Darkglass Anagram - NAM profile player & multieffects unit

Yes I tried to perform a null test and yes let’s say I wasn’t able to have some acceptable results. But I used the AD to reamp and that’s why I guessed it was 3 possibly causes :
1- NAM LV2 implementation used in the Linux based device, baked by Olifan is not giving the same results as Steve’s plugin DESPITE using the same NAM core (or from what I can understand) : can be solved with future but need some work., and not 100% sure
2- Darkglass is using some instant conversion or trickery to simplify the nam cpu load (which I assume not being the case, trusting their marketing campaign) and the company overall
3- It’s just a simple audio path/AD conversion divergence, or maybe a bass tuning trick I don’t have the knowledge of, or maybe some focus on some frequencies?

So I guess Leo will be able to figure it out, if results are different with reamping USB, or if it’s a firmware update that is needed and so null testing right now is pointless, or anything else.

But let’s say for now only Dimehead has really proven but independent testing that they can deliver pure standard native NAM in a pedal.

for us guitarist right now the question of ANAGRAM is : is that the first high end multi NAM std 1:1 loader made by a renowned company. For now effects are just ok, not justifying the price tag on their own for a guitar player imo.

What’s sure is that they put enough power in the box to deliver what they promised.

Also the silence of Leo is strange, as he clearly doesn’t say anything about null testing despite being the most reliable of YT « scientific » guitar geek.
 
BenIfin said:
As to this "AT" guy - what's his game/angle or is he just the classic old school Forum b.s troller-poster ?
Click to expand...

He joined this forum here to pretend he's a fanboy/member of his own anti-alias cult, where he was spamming about 'super input' - which didn't work as advertised and was dismissed by everyone except his most loyal cult followers.

He's now back pretending that he "knows" something no one else does about something else, which we can also safely dismiss as nonsense.
Merely an attempt to shit stir.

I have observed his behaviour for years at this point.
Outside of the guitar/audio communities he goes around posting the most vile, divisive, racist/bigoted/crap.
A horrible person if what he posts is his actual way of looking at the world.

He's banned from many communities now because he gets angry/rude/spiteful if he's called out.

I am holding back so much here, and I wish I didn't even have to mention any of this, but, here he is still trying to waste everyone's time.
 
Dom Mcsweeney said:
I have observed his behaviour for years at this point.
Outside of the guitar/audio communities he goes around posting the most vile, divisive, racist/bigoted/crap.
A horrible person if what he posts is his actual way of looking at the world.
Click to expand...

Ah so this is the NAM S.I better aliasing person !!! Now I get it .... the "solution looking for a problem" guy !! where the solution didn't actually "solute" anything !!!!!!!!!!!!!!!

^^^ This stuff ..... is f*cked up ..... maybe he'll get the hint and just disappear out of here.
 
master wrestler GIF


GIF by The Jerry Springer Show
 
But for sure I solved the aliasing thing in NN. The fact is that sine sweeps in various levels, helps wavenet and LSTM arch to autocorrect their aliasing. Of course Steve found it wasn’t worth the try more than that, because he constantly stated it wasn’t working without even going deep into that. I had the confirmation from wavenet developers is is the way to get rid of aliasing to do that. But fortunately it doesn’t change much in terms of null testing, so for you don’t use high harmonics contents as input in inference (using a drive pedal before the nam) it doesn’t change a lot, But gain staging nam will greatly benefit from super input nam trained.
I’m also glad ToneX/NDSP people and many more got the info and plan to mimic the effect of this auto correct aliasing trick without the need to have 10 mn data.
Steve is not that genius after all and that’s what I was the most disappointed about alongside with the lack of open mind.
But at least nam is open source and Steve was able to bring some great competition. Maybe he though that the project would have been joined by some great machine learning developers but NAM is essentially Steve and Olifan efforts, with some fanboys around ;)
And I’m a pharmacist, not a NN dude so I won’t help that much. But at least Cliff from FAS now can’t say anymore that NN capture are doomed and full of aliasing ;)
 
If you wanna do a null test you have to realign the audio. It's key to getting accurate results. Each unit will have a different latency, so you can't just invert the model recording vs the original recording and get the result. You have to perfectly align them in the timeline. Having one of the recordings off by only a MICROSECOND will make that result significantly worse. That's where my trust in the results of other peoples null tests is limited.
Besides that, I doubt the sense of null tests anyways. It sounds good or it doesn't.
 
Rev. Irreverent Revenant said:
FYI - it looks like Marcos is doing an AMA on r/Darkglass.
Click to expand...

hunter said:
If you wanna do a null test you have to realign the audio. It's key to getting accurate results. Each unit will have a different latency, so you can't just invert the model recording vs the original recording and get the result. You have to perfectly align them in the timeline. Having one of the recordings off by only a MICROSECOND will make that result significantly worse. That's where my trust in the results of other peoples null tests is limited.
Besides that, I doubt the sense of null tests anyways. It sounds good or it doesn't.
Click to expand...
Thanks but let’s say I know how to do. I even made an html5 to auto align, phase invert and normalize 2 wav. And of course I have all the DAWs to perform it well.
But as I said using guitar DI and nam reamp is giving me a bad null test (-23 LUFS on 10 mn test). And A/B testing is not very good too, doesn’t feel exactly like nam plugin.
It’s a matter of day before Leo tell us the right results. He’s in close relationship with Darkglass so I will wait.
I started a « scientific » YouTube channel where I tested aliasing on flagship modelers but didn’t publish it. I tend to stop making video that can arm the device reputation because most of guitarist are not able to make any difference and we need people to buy modelers so the brands can spend in R &D.
 
hunter said:
You have to perfectly align them in the timeline. Having one of the recordings off by only a MICROSECOND will make that result significantly worse.
Click to expand...

Alligning clear attacks with sample accuracy isn't an issue in any DAW these days.
I think that getting a proper DI signal into any hardware to compare it with the NAM plugin is much more of a potential issue possibly causing sound coloration. If I was to do such a nulltest, I'd split the signal beforehand and run one side through the HW, the other side into the Hi-Z inout of my interface and straight into whatever NAM plugin. I would then bounce the NAM plugin, align the HW signal with the bounce and perform a nulltest. Still leaving you with input level calibration, but that could possibly be taken care of beforehand in case the tested device also functions as an interface.
 
You must log in or register to reply here.

Similar threads

texhex
Kemper Rack vs Stage
2
Replies
24
Views
3K
Antipodes
A
laxu
Revisiting the Strymon Iridium (vs Axe-Fx 3)
2 3
Replies
41
Views
9K
MirrorProfiles
MirrorProfiles
Back
Top