Note; this post isn't intended to evaluate the digital audio emulation as an "evolution" of the analog gears, the delicious hardware that almost every human would like to have & play with while billons can't afford, like me. This is not a religious discussion, but a chance to share impressions, thoughts & experiences about the chances of digital modeling based on capturing hardware in order to achieve more or less an "enjoyable quality" for play & production without thousands of dollars or any other coin.

For thousands of guitar players around the Earth (includin' me), & producers, Neural Amp Modeler means a very accuracy way to achieve impressive amp emulations, I feel it as never before. Not every captures are great, obviously it relies on patient dedication, passion & knowledge about the topic, but you can find a high number of jewells in ToneHunt for free as those made by Tim Robertson, Phillip "Pipp" Priss, SaschaS, Tudor (2dor), Raksha, Arlington Audio, SlamminMofo, AnthillAudio, Res, nicolasrivera & many more. Paid packs produced by great studios are available too, however I've found simmilar quality levels between those and the free ones, discussions about the eternal problem of the not recognized efforts of the artists & sound engineers are allowed too...

NAM is not just about guitar & bass great amps neural capturin', but almost any piece of real hardware. PastToFutureReverb, Neutron Audio & other pro groups are achieving some incredible stuff from Fairchild compressors to ancient tube consoles, real speakers modeling and much more.

A very recent project promises to blow up the entire digital audio universe; GPU AUDIO.

They talk about dozens, even a hundred of NAM instances running in REAL TIME through GPU management, instead of our throtled CPUs...; can you imagine on that revolution? Lets talk about it before this Sunrise.

Of course, I guess GPU powered NAM won't be "free", but I'm sure will worth it.

Let's hear to those crazy geniously audio scientists;

EdiT; a few weeks ago on the NAM site it was mentioned that it would no longer be open source, I don't know what happened but that phrase isn`t present anymore, which however I feel as something positive.

 

Last edited: Feb 17, 2025 at 1:20 PM

I don't know what's better. That they actually have someone called a CTO, or that people really think that this shall be better than Amplitube 2 or whatever.

 

You have a smart business idea and the know-how about GPU! Of course you want to make money. I don't need it!

 

if it can be done "better" on the gpu , it can be done better on the cpu. the gpu is not some analog mojo - pumping device. if ampsims first, then mxing plugins should/will follow. it's just some nonsensical marketing bollox. i'm jamming here on my ampsims near zero latency, what can be better than zero latency?

 

A couple of weeks ago I've quickly made a simple guitar chain to demonstrate a "very forgetable" procedure about Loc Level using it as a preamp, the NAM capture is a very humble one of a BadCat 100watts real amp, not one of the best captures that I've downloaded from ToneHunt. Forget forever my foolish and reckless execution, but, did u really tried NAM?

 

I've seen a BUNCH of so called "GPU for audio".
It always ended fast.
Because needing to maintain TWO ways is always worse than ONE way to do things.

Some reading (and there are more) :
https://audiosex.pro/threads/gpu-for-audio.57679

At the end, convenience always win over power.

 

The wheel has already been invented. You can't reinvent the wheel twice. This is about building up hype to attract attention, because where else is the money going to come from but from users who are lured in with a promise.

All the great guitarists over the last 40 years got by without these things and they sounded great.
They could play, knew their amp and pickups and used high-quality hardware. That's all it takes.

Eddie Van Halen - Eruption Guitar Solo (Live in New Haven 1986) (WITHOUT NAM)

Gary Moore - Still Got The Blues (WITHOUT NAM)

 

Eddie is one of my forever idols, of course, I'll never reach some like that, also Gary, great Gary. So here we're talking about impressove analog gears too, something which I'll never could reach either. This thread could be a multiple pointless if all of we could reach that kind of real equipment, even worst about "talent"

 

Oh, yeah, I'm aware even since the start of it, as I've played amazed with Guitar Rig III many years ago... Now things moved forward a lot, is the nature of things.

 

Another Review of Neural Amp Modeler

 

Everybody knows tonewood is where its at. I want a pedal made out of like lebenty-seben kinds of rare tonewoodz and when i stomp that pedal box, BAM!!! more tonewood. I don't need no stinkin GPU, just some termite proof tropical wood. and a microphone/transducer and dsp chip and some a/d's and some buffers and half a dozen 9 volt batterys and cool finish with lots of sparklez, and a light so I can find it in the dark. oh and there should be an app to download more wood on my phone, cause why bother using it for phone calls?

and a really nice box so people can spend a half an hour unboxing it on the yew-tube would be ahhsummmm...

 

No, I exclusively use 3 (two of them 80% of the time) different amp's depending on the situation for everything - ever. Neither amplitube or nam or neural whatever.

Doesn't matter anyways because amplifier simulations will not get "better".

 

...well, I really think in money as a fair retribution for a good work, obviously, if someone's job/improvement helps me in some way, is it bad? Sadly I'm still poor enough to buy a decent gear but I can't be thankfull enough for all that TBB & I hope to retribute soonest I can all that I feel owe...

 

scientist at Gigglelabs (not associated with GoogleLabs in any way,shape or form) are working on cloning your own ears, to add a 2nd stereo pair, allowing humans to listen to quadraphonic sound in real time, also you will be able to wear 2 pairs of glasse at once (2nd nose is optional and payment is extra).

 

Nothing like the real stuff, of course, that's why digital audio has modeled sons & daughters.

 

Useless message - I removed

 

Last edited: Feb 16, 2025 at 11:21 PM

If they finally have a product on their hands that makes people go "yeah, that's useful" - good on them. But.. they've been hyping their tech since 2010 without ever showing products that show real advantages over standard CPU implementations ("GPU Audio" isn't their first rodeo).

chorus, flanger, delay -> can run a thousand of those on CPU anyway.
convolution -> who cares, 0.2% on CPU.
synth -> what happened to that one?

build hype -> secure more funding -> scale back -> product works better on CPU anyway -> find new application for tech -> repeat.

not saying it won't happen, but adjust expectations imo.

 

Yes, u'r right about most of the common VSTs use, the challenge borned from the bottlenecking happened while loading more than 2 NAM instances, as with some very heavy synths & some Acustica Audio plugins (far beyond many people find those AA absolutely dispensable). I'm experimenting with NAM & some astounding free profiles a lot, I can load 2 instances of slightly coded profiles, but there are complex profiles that sounds impressive & my current CPU can't decently handle them. I would like to experiment NAM without such kind of limits.

 

it would be far more effective to have an AI double tracking your guitar in real time, than yet another series of effects to stack on top of your guitar performance.

an ai that could provide a decent bass line and a tasteful piano or organ riff here and there, would be far more helpful and productive. and yes I've tried the band in a box type pedals that add drum and bass and they all have too many negatives to their handful of positives,its just easier and faster to DIY in a DAW than use such pedals for composition.

I think they are stumbling down the wrong path, but they must have done some research to see if their was a market for such a product, or else why would they bother?