NAM, Re-Amping, and going down the rabbit hole

There is this amazing open source project out there, called Neural Amp Modeler. The idea behind that is to use machine learning to capture the sound of guitar amps (or other gear) and use it in your DAW.

When I first discovered it, I instantly fell in love with it. I tried Kemper, I tried Quad Cortex… However, NAM was the first that literally felt and sounded right. Other than the aforementioned solutions, you need to do the capturing using your computer. Afterwards, you can choose to either let your computer do the number crunching (training the WaveNet neural network) or use google collab. The solution has quite a bit of a learning curve, as there’s no really stream-lined software like for example ToneX offers. However, I think the quality of the results is worth all the effort!

A few months ago, I discovered that a German company named Dimehead managed to put NAM into a pedal and I immediately ordered it. It arrived in the meantime and it is fun!

However, my honeymoon phase recently ended, when I plugged straight into my beloved Friedman Smallbox 50 and then used a model I created from the same amp right afterwards. They sounded different! They even felt different while playing. And that’s when I went down the rabbit hole.


I re-setup my re-amping chain, using my RME Fireface UCX II along with a Palmer trave re-amp box. Between load box and my interface’s input, I’m using a Palmer PAN01 to capture the DI signal from the amp. I sent that through my favorite IR on my PC. Then I plugged my guitar into the HiZ input of my interface and sent it through the Palmer trave to my amp. I compared that to plugging straight into the amp. The re-amped signal was lacking some presence, but it actually matched the NAM profile perfectly! The next step was to attach my oscilloscope to the trave’s output, sending out a sine wave at 1kHz at 0 dBFS, and adjust the re-amp box’s level to 3.08 volts RMS, which equals 12dBu. That should leave enough headroom for, for example, direct recordings of humbucker guitars. Still I couldn’t get closer to the experience of plugging straight into the amp.

The next step in my journey was trying a Lehle P-Split III I had lying around for a different purpose. Here, no adjustment on the box itself was possible, so I had to adjust my audio interface’s output accordingly. Still not getting closer to plugging straight into the amp.

The last thing I did was ordering a few more re-amp boxes and testing those.

The Boxes I have tested

To be honest, between the re-amping boxes, I couldn’t hear that much of a difference. So I ended up sending frequency sweeps through all of them and looked at the results. Unfortunately, I didn’t pay much attention to calibrate their outputs to 12 dBu, like I did before. So, pardon me for having the levels a little all over the place. I might re-do those tests in the future.

First, here’s a loopback sweep of my interface’s instrument input. The input was set to 13 dBu with the gain set to 0. The output of the interface was set to 13 dBu as well. REW’s sweep generator was set to -12 dBFS.

Now here’s the response of my Palmer trave:

I also tested a United Studio Technologies Replay Box. Like mentioned before, I didn’t calibrate it to output 12 dBu. However, I adjusted it by ear to get as close to plugging straight into the amp gain-wise, which coincidentally resulted in the box’s level knob to point exactly at 0 (on a scale from -inf to +4).

Transparent mode:

and here’s the harmonic mode of the same box:

Now, the Lehle P-Split III, which isn’t per-se a re-amping box, but works for that as well:

And finally, here’s a Walrus Audio Canvas Re-Amp. Level knob cranked:

I don’t know where those “waves” below 40 Hz are coming from. However, it’s telling me that you can’t go wrong with any of these boxes, with the Canvas being the exception. Your re-amped signal might actually sound way too bright/brilliant with that.


I couldn’t solve my initial problem of not achieving the same tone of a guitar plugged straight into the amp when I either re-amp or use a NAM model. However, after putting a bit of thought into that: When are we actually plugging straight into some amp? We often tend to use our favorite pedals in the chain, cables of different quality or length. All that clearly alters the sound. And what do we do? Of course! We adjust our amps to achieve a tone we connect with. The same can be done when we are re-amping. Adjust everything so you have a tone you love and then you should be set for capturing for your NAM models.

EDITED: After some feedback, I altered the article to better reflect the chain of events. Thanks for all the input! 🙂

The RME Babyface Pro FS and Linux

RME Babyface Pro FS

I recently got this beautiful little RME Babyface Pro FS and I’m quite stunned by build-quality and the sound quality. Using Linux most of the time, I’m really glad it has a Class Compliant mode. However, it features a DSP, allowing some basic mixer functionality like EQ, gain control and routing.

So, after using OpenViszla to sniff on it’s USB communication, I got in touch with RME. The result is a kernel patch that enables you to access the routing part of the DSP, as well as changing clock source, etc.

I’m also working on a (so far rather basic) GUI, since using standard ALSA-mixer with that many controls can be cumbersome. Also, the patch isn’t in mainline yet, so you’ll still need to patch your kernel yourself.

Instructions can be found here:

The GUI can be found here:

This also works for the Babyface Pro (without the FS).

Stranger Things – Kids – An Attempt to make a REV2-Patch

I really enjoy creating my own patches for the Prophet REV2. Dave Smith really knows how to design intuitive user interfaces, and that way, creativity and inspiration aren’t being obstructed by menu-delving.

The music from Stranger Things keeps inspiring me and so I decided to do a little experiment. I tried to re-create the synth sound on the hook-line of “Kids” on the REV2. I think, I got close, all though there’s still some high-end sizzling missing and probably a lot of LFO-modulation. Nevertheless, I’d like to share that patch with you. Perhaps it’s a kind of starting point.

In the video below, you can see how I approached the song on the REV2. If you can’t wait to check out the patch, here it is:

Nevermind the rainy weekend: Building a Tube Town PoS 100

No obligations, this weekend… And it’s raining. So I decided to assemble one of Tube Town’s great DIY Kits. This time it’s a Power Soak for my guitar amp. Here are some pictures. I’ll do some thorough testing later and post some results.

Here are some example files. I didn’t use a looper, as I wanted to test not only how it affects the tone, but also how it affects my playing. Well, it does. But hear foryourself:

First my amp with the Master Volume set to “Bedroom Level” without the PoS in the Chain:

MV_BedroomLevel (FLAC, 24 bit)

This is with the MV set to 1 o’clock (the level I use on the Klanghoehle Jam-Sessions) and the PoS reducing it to “Bedroom Level”:

MV_JamsessionLevel_PoS (FLAC, 24 bit)

And now with Master fully cranked and PoS reducing it to “Bedroom Level”:

MV_Full_PoS (FLAC, 24 bit)

Tell me your opinion in the comment section. 🙂

Tube Screamers rock!

Ever since I got my JTM-45 clone in summer 2015, I thought perhaps the Tube Screamer from Ibanez would be a nice addition, but I somehow hadn’t been motivated enough to buy one… until last week. I ordered a clone, The Screamer from Das MusikDing.

I had assembled a couple of kits from that shop before, but when I started soldering the Screamer, I was surprised about a small, but very convenient improvement over prior kits: Along with the various parts, a small PCB is being delivered. You solder the nine lugs of the footswitch to it and connect it to the input, output, LED and – of course – the main PCB. This completely removes the burden of spending an hour soldering various wires to the footswitch. Thumbs up, MusikDing! I also modified the circuit a little, making two resistores switchable, so you can choose between a TS808 and TS9 voicing.

After finishing assembling the kit, it instantly worked (Yay! Getting better at that. ;)) and as I cannot make much noise in my flat, I test-drove it using some impulse response and load box behind my amp. It sounded awesome! But that assessment turned into “Holy f*ck, this sounds frigging marvelous!”, when I used it at my friend’s rehearsal room. Especially on the TS9 setting. It’s exactly the overdrive pedal I’ve been looking for. It adds warmth, but also some grainy, crunchy cream topping. It works perfectly in front of other overdrives to create everlasting sustainy distortion sounds. Here’s a small blues jam from our last Klanghöhle Session. Usage of the TS starts at about 02:05.

Drums/Engineering: The C.L.A.
Guitar/Bass: Bollie

Enjoy. 🙂