• TalkBass has been independent since 1998. Add your voice.
    Create a free account to reply to discussions, view embedded media, and browse with fewer display ads.
    Join freeLog in
    Want zero display ads or expanded classifieds tools? Compare plans.

NAM (Neural Amp Modeler) big thread

I am a Tonex user but I see that NAM has interesting and growing community.

I’ve done some captures from my amps and stomps. I’ve checked Sansamp and Bass Attack with NAM profile and it works nice. Anyway it should work in context and mostly helps you to not take giant amps and cabs.
 
I am a Tonex user but I see that NAM has interesting and growing community.

I’ve done some captures from my amps and stomps. I’ve checked Sansamp and Bass Attack with NAM profile and it works nice. Anyway it should work in context and mostly helps you to not take giant amps and cabs.
Grabbed these yesterday and used them immediately! Cool stuff!
 
  • Like
Reactions: kevk and eve_ripper
So I tried a few test captures, in the sense that I loaded things into Audacity, applied an effect. I want to see what it might be possible to model and might load on a Valeton GP5 or Sonicake Pocket Master to get around the lack of bass effects

* No effect: works fine. (always a null test)
* Overdrive: fine.
* Octave: no

And then via REAPER and Guitar Rig:

* Filter sounds: no (time dependency)
* MIDI triggered sounds for fun: no

I'd love to find a way of capturing my Bass V-Amp synth sounds as they are remarkably good for a budget unit, but that fails on the time dependency, I suspect, even if turning time dependency checks.

My background for 25 years was research into what is called AI, not that we called it that then, so when I get time from not working, making music, family stuff, pets, family history, buying and selling musical kit and DIY I'll look at the code. So the Christmas break when I'm fed up of mince pies and reruns on TV.

I suggested using 'AI' to model amps in about 1998 but did nothing about it back then. The hardware wasn't there to support it effectively real time so it seemed fairly pointless.
 
I've been continuing to push the boundaries.

I can get Moog-type filters to work and I am working on using some of the Saffron series of soundfont players to see what I can make it do. I will need to download some sound fonts that aren't just different bass sounds, though. But since Mike Oldfield will be touring Tubular Bells I-III soon I am just about to try a bell soundfont.
 
Having probed the limits (Moog synth sounds possible but that's about it, which didn't surprise me) next is:

* Can I get the NAM plugin working under Linux

* Can I get a VST host to create a live setup working with Linux on ARM and boot reliably into it (emulated at first as I don't have a spare RPi3)

* Can I control it via a M-Vave Chocolate?

If so, I might be able to create a set up in about the same footprint as a Valeton GP-5 but with the option of running other VSTs like the Saffron plugins with some sound fonts as well, so more flexible than a GP-5, although I'd have worries about the reliability on a gig and it wouldn't easily be battery powered. But it might be fun.

Trying to put any sort of shim for showing patch names on a tiny display would be tricky. So, I would just have to remember what patch 3d was. A VNC server running at boot to allow access would be sensible.

Arguably, it would be more productive to volume level patches on my Zoom B3n and practice more but this is a shiny technical challenge.
 
Having probed the limits (Moog synth sounds possible but that's about it, which didn't surprise me) next is:

* Can I get the NAM plugin working under Linux

* Can I get a VST host to create a live setup working with Linux on ARM and boot reliably into it (emulated at first as I don't have a spare RPi3)

* Can I control it via a M-Vave Chocolate?

If so, I might be able to create a set up in about the same footprint as a Valeton GP-5 but with the option of running other VSTs like the Saffron plugins with some sound fonts as well, so more flexible than a GP-5, although I'd have worries about the reliability on a gig and it wouldn't easily be battery powered. But it might be fun.

Trying to put any sort of shim for showing patch names on a tiny display would be tricky. So, I would just have to remember what patch 3d was. A VNC server running at boot to allow access would be sensible.

Arguably, it would be more productive to volume level patches on my Zoom B3n and practice more but this is a shiny technical challenge.
I'm not sure there is currently a VST compatible pedal, at least not fully. I think a lot of us would like to be able to take the VST plugins they are using with they DAW, and use them live in a pedal (without a computer).

Here's what ChatGPT says:

Option 2: Use Hardware VST Hosts (Standalone VST Pedals)
Some hardware units are designed to run VST plugins without a computer:

Examples:
  • Neural DSP Quad Cortex (doesn't run VSTs, but functions similarly)
  • MOD Devices Dwarf or Duo X (runs LV2 plugins, but not native VSTs directly)
  • ELK Audio OS devices (runs native VSTs on custom Linux hardware)
  • Aurora DSP ReVolt and similar emerging VST-capable hardware

But I haven't seen ELK Audio OS (or Aurora DSP) compatible pedals yet. Please correct me if I'm wrong
These things evolve quickly, it might change in the near future.
There are already a few NAM compatible pedals, they might evolve to allow for VST plugins by software update in the future?
 
I'm not sure there is currently a VST compatible pedal, at least not fully. I think a lot of us would like to be able to take the VST plugins they are using with they DAW, and use them live in a pedal (without a computer).
Indeed. Hence RPi seeming to be the only option here at present, unsatisfying though it is. I expect I can get things working with effort but I'd worry about it not booting or bringing up all the audio stack at a gig for reasons and then having to debug it. So it would be a project for fun as I can probably get acceptable noises from by Zoom B3n. The weak link is between my brain and the strings anyway.
 
  • Like
Reactions: gtirard
Looking at current RPi 5 prices, getting a second hand laptop or NUC sized PC might be as well to run NAM plus effects.

MODEP – Virtual Pedalboard for Raspberry Pi is what I was looking at.

I'll open a separate thread on that.
Looks fun, but be careful, NAM is very CPU intensive, I'm not sure a RPi can run a NAM plugin easily.
Anyway if you're able to build a RPi based pedal than can run NAM models and VST plugins, please report and share.

:hyper:

That would be really cool.
 
  • Like
Reactions: GotGroove
I have a question:

When capturing, will noise from bad power supplies in the chain you are capturing be an issue?

I am asking because I ponder on investing in a pedal that I capable of running NAM captures, to capture my "amp-less" basic "clean" tone chain, which consists of several pedals/preamps, to be able to cut down my overall effect chain significantly, but I don't currently have my setup hooked up on my board that have proper isolated and filtered power supply installed, so currently it is quite noisy.
 
Last edited:
  • Like
Reactions: -Asdfgh-
Looks fun, but be careful, NAM is very CPU intensive, I'm not sure a RPi can run a NAM plugin easily.
Anyway if you're able to build a RPi based pedal than can run NAM models and VST plugins, please report and share.

:hyper:

That would be really cool.
I've only run them on my desktop so far. I'm going to be checking the load and creating an RPi VM to check this all first. Memory is also a concern.

I'll report back, but it might in part be via the laptop gigger thread is I can find it.
 
I have a question:

When capturing, will noise from bad power supplies in the chain you are capturing be an issue?

I am asking because I ponder on investing in a pedal that I capable of running NAM captures, to capture my "amp-less" basic "clean" tone chain, which consists of several pedals/preamps, to be able to cut down my overall effect chain significantly, but I don't currently have my setup hooked up on my board that have proper isolated and filtered power supply installed, so currently it is quite noisy.
You can test it.

Record a track in your FAQ

Create a new track and add a media file of noise.

Render both tracks down to one mono result.

Go the the NAM website and fire up the model generator and build a model and download it.

go back to your DAW and with your original recording solo, add in the NAM plug in with your model and apply to that track. See if the result sounds noisy.

That's assuming it's constant noise like mains hum

I'll test it in a bit
 
  • Like
Reactions: Ritter667
I really do think I may shell out for the Anagram.

Right now my amps are largely a series of very old, heavy Peavey heads, Bi-Amped. I'm constantly experimenting with different cabinets. The weird EQ on the Peaveys is crucial to my sound and I haven't found another way to get it in a compact package, minus a custom pedal. I also like being able to experiment with other stuff. All told, at the moment, my current rig weighs 230lbs.

My thinking is I can capture the two heads and run them in NAM with different impulse responses for recording, and then also run them into something like the Anagram or the Dimehead in the original post. Both of these solve the issues I've had with other hardware modelers - they're either too complex (I don't need all the features of a Kemper or Axe-FX and find their UIs difficult) or too simple (I don't like IK Multimedia's customer support/licensing and the ToneX don't have an XLR out to run to FOH).

Is there something I am missing here, or am I pretty much the ideal candidate for NAM?
 
  • Like
Reactions: gtirard
You can test it.

Record a track in your FAQ

Create a new track and add a media file of noise.

Render both tracks down to one mono result.

Go the the NAM website and fire up the model generator and build a model and download it.

go back to your DAW and with your original recording solo, add in the NAM plug in with your model and apply to that track. See if the result sounds noisy.

That's assuming it's constant noise like mains hum

I'll test it in a bit
The result is that you can capture if it's continual noise.

I added the same amount of noise as from some thirty year old recordings have because I grabbed the noise from a thirty year old recording. The result, though, is a fuzziness to a bass track more than the bass track overlaid with the noise.

So, whilst you can capture it, it sounds pretty bad. If you take the noisy signal and noise gate it, then it sounds okay. So those trying to record noisy amps will need to experiment with noise gating their recordings prior to training, it seems.
 
  • Like
Reactions: Ritter667
First is clean bass with artificially added hum captured from an old recording. The second is the model built from the NAM input file and the same hum, with the same clean bass recording and no added hum. As you can tell, it's made the bass sound fuzzy. If you put a noise filter on the noise to take it out from the input file, it is probably fine, although I haven't tested it.
 

Attachments

I really do think I may shell out for the Anagram.

Right now my amps are largely a series of very old, heavy Peavey heads, Bi-Amped. I'm constantly experimenting with different cabinets. The weird EQ on the Peaveys is crucial to my sound and I haven't found another way to get it in a compact package, minus a custom pedal. I also like being able to experiment with other stuff. All told, at the moment, my current rig weighs 230lbs.

My thinking is I can capture the two heads and run them in NAM with different impulse responses for recording, and then also run them into something like the Anagram or the Dimehead in the original post. Both of these solve the issues I've had with other hardware modelers - they're either too complex (I don't need all the features of a Kemper or Axe-FX and find their UIs difficult) or too simple (I don't like IK Multimedia's customer support/licensing and the ToneX don't have an XLR out to run to FOH).

Is there something I am missing here, or am I pretty much the ideal candidate for NAM?
I do think you're the perfect candidate indeed, at least if you're always using the same EQ settings on the Peavey head - make one capture for each favorite EQ setting and call it a day. If you are used to tweaking the Peavey's weird EQ each time you use it, you won't find a virtual reproduction of the Peavey's EQ on the Dimehead or the Anagram.
 
Last edited:
I've just finished capturing my amps, preamps and pedals:


Feel free to comment (some of them may need improvements)