The most accurate capture system: A frequency analysis

[RiffDuck]

God I fucking hate that guy. Literally the least scientific person on the planet.

 

[OneEng]

A null test is the ultimate test. If something matches in the time domain it will match in the frequency domain. The converse is not true (assuming you're not comparing phase which no one ever does).

The issue is interpreting the error signal.

Whether to use LUFS vs. some other weighting (or no weighting at all) is a secondary issue. LUFS is just one method of determining "loudness" which means weighting the spectrum and integrating. You could use no frequency weighting at all and use an RMS measure.

The ultimate measurement would be a peak and average difference. LUFS gives you a moving average of the error. A measurement that also showed instantaneous peak error would be a more complete solution. If a capture device doesn't handle palm mutes well this would show up in the peak error. Or you could use two different time constants. A short one for transient error and a long one for average error.

I've been seeing people claim that LUFS is dependent upon the source level. If done correctly it shouldn't be. You should be looking at the DIFFERENCE between the source LUFS and the LUFS of the error signal.

All that said I had a good laugh the other day. Someone was posting that one of the capture devices, I don't remember which, was displaying -20 dB LUFS error. I took the latest Axe-Fx firmware and compared it to our Deluxe Reverb and Plexi reference amps and was seeing between -26 and -30 dB.

I think your ears and your fingers are the ultimate test; however, I don't disagree with most of what you said.

Still, there are things that throw a NULL test way off:

• Phase/Time alignment issues
• Frequencies that are outside of hearing range
• Level matching issues
• others?

What I found interesting about THIS particular video is that the graphics matched what my ears were hearing. This is NOT always the case with NULL tests for whatever reason there is for it.

What I personally believe is that modelers and captures have gotten QUITE good and are likely to get even better. We should be spending more time evaluating how easy it is to get a great tone out of a device, or how well the device works in a live situation, or how good it is for recording. I see a billion videos comparing accuracy, and very few talking about how well everything works together to meet a particular users needs.

For me, based on all I have heard, the features, sound quality, and usability in the Stadium are near to or actually best in class in the market. Some devices still have more things they can do or amps they have in the arsenal, but this may not be the case in a year.

This video didn't change my mind on Stadium, or if anything, further made be think it's a pretty well thought out device.

 

[banned]

Why don't folks also compare RMSE error on a time-series basis? (I.e in addition to lufs)

That can provide a comparative indication of time-series error. Wouldn't it?

You could do it with a few different playing styles to get an idea.

 

[OneEng]

Why don't folks also compare RMSE error on a time-series basis? (I.e in addition to lufs)

That can provide a comparative indication of time-series error. Wouldn't it?

You could do it with a few different playing styles to get an idea.

True, and not a bad idea in theory (to take time series errors down to where they are happening).

I still really question how much it matters if you can't tell the difference in a raw recording without computer digital analytics. Once you add effects and put it in the mix, it will most certainly be indistinguishable from the original IMO.

 

[laxu]

I'm being a bit flippant of course. But in a general sense, I just love guitar amps as a system. Rather than cobbling together a modeller, a poweramp, IEM's, some kind of FRFR monitor, or wedge. I truly think that by the time you've shoved all that stuff together, tested it, got your tones... most people would've been better off with a basic 50-watt head or combo; easier to deal with than a modularized setup that a lot of people are going for these days.

I agree. At some point it just gets too complicated to manage - or people struggle with some specific aspect of it like they don't like the FRFR thing, don't like whatever poweramp that is practical etc.

That said, a lot of digital gear quickly gets into this. I've got a Fractal AM4 on my desk and it's a fun unit, but it really drives home how I don't need every tone known to man at my fingertips. It's fun to play with, and a good way to get rid of GAS by just trying some new amp model, but ultimately I always just end up at square one: a medium-to-high gain modded Marshall tone with some pretty barebones fx.

At the same time, for touring... something like Quad Cortex or Stadium... in a bag... jump on a plane... plug into the PA at the venue. Done. That part of it is really cool.

Which is why I love the BluGuitar stuff. It's right in that in-between space where they're just super simple to use, real amps that you can just throw in a backpack because they are so small and light.

Nothing beats the charm of tube amps though. At this point it's not IMO even about how they sound or feel, but are kind of like an experience unto itself. It's become almost like listening to vinyl records. Yes, you could argue that this and that gear is more practical, does more, costs less, is smaller etc, just like you could argue a CD or streaming service is more practical or even better sounding. But it's still more fun to put on a vinyl record as a ritual, or wait that tube amp to warm up, smell the dust baking off the tubes and then rock the shit out of it.

 

[NKP]

Interesting to see that the lest expensive one (Tonex) does the best capture . Also nam too but that’s not a hardware