The most accurate capture system: A frequency analysis
- Thread starter eon
- Start date
James Freeman
Rock Star
- Messages
- 4,436
Why Miko... WHY??
You're not even Gen Z.
You're not even Gen Z.
- Messages
- 11,270
It is still a 1-dimensional reading, and no more superior than using LUFS to be honest. The reason being, the frequency content doesn't tell you much about transient response, it doesn't tell you anything about sag, compression, or other non-linear phenomenon going on in the circuit.Good stuff. Man, i'm so happy someone finally decided to compare profilers by frequency content rather than LUFSs
It is mildly better. But mildly mild.
7th order Hammerstein plots would tell you much more.
MirrorProfiles
Rock Star
- Messages
- 5,544
This. Too many non linearities and time domain information for them to be that helpful (because they’ll skew what you see). But it’s more meaningful than null+LUFS.
Lysander
Rock Star
- Messages
- 3,563
Disagree. It's way better because it provides context, instead of trying to reduce everything to a single (useless) figure.
For example, this shootout clearly showcases *why* Kemper is perceived as less accurate than the alternatives, in a way that "-40.6 vs -33.8 dBA LUFS" simply does not.
- Messages
- 11,270
Okay - just one question ... How do you read whether a transient has been smoothed out in a frequency plot?
- Messages
- 11,270
@Lysander for example... around 10minutes in, he claims Neural's v1 and v2 have different amounts of saturation. Fine... I can even hear it with my ears.... but he's using a frequency plot to judge saturation... which... well... a frequency plot alone is a pretty weak tool for judging saturation.
To measure saturation you'd use proper harmonic analysis, measure transfer curves, send in multi-tones to analyse intermodulation distortion, and send in various levels of dynamic signals to analyse the system.
And you know what the whole irony of this all is??
THAT'S EXACTLY WHAT THE CAPTURE TECHNOLOGY IS DOING IN THE FIRST PLACE!
To measure saturation you'd use proper harmonic analysis, measure transfer curves, send in multi-tones to analyse intermodulation distortion, and send in various levels of dynamic signals to analyse the system.
And you know what the whole irony of this all is??
THAT'S EXACTLY WHAT THE CAPTURE TECHNOLOGY IS DOING IN THE FIRST PLACE!
Lysander
Rock Star
- Messages
- 3,563
I... don't? Not sure what that has to do with a frequency plot being better than a null delta LUFS thou.
- Messages
- 11,270
We're talking about whether the frequency plot is still a 1-dimensional reading or not. And it is.
I could probably agree that it provides more context on a frequency basis. But it is still only showing you frequency. It isn't showing you how the system behaves; which is the entire point of doing an A/B comparison or making any kind of accuracy claim in the first place.
LUFS alone == shit.
Frequency response alone == shit.
Lysander
Rock Star
- Messages
- 3,563
No mate, you're talking about it
I never claimed frequency plots to be the end-all to profiling accuracy.It's still much better approach than a LUFS metric without baseline.
- Messages
- 11,270
So... why did you quote my comment about it still being a 1-dimensional reading then?No mate, you're talking about it
It's still much better approach than a LUFS metric without baseline.
I said two things:
1. It's a 1-dimensional reading.
2. It's no more superior than using LUFS.
And you said you disagree. It's way better because it provides context.
So I asked you for a specific factorial within the wider context, and how you'd measure it? And your response is... er.. don't?? lol
I already said that I thought that the frequency response information was mildly more useful than LUFS. But yes... mildly. There's so much it doesn't tell you, because it simply isn't trying to tell you. So why he'd rely on it in his video is anyone's guess.
The simple truth is; the video is just as unscientific as any of Leo's is. He's just happened to get a bit closer to pinning the tail on the donkey.
Lysander
Rock Star
- Messages
- 3,563
- Messages
- 11,270
Yeah, I got that. I want to know, if it provides more context, how would I answer an essential question relating to amp capture accuracy; how would I measure or compare transient response in a frequency response plot?
I'd assert that if it can't answer that basic a question, then the context it provides is just as useless as giving a flat LUFS value.
Frequency response, and the differences between two signals, when taken alone is simply not enough to make an accurate judge of whether two amplifier systems are equivalent or not. I don't know what is hard to understand about that?
Chocol8
Rock Star
- Messages
- 3,451
I would say it is slightly better, but it's still only measuring on piece of a much more complex puzzle. For example, when players with a lot of tube experience complain that digital doesn't "feel" the same, it's not the frequency response they are reacting to. It is the transient response and how what it is played now is impacted by what you just played.
To do truly helpful measure of how a capture is performing is going to require a variety of tests and measurements.
paisleywookiee
Goatlord
- Messages
- 10,084
Oh man, I'm TOTALLY going to trust what this guy says!
Lysander
Rock Star
- Messages
- 3,563
But, so does frequency response? If you ever heard anyone describing captures as "fizzy", "bloated" or "muddy", well, this is it.
I don't know why some got offended as if i presented these as a benchmark golden standard or anything. But still, a frequency content analysis is vastly more useful than blindly ranking profilers by LUFS on some test signal - simply because it highlights, and contextualizes, perceived differences between profiling devices.
Similar threads
