St. Rock Amperium: Owners Club

Not sure if I completely understand but sometimes when changing power amps, it can mess with the volume and clipping. Also, my headphones clip but I need to get some new ones so I'm not sure how you are monitoring
Yes, that happens sometimes, but my question is about another issue, when I select for example bank 98 (an empty bank) that dry singal is clipping, but when I begin to activate blocks, stops the clipping (maybe my guitar is too hot?) so the question is, does that initial clip affects the signal lvl incoming to the next block? if thats the case, Im sending a clipping signal through all the chain, or im talking nonsense?
 
ohhh....

any ADC have low input level - about 1.25V.
Guitar have amplitude on attacks 6-8V.
Amperium recreate the input level to analogue levels for correct work with a drv\preamps\wah etc.
Output of all blocks have compensation to digital 1V level.
So....
When no one block is enabled - you have real level in chain. And it's more than 1V - in this case output will clip on attacks.
When you enable some block - then output level will decreased to right value.

I don't know how explain more easy.
 
Yes, that happens sometimes, but my question is about another issue, when I select for example bank 98 (an empty bank) that dry singal is clipping, but when I begin to activate blocks, stops the clipping (maybe my guitar is too hot?) so the question is, does that initial clip affects the signal lvl incoming to the next block? if thats the case, Im sending a clipping signal through all the chain, or im talking nonsense?
Like @st.Rock already said, the processed signal can sometimes be a lower amplitude than the unprocessed signal. It's completely within the realm of possibility that your DI (unprocessed) signal can clip the Output, but the processed signal (with blocks active) will not.
 
if you need simple answer :
This is absolutely normal and right if you have output clipping on bare chain.
 
if you need simple answer :
This is absolutely normal and right if you have output clipping on bare chain.
Cooool and how do I know that I’m sending the right amount of input signal to the chain? I want to send the right amount to not overgain effects or amp
 
We recreate the original level by default. In the beginning of the chain you have what goes from a guitar. You can increase or decrease by Input knob. It's advantage of hardware solution.
But you are think in planes of soundcard-vst area. We have already care about correction of right level.
Now you can breath easy. -)
 
Here's the current live setup. Honestly it doesn't have to have the Devil's Triad pedal in there but I just couldn't justify not using it, and it is a great pedal in its own right. The delay and reverb are going into the Amperium's loop and the OD and boost are in front. One output goes to FOH and has an IR, the other to the Baby Bomb power amp to my cab on stage. All of it sounds great.
20240121_154659.jpg
 
We recreate the original level by default. In the beginning of the chain you have what goes from a guitar. You can increase or decrease by Input knob. It's advantage of hardware solution.
But you are think in planes of soundcard-vst area. We have already care about correction of right level.
Now you can breath easy. -)
Holy sh*t that's was a clever move, I'll just let the input knob at 0, I was experimenting today with some high gain stuff, I still can't understand how the hell the high gain stuff sounds so damn good, where's the magic? BTW why there's three 5153 poweramps? 5153, 5153 100G and 5153 100BR?, I prefer the 5153 100G with the Leon's impulses.
 
St.Rock - Any chance you could make a Roland JC120 model, perhaps with matching chorus effect?

Also, we desperately need a youtube video to show us how to use the training to make our own nnm profiles. I am confused with how to do it.

Update: With some trial and error, I have finally got to the stage where it's training the profile. For reference, each Epoch looks like it takes about 26-30mins or so.
 
Last edited:
NNM training - If I get disconnected, how do I continue from where I left off?

I get this error;

NameError Traceback (most recent call last)

<ipython-input-4-0c2e3db9586d> in <cell line: 12>()
10 epochs = 2 #@param {type:"integer"}
11
---> 12 train_model(x_train, y_train, x_test, y_test, model, epochs, batchSize)
13
14 mfname = name+'.nnm'


NameError: name 'train_model' is not defined

Which I assume means I've got to reload everything?

If so, how do I resume training for a disconnected session?
 
St.Rock - Any chance you could make a Roland JC120 model, perhaps with matching chorus effect?

Also, we desperately need a youtube video to show us how to use the training to make our own nnm profiles. I am confused with how to do it.

Update: With some trial and error, I have finally got to the stage where it's training the profile. For reference, each Epoch looks like it takes about 26-30mins or so.
It seems that GPU acceleration was not activated in colab.
Make sure to get a GPU! (Runtime->Change runtime type->Select "GPU" from the "Hardware accelerator" dropdown menu)
As stated in head section of NNM_colab.ipynb


ps. What part of initial video I should remake/revise or clarify?
 
BTW why there's three 5153 poweramps? 5153, 5153 100G and 5153 100BR?, I prefer the 5153 100G with the Leon's impulses.
5153 means 50W version with C-type presence pot (it`s like standard presence at settings below 50%).
5153BR means 100 W with standard presence pot in Blue and Red modes, 5153G means 100W Green mode. There is some slight difference.
 
Guys I found a dude who made React IR compensated IRs these are crafted to react well with the impedance curve simulation of the React IR and by default with the ReactIR curve of the Amperium, bought the mesa trad ones and they sound fantastic.

 
Guys I found a dude who made React IR compesated IRs these are crafted to react well with the impedance curve simulation of the React IR and by default with the ReactIR curve of the Amperium, bought the mesa trad ones and they sound fantastic.

The dude abides and can comment further in this very forum.

@MirrorProfiles
 
Hey @st.Rock @deLuther The latest 1.92 beta is from January 7 and I was wondering if you have any news about upcoming updates and what features you have planned for the future. I hope you are both doing well.
 
Back
Top