I wasn't talking about resonances, just group delay by itself. I mean that it is the only time domain value we can't easily see by using our eyes and the naked FR (audibility matters put aside).
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
NHT Pro M-00 Powered Monitor Review
- Thread starter amirm
- Start date
- Joined
- Jan 14, 2020
- Messages
- 30
- Likes
- 20
I really would like to see phaseplot and impulse response, because they show quite well, how well drivers are time alligned, and other neat thingys.
And seeing harmonic distortions would not hurt either.
Oh and thank you, for your amazing job!
And seeing harmonic distortions would not hurt either.
Oh and thank you, for your amazing job!
- Joined
- Nov 22, 2018
- Messages
- 1,137
- Likes
- 1,509
Maybe true at JBL/Vanatoo levels, but it's still quite easy if you're willing to spend a lot. At least, that's how I read Stereophile's measurements of expensive speakers...
Ignoring phase makes sense to me. A reflection's phase will change if the speaker-boundary distance varies. The only way to standardize would be to use given room dimensions AND fixed speaker positions.
And diffraction-type issues.
- Joined
- Nov 22, 2018
- Messages
- 1,137
- Likes
- 1,509
What does Klippel recommend here?
Yep, I spent a long time trying to make the ER curve work with phase and it always just dipped far too much in the treble so that just about confirmed it.
OP
- Thread Starter
- #127
I have not found anything outside of driver tests. There, they have sophisticated laser scanners and such that fully model the driver. But not the whole speaker. I am trying to keep my chips for critical things in contacting them. Don't want to wear out my welcome too quickly.
And why would you like to see group delay measurement? It is not part of the CEA-2034 because it has not been related to audiblity issues with listening tests. Do you know something those scientists are not aware of?
restorer-john
Grand Contributor
So independent "scientists" proposed and established the CEA-2034 guidelines did they? Seriously, I'm asking.
It was an industry body, with vested interests, the Consumer Electronics Association, now the CTA?
The point here was it has been well established through multiple listening tests by Toole, Olive and others that time domain parameters do not influence how we perceive music thus asking Amir to measure GD, which is not part of standard Klippel measurement as it is based on CEA-2034, is pretty much meaningless.
restorer-john
Grand Contributor
Right. OK.
So when people proposed multi-tone IMD measurements vs single or dual tone measurements, some people said it was meaningless.
Who cares what Klippel put in their standard measurements? It is but ONE (expensive and awesome) device and does not measure every parameter important to loudspeaker characteristics. Perhaps when Amir combines the AP, the Klippel and a bunch of standalone tests, we might be able to get a better understanding of how a loudspeaker will perform in a typical usage scenario. Right now, it is a work in progress.
It's early days and there's lots to learn.
Right. OK.
So when people proposed multi-tone IMD measurements vs single or dual tone measurements, some people said it was meaningless.
Who cares what Klippel put in their standard measurements? It is but ONE (expensive and awesome) device and does not measure every parameter important to loudspeaker characteristics. Perhaps when Amir combines the AP, the Klippel and a bunch of standalone tests, we might be able to get a better understanding of how a loudspeaker will perform in a typical usage scenario. Right now, it is a work in progress.
It's early days and there's lots to learn.
There is a huge ammount of AES papers claiming that phase/GD don't play a role with SQ and not a single one that confirms it does. IMHO that makes measuring GD as usefull as measuring how thick the paint on the front panel of the speaker is.
restorer-john
Grand Contributor
Would you buy a D/A converter today with significant group delay or inter-channel phase differences, particularly errors that change with frequency? We all did that with CD players in the early 80s, ones that used a time shared D/A converter. Should we have known better? Perhaps.
So, how much more useful is measuring THD % to three decimal points before zero, or four? I look at 0.00042% and think, am I kidding myself? Yes, I am.
Not really a good analogy - if speaker drivers have GD related issues that would be an audible problem, but it has been proven by listening tests they don't. So why measure it?
From the audibility perspective it is absolutely a waste of time. But it does come handy if you want to evaluate engineering quality of the device.
I doubt it's that definitive if there are still papers coming about it. But I said audibility matters aside because I didn't read those papers, since I'm not an AES member; which is also why I asked if anybody read the recent Genelec paper about it (https://secure.aes.org/forum/pubs/conventions/?elib=19404) and could give a summary.
Especially since the abstract says: "Our results suggest that when the group delay in the frequency range from 300 Hz to 1 kHz is below 1.0 ms, it is inaudible.", which is not reached by any speaker I know of (for example, the 8351B still has 3ms of group delay at 300Hz).
Ok, here is GD graph of Performa3 F208, taken from 3m, gated at 20ms with FDW of 15 cycles. Tell me, what do you make of it?
That measuring group delay in non anechoic conditions is useless. What are you even trying to prove, anyway?
Anyway, that Genelec paper gave an interesting conclusion and I don't see how you can say "problem solved, group delay never matters" that fast.
It's a gated measurement, so "quasi anechoic". If you want me to apply different gating just say which one.
That you were asking for a GD measurement but you have no idea what to make of it.
Btw, you are listening in the room, not in anechoic chamber and you don't seem to realise that reflections will spoil any nice looking GD graph made in anechoic chamber as they spoil the FR.
Sure, if you carefully EQ phase at your LP you will get nicer curve, but here is how uncorrected GD of Revel Performa3 F208 looks in a room.
So, once again, tell me what do you make of it?
Last edited:
Doesn't look like any anechoic GD graph I've ever seen, which always look like f(x)=1/x. Where did I ask for it? I only said "So only group delay could matter in this case" ("this case" being CSD not being that useful) with emphasis on could. You still didn't answer the fact that you boldly said that any quantity of group delay is completely inaudible while I linked to a paper pretending the opposite.
Mate, do you have an axe to grind, or what? I'm just wondering if some measurements could help make those reports more exhaustive, nothing else.
Similar threads
