DCAudioDIY.com

More on the pink vs. white noise issue, after a few cups of coffee....actually several gallons since that last post.

This approaches a topic of great interest to me and like many things is not that straightforward: Is flat what we want?

The idea of using pink noise is that constant power per octave is "how our ears hear," power per octave linear on a logarhythic scale. Since power density/oct declines with frequency, 1/f, pink noise exhibits a 3db/octave dropoff. In other words, the octave gets a lot wider as frequency increases, so the spectral power density is lessened.

Well, that's cool and what is interesting in this to me is how such a dropoff approximates the natural falloff of HF in a sizeable hall. Bruel and Kjaer and Roy Allison among others have published landmark papers on this topic.

I noticed that the response of a WE728B 12" with no tweeter sounded very close to what I was hearing at Kennedy Center concerts 15 minutes earlier. Then I took my RTA to a few shows and saw graphically what I was hearing. Serious HF rollage. Very little 10k.

This raises the issue of accounting for differences between hall sound and living room sound, right? With less distance and dispersion, the natural rolloff is far less. To get hall sound in you r home, you would need some wicked rolloff.

OK, lets consider that pink noise is rolled at 3db/octave or 10db/decade.

If you EQ for flat response using an RTA and C-weighted pink noise, per the convention, at the listening position, you are beefing up the highs from the speaker.

If you use white noise with a constant power vs. unit frequency distribution, there are more highs in the test signal so the required HF EQ is less drastic, less highs out of the speaker.

It seems to me that what is required is to use white noise and EQ for the desired rolloff in order to achieve a natural hall EQ.

White noise is what is typically used for spec analyzers, whether old fashioned roll chart type or modern FFT jobs, both of which look at a swept frequency one at a time, so to speak.

Old school RTAs like the 15 pound huge antique B&K that I used in the 80s had a bank of LC bandpass filters. I don't know how something like a phone app RTA actually samples, probably discrete frequency sweeps then some kind of mathematical integration to approximate an octave bandpass function. That B&K had an accompanying oscillator box that output C weighted pink noise.

I suppose we have to trust the RTA app writers to have worked this out.

However, when putting these tools to use, it is helpful to understand the assumptions and structure of the test relative to goals. RTA is supposed to approximate the ear listening in real space as well as "real time." White noise is a constant "instrumentation" test signal and is basically equivalent to an omni-frequency "sweep"

Neither approach takes into account the effect of the room on EQ. This is something that we have to do ourselves.


The combination of close micing in the recording process as opposed to ambient stereo mic techniques and flat equalization, moreso with RTA techniques, and super-extended tweeters creates a situation with way more high frequency than live music, the reproduction of which is supposedly the goal. Or is it? The standard approaches seem like a recipe for in-your-face sound.

So I actually do use my RTA with white noise input and all of my stuff will likely have rolloffs before 10k. I use the RTA to see where it starts and to ensure that my HF and LF are at the same level at crossover. I left my Clio rig in Korea and I don't bother hooking up a mic and REW or whatever. I use my phone under the conditions described.

Still RTA is a standard measurement and if you want to do it by the book, pink noise is employed....but it is useful to understand what is going on with this test.

Yet another test that may be "controlled" but is crucially different from messy reality in important ways, just as all tests are. That why they are called "tests."

Anyway, here's a sweep of a 32 and 802 combo that I found somewhere online. Forget where, maybe Earl Geddes or Zilch. This EQ looks good to me for a natural "hall" sound, if one can deprogram the modern hifi listening habit and get into it for what it is.

I suspect this is exactly the sort of response that Bell Labs scientists were looking for when they developed the WE32A. Looks a lot like the response of a WE 728B, in fact. and a lot like what you will see if you tote your RTA down to the Kennedy Center

I have always found that live music (no sound reenforcement) does have less HF energy than most commercial audio gear. So the question to me is why do manufacturers do this?
Where did this over emphasized tizzzies come from?

To accurately (Hopefully) reproduce the material on the disc/file/record/tape/whatever. If the recording is good and mastered well the playback system should be faithful to the original.

_________________
I have too much stuff - https://www.pleasebuymystuff.com

The fact that very few people are talking about this gives you the answer, in part.

Hifi sound has always been "it's own thing" while practitioners go on about how their goal is to reproduce the exact experience of live music. Blah, blah, Blah...

Yeah, it sounds good on paper. So does 20-20 response.

Although a problem for all of my audio career, back to the 80s, it is even worse than ever now, probably due to the proliferation of cheap dome and ribbon tweeters that can play bat songs.

However, the core of the issue is cultural...tizz has been absorbed into the audiophile aesthetic as a desirable thing, and then marketing does its thing.

The sad fact is that if you are demoing two speakers, cables, whatever for customers, the one that sounds brighter and more "detailed" will win most of the time. That's one reason to be wary of A-B tests that set up a testing situation removed from any external references. More bass or more highs is easily confused with better bass and highs. More highs will increase the apparent detail until you have 10x more detail than music itself, thanks also to close mic techniques for that. And the spatial illusion created by microphone artifacts will be enhanced by a nice HF boost.

I deprogrammed myself by listening to older gear and systems. To me, a high grade system with what would be considered "subdued" treble can sound more natural and realistic to my ear.

Bell labs decided through human experimentation in the 30s that 13k is the required HF extension to properly reproduce orchestral music, gets the first few harmonics of most instruments.

I think that greater extension is OK but it is useful, in the quest for a natural perspective, to have a gradual falling response from 5-8k to wherever the tweeter runs out. Try it! You will find that the music is not in your face as much and maybe not as exciting, but you will be forced to listen more actively, like you do at a live performance, and maybe a little more deeply to pick out the details that modern systems spit out at you. I find that I have to "listen harder" at a live show than I do to a hifi rig. This changes the whole dynamic of music listening.

And even if I listen real hard, with great intent, I'm just not hearing the highs and detail that goes with the territory in high-end audio, which I like to call that because there is too much high end!

And then if a fool like me makes this kind of argument, scientists will come out of the woodwork to protest.

Well, it's like this: The spectrum analyzer and music listening ear are from two different worlds, with different goals and attitudes.

You can't judge one game using the rules of another.



An then you will hear unassailable statements like the above. How do we know if the playback system is faithful to the original when the "original" has no sound without a playback system?

And who says the "original" really mirrors the actual recorded event, which these days usually isn't even an actual historical event.

Although way more slippery and hard to define, I'm looking for a system that makes me feel like I am listening to real music, which I do a lot.

The best solution to this deep philosophical/technical dilemma, is to just listen to what you like. This is still legal, for now. Might already be illegal in the EU...

_________________
Walt

I agree with your assessment of RTA vs ear as I mentioned above. However, your statement about ABX testing is misinformed. ABX testing is not about which sounds more accurate or more pleasing. It SHOULD solely used to determine if there is a difference. If it is determined by double blind testing that a difference exists you can then go into the search as to which sounds "more accurate", "better", etc. You don't need blind testing for that and very ofter which sounds better will depend on your mood at the moment, the music used, the equipment playing it back, your preferences, etc. If a difference cannot be determined it only means for that test with those parameters and listeners that there was not an audible difference. It does not rule out that there can never be a difference.

The value of ABX testing is that a positive result gives you confidence to go the next step. Even with a negative result, if further testing is done changing the listeners, music, rooms, etc. and no difference is found, it still does not guarantee that under some situation a difference could be detected. That said, if after many different tests you still cannot get a positive result it should cause you to evaluate the magnitude of any difference that might occur and what amount of effort should be spent in that direction when so many gross differences do exist in other areas.

_________________
Walt

The question of frequency response shaping is one thing.

The quality of the produced HF signal is another.

I seriously doubt that most extended tweeters out a signal that has much to do with the input. Various dome tweeters put out 20k but how well correlated is it? It often sounds like ringing, out of time with the music, and probably is.

I think stiff and light titanium and beryllium diaphragms play into this trick because they can resonate at higher frequencies, doing an even better job of LARPing as wide bandwidth reproducers when they are actually resonating producers working on a skewed time base.

Even if careful engineering can yield tweeters that reproduce inputs reasonably faithfully with steady state test signals, how do we know they are so well behaved with complex transient musical signals. I don't think we do and to me it often doesn't sound like it.

Flat or rising HF tailoring to emphasize such time smears can't end well.

I grew to become obnoxious and contrarian on this topic after many years listening to the best pro stuff of the 30s and 40s. If these setups did 15k that was really extended, even for the top studio equipment. Little of it measures Stereophile flat. Some of the most satisfying things I have heard are from this bygone universe.

Take Altec, they didn't even have what we'd consider a tweeter until they designed the 3000A for hifi. For the pro stuff, a 288 or a 802 driver WAS the tweeter. Western Electric had the 713C compression driver in the late 40s, which measured to 20k, on the bench maybe, but it didn't reach beyond 15k in any of the horns made for it. Most of the "good stuff" was flat to 8-10k then rolled off smoothly....pretty much what an equalized 802/511 will do on axis at listening distances.

I see it as a two-part problem. Quality of reproduction, call it fidelity to the input signal, and then there is the question of proper system equalization for realistic, natural in room balance. That is to say, fidelity to the presentation of musical sound in actual real life contexts.
Fidelity to how real music makes me feel.

So, I am of the mind that a bad tweeter is worse than no tweeter, and even a good tweeter must be applied with care .

Food for thought, y'all.

As DIYers, we can put ourselves anywhere on the continuum that we want to be. If you're shopping Sterophile ads, you get what they are selling you.

"So, I am of the mind that a bad tweeter is worse than no tweeter, and even a good tweeter must be applied with care ."

Amen to that Joe. Many tweeters that go out to very high frequencies suffer from at least two problems. The first is that the materials of the diaphragms (cones, domes, etc) have poor characteristics to deal with high frequency reproduction. The second is there is often an impedance mismatch at the terminations of the diaphragms that leads to energy being reflected back into the diaphragm causing resonances. That generally accounts for the resonances you see in titanium and aluminum cone tweeters. A similar thing happens with paper and cloth although both of those materials tend to damp the resonances to a greater degree.

I'm including two references that may be of interest. Also if you look at the response of the tweeter I'm using in the "eggs" you will see that there are no resonances out to 40K. It does sound very smooth although if it had a problem at 40K I couldn't hear it anyway.

Tom I am really sorry I did not get to hear the eggs for more than a few miniutes since I was so busy with the show, but what I heard for that few miniutes I liked very much.
The sizzle,tizzzies were live and well at CAF and was very disappointing. I thought there was much better overall sound at the show in July2016.
Even the rebuilt Quads sounded bad to me, and I know the HF is limited on those to less than 15k
My top end hearing is getting less as I get older, l know in a hearing test I'd be lucky to have hear pass 10k but I sure can hear the glassy sizzle, tizzies.