Stompstation by Sonulab (new NAM player pedal)

2dor

Roadie
Messages
843

Stompstation by Sonulab is a minimal and compact guitar pedal with these features:
  • NAM Profiles Player
  • IR Loader
  • FX: Noise Gate, Comp, EQ and Reverb
  • Tuner
  • MIDI, Bluetooth, WiFi, USB
  • 2 In (with Pad)
  • 2 Out (TRS Balanced)
  • A/B or ON/OFF Mode
 
1729751988620.png
 
Assuming (?) this proceeds to completion, manufacturing and then sales, and

(a) it can load any 3rd party NAM Captures
(b) it can load and play the highest resolution / highest quality NAM Captures
and
(c) it remains at no more than EUR $250

It will sell like the proverbial hot-cakes.
 
Assuming (?) this proceeds to completion, manufacturing and then sales, and

(a) it can load any 3rd party NAM Captures
(b) it can load and play the highest resolution / highest quality NAM Captures
and
(c) it remains at no more than EUR $250

It will sell like the proverbial hot-cakes.
Apparently it can load the highest quality NAM profiles that come with the NAM code by default; those use, what Steve calls, the "STANDARD" architecture. The other architectures / profile types (FEATHER, LITE & NANO) are lighter to run bun aren't as accurate.

I doubt this will be able to run custom architectures. I've trained a few models which are 10x below the "Great" threshold established in the NAM code already; the Dimehead NAM player can't / won't run them.

I think any solution that comes out will only support or retrain the models to one of the "canned" architectures Steve supplied; maybe a little wiggle room to spare but my hunch is they'll be at the same level as the "standard" architecture or below. Even if that's the case, I'd still be alright with this stuff as the "standard" architecture already blows the socks off anything else that does profiling at the moment.
 
What's interesting is the 2 inputs / 2 outputs.
Guess it can either run 2 parallel paths or a stereo one.
 
Well it looks like it'll have some sort of app; somewhat of a redeeming quality.

Seriously, while I like wireless apps, a pretty much "amp only" box should allow for better adjustment straight on the unit. I mean, what's next? You'll be adding a GT-1000 or whatever for anything else and edit that using another app?
One single app to control your entire sound at once is great, tons of apps to adjust individual portions of your signal path just suck.
On an amp-only pedal, I expect to see at least 3 encoders. Gain and volume should always be adjustable simultaneously without any additional actions, same goes for the BMT part of the tonestack (HX Stomp, looking at you as well, when will this be adressed?). Ideally, there should be at least 5 endless encoders with readouts. Gain, volume, BMT. And an analog pot for the overall volume. That would make such a device usable, anything else ends up in whatever crutch approaches.
 


photo_2024-10-24_10-57-07.jpg

The board and the DSP module cost 83€, couldn't find the specific enclosure for sale but, for probably less than 100€ and if the firmware with the nam loader can easily be installed on a diy one, I'll definitely build one myself.

Well it looks like it'll have some sort of app; somewhat of a redeeming quality.
Yes, the development kit has an app: https://play.google.com/store/apps/details?id=com.voidx.dsp_nodes
 


View attachment 31635

The board and the DSP module cost 83€, couldn't find the specific enclosure for sale but, for probably less than 100€ and if the firmware with the nam loader can easily be installed on a diy one, I'll definitely build one myself.


Yes, the development kit has an app: https://play.google.com/store/apps/details?id=com.voidx.dsp_nodes
It's pretty cool.

That input headroom is a bit low but yeah - I could work with it.

I'm now curious what corners they need to cut in order to run standard NAM profiles on the thing.
If I'm to get a NAM player, I'm not keen on sacrificing the main quality of NAM: fidelity to the source.
 
It's pretty cool.

That input headroom is a bit low but yeah - I could work with it.

I'm now curious what corners they need to cut in order to run standard NAM profiles on the thing.
If I'm to get a NAM player, I'm not keen on sacrificing the main quality of NAM: fidelity to the source.
1729762263101.png
 
A good mate of mine - Alex (y)- emailed Sonulab via their current website with some questions and got this response ... he passed it on to me, but it doesn't mean a lot to me as I'm not into A.I Capturing methodologies and techniques etc... but those of you that are will probably understand it:-

" ..... Sonulab employs VxAmpEngine technology for NAM model conversion, a black-box system for modeling guitar amplifiers. Included with StompStation is a conversion tool that identifies NAM models and translates them into a computationally more efficient format for direct execution on an embedded system. The conversion process extracts 99% of the NAM model’s characteristics, so slight differences may be observed. In fact, the neural network is converted into a block-based model, reducing nonlinear operations and allowing them to be executed in oversampling. Numerous tests have shown that the result will have less aliasing and may, in some cases, appear slightly more defined compared to the original model. In addition, we are excited to announce that the product will be available for commercial release in the early months of next year. Stay tuned for updates, which you can find on our website ..... "
 
A good mate of mine - Alex (y)- emailed Sonulab via their current website with some questions and got this response ... he passed it on to me, but it doesn't mean a lot to me as I'm not into A.I Capturing methodologies and techniques etc... but those of you that are will probably understand it:-

" ..... Sonulab employs VxAmpEngine technology for NAM model conversion, a black-box system for modeling guitar amplifiers. Included with StompStation is a conversion tool that identifies NAM models and translates them into a computationally more efficient format for direct execution on an embedded system. The conversion process extracts 99% of the NAM model’s characteristics, so slight differences may be observed. In fact, the neural network is converted into a block-based model, reducing nonlinear operations and allowing them to be executed in oversampling. Numerous tests have shown that the result will have less aliasing and may, in some cases, appear slightly more defined compared to the original model. In addition, we are excited to announce that the product will be available for commercial release in the early months of next year. Stay tuned for updates, which you can find on our website ..... "
Doesn't disclose too much. There was a very intriguing post this month in the NAM group about something similar:

1729948012257.png
 
Back
Top