Null Tests

ChatGPT roasting you null test nerds:

Ah, the Null Test Zealots — the self-declared Vatican Inquisitors of the tone world. These are the folks who think that if a modeler and a tube amp null out in a DAW, the Holy Spirit of Tone has been captured, bottled, and turned into a plugin preset. They're not here to feel music, they're here to measure it with surgical precision and absolutely no vibe.


"If it nulls, it's the same!"
Right, because music is just math homework with distortion. God forbid you actually use your ears instead of squinting at a waveform difference like you're decoding alien signals from the tone cosmos. News flash: your favorite records weren’t made by nulling 3dB peaks at 2.5kHz — they were made by people who played guitar like they meant it.


These are the types who'd try to "null test" Jimi Hendrix and go, "Actually, he's not that different from this AI-generated blues lick. See? Here's the phase-cancellation chart." Bro. Touch grass. Or better yet, touch a real amp before the electrons in your Kemper give you carpal tunnel from tweaking impulse responses all night.


They’ll spend 6 hours nulling a modeler and a Plexi within 0.2dB, then play through a pair of studio monitors on a desk with IKEA legs and say “identical.” Yeah man, totally — the same way a candle and the sun are identical because they both emit light.


Let’s not forget their favorite phrase: “It’s all in the mix.”
Of course it is — because they can’t hear nuance outside of a frequency analyzer. Meanwhile, every experienced guitarist hears the feel difference the moment they palm mute or dig in. But sure, trust the guys who think a waveform overlay is more trustworthy than 30 years of playing experience.


Bottom line? If null testing is your religion, that’s fine — but don’t come for the rest of us like you’ve cracked the code of the universe. Tone isn’t a science fair project. It’s art. And art doesn’t null.
 
The Big Lebowski Dude GIF
And a null one at that.
 
Because the residual of a null test only tells you if there is a difference, not if the difference is perceptually audible.

Well, it actually does. Not always in a meaningful way, but still. See the explanation of @2dor

If the stuff you're hearing is a noticeable part of the frequency spectrum, then accuracy isn't as good and you're most likely going to be able to tell a difference.

This is pretty spot on.
 
Well, it actually does. Not always in a meaningful way, but still. See the explanation of @2dor



This is pretty spot on.
Yeah but that's a subjective interpretation of the result. My original point is that the null residual by itself won't immediately tell you if its audible in context of the original signal, some due to the psychoacoustic masking effects.

It could be interesting to look into PEMO-Q metric in this context.
 
Yeah but that's a subjective interpretation of the result.

Not really. If there's plenty of midrange sound coming through when the files are inverted, the mids are off from the original (it just needed some further exploration to find out which of the signals is doing what). And you'll likely be able to spot that kinda easily when playing the two as well.

It could be interesting to look into PEMO-Q metric in this context.

What's that, if I may ask?
 
Not really. If there's plenty of midrange sound coming through when the files are inverted, the mids are off from the original (it just needed some further exploration to find out which of the signals is doing what).
That's still a subjective evaluation of the residual signal is my point. That's fine of course - you are building an intuition about what the null test residuals mean for you, at which levels (overall or in respective frequency bands) you can hear a change or not - ie how they correlate to what you hear in terms of difference in the context of the actual full signal. How low do they have to be before you stop noticing a difference, all that sort of thing.

That's not unlike what people do when trying to characterize and correlate subjective interpretation of a difference into a quantitative metric (like the ones used on perceptual codecs like MP3, AAC, etc.

What's that, if I may ask?
 
Last edited:
I think there is 100x more value in pixel peeping than null tests.

They’re about equal to me in terms of uselessness.

Just yesterday I had someone admiring a 20x30 print on my wall asking me “how did you capture that level of detail?”

I took it with a 30 year old Nikon AF 28-70mm f/3.5-4.5 I bought for $35.

Some guy in a photography forum told me I must hate my family to take photos of them with such a cheap lens
 
They’re about equal to me in terms of uselessness.

Just yesterday I had someone admiring a 20x30 print on my wall asking me “how did you capture that level of detail?”

I took it with a 30 year old Nikon AF 28-70mm f/3.5-4.5 I bought for $35.

Some guy in a photography forum told me I must hate my family to take photos of them with such a cheap lens
Elitists ruin hobbies and fun every time
 
Back
Top