/delivery/presentation. The unique strategy started with Plini I think… maybe before even. But that’s when I saw their “big picture strategy”. Basically it’s about letting social media do the work with hype, reviews and spread of word. First few well placed reviews sells a shit ton, well placed arrows. Everything that follows is usually an enormous amount of shorts/reels tagged with the plugin flooding the social media flows. Sells.
Undeniably the easy plug and play focus of it also sells. 90% of the buyers don’t really bother to go beyond the presets I think.

Just my view on it. I can’t say anything bad about the plugins as I bought both Plini and Nolly a few years ago. The both sound great expect there’s overlap yet to little for any of them doing what I want on their own.