Blind ABX test shows difference between 44.1 and 88.2

[davidsrsb]

actually I would have expected that any speaker would have peaks and troughs of that level in room at your ears because of comb filtering. They will have peaks and troughs like that even below 20Khz.
Anyway I thought my speakers were supposed to be rolling off not peaking.

Room modes and comb filtering happen at much lower frequencies and are reasonably linear with volume. Domes can get into horrible break up modes, where the the sections of the dome move in a way that beams on axis and almost nothing off axis - the efficiency measured in total has not magically jumped 30 dB. This is a very non-linear effect, modulated by all the audible treble motion the dome is making

A similar phenomenon happens to the mid-bass too, and the preference of simple or non existent crossover low pass vs high order is not settled either

 

[merlin]

Thank you David.

 

[adamdea]

Room modes and comb filtering happen at much lower frequencies and are reasonably linear with volume. Domes can get into horrible break up modes, where the the sections of the dome move in a way that beams on axis and almost nothing off axis - the efficiency measured in total has not magically jumped 30 dB. This is a very non-linear effect, modulated by all the audible treble motion the dome is making

A similar phenomenon happens to the mid-bass too, and the preference of simple or non existent crossover low pass vs high order is not settled either

Does comb filtering stop at/below 20Khz? I seem to rember that the measurements I did for DRC looked horrific over 10Khz, but then they did on my proacs too. I thought that this was the result of comb filtering or at least some form of interference.

 

[davidsrsb]

20 kHz has a wavelength of 15mm, so any discontinuity of this sort of dimension near the tweeter OR microphone can cause ripples n the frequency response.
I have never seen a measurement microphone without its own problems above 20 kHz.

Things like abrupt edges on tweeter horns, fixing screws with raised heads, badly designed grilles, cavity reflections and many other things make getting a smooth response hard - felt strips near the tweeter is a favourite
The saving point is that these ripples are fairly linear and don't cause intermodulation

 

[adamdea]

20 kHz has a wavelength of 15mm, so any discontinuity of this sort of dimension near the tweeter OR microphone can cause ripples n the frequency response.
I have never seen a measurement microphone without its own problems above 20 kHz.

Things like abrupt edges on tweeter horns, fixing screws with raised heads, badly designed grilles, cavity reflections and many other things make getting a smooth response hard - felt strips near the tweeter is a favourite
The saving point is that these ripples are fairly linear and don't cause intermodulation

Yes, well given the inevitable discontinuities near the ear, I can't really see how a flat frequency response at the listening position could be possible. As a matter of interest are you aware of any speakers which have a flat anechoic response on axis out to 96 kHz ? Assuming that is that there are any mics capable of reliably measuring this.

 

[davidsrsb]

I think only piezo technology can go much above 20 kHz. I always regret Audax never solved the deflating HD-3P. That was a far better unit than any moving coil dome

 

[merlin]

There are now numerous tweeters and supertweeters on the market with clean extension out to beyond 30khz which is a good thing. With more materials research comes greater extension.

Sony introduced a supertweeter for the SS9 fifteen years ago, clean and flat to 50khz - although in the field measurements of that were understandably difficult

 

[F1eng]

I have a pair of NXT super tweeters which a friend made. They extend to 50kHz. My friend thinks they are a big improvement, I can not tell whether they are on or off.
Modal panels like this (and the BMR) have the huge benefit (or disadvantage depending on POV!) of not being very directional.
Most, if not all, conventional tweeters get increasingly directional with frequency such that it is quite likely that unless you sit on the tweeter axis you have never been exposed to the highest frequencies, whether you can hear them or not.

 

[merlin]

What's the HF unit on the Animas Frank?

 

[Julf]

Unfortunately most studio mics don't go much beyond 20 kHz, and neither did tape back in the days of magnetic tape in the recording studios...

 

[adamdea]

There are now numerous tweeters and supertweeters on the market with clean extension out to beyond 30khz which is a good thing. With more materials research comes greater extension.

Sony introduced a supertweeter for the SS9 fifteen years ago, clean and flat to 50khz - although in the field measurements of that were understandably difficult

Well if they are only flat to 50kHz then would they meet your requirements to test the difference between 24/96 and 24/192 or between either of them and dsd?

In the infamous 24/176 vs dsd test
http://old.hfm-detmold.de/eti/projekte/diplomarbeiten/dsdvspcm/aes_paper_6086.pdf
They used speakers which were
"distinguished by their very precise impulse
response and frequency response up to 35 kHz"

Could this have been adequate to test the effect of the anti alias filter in the 24/176 recordings? Of course they also used headphones in some cases, but we know from Oohashi's later work that headphones do not seem to stimulate the mysterious ultrasonic brain effect.

 

[adamdea]

I have a pair of NXT super tweeters which a friend made. They extend to 50kHz. My friend thinks they are a big improvement, I can not tell whether they are on or off.
Modal panels like this (and the BMR) have the huge benefit (or disadvantage depending on POV!) of not being very directional.
Most, if not all, conventional tweeters get increasingly directional with frequency such that it is quite likely that unless you sit on the tweeter axis you have never been exposed to the highest frequencies, whether you can hear them or not.

Surely all right thinking forum members sit exactly on axis, Frank?

 

[merlin]

The point here is that if you overlay the read outs from both a Red Book and High Resolution file, you will see that if there are differences, these occur between typically 18khz and 22khz.

If your loudspeaker is -3db at 20Khz you are substantially reducing any possibility of identifying differences.

In order to hear potential benefits with high resolution you need to be able to reproduce the area where differences might exist.

 

[merlin]

Well if they are only flat to 50kHz then would they meet your requirements to test the difference between 24/96 and 24/192 or between either of them and dsd?

I'm not with you. Are you suggesting someone is claiming frequencies between 48khz and 96khz have any impact whatsoever on replay?

 

[adamdea]

The point here is that if you overlay the read outs from both a Red Book and High Resolution file, you will see that if there are differences, these occur between typically 18khz and 22khz.

If your loudspeaker is -3db at 20Khz you are substantially reducing any possibility of identifying differences.

In order to hear potential benefits with high resolution you need to be able to reproduce the area where differences might exist.

The differences will be in the stop band of the anti alias filter and to some extent in the transition band. If the AA filter is done correctly there should be effectively no difference below 20Khz.

I am not sure what speakers you are referring to. You linked to this measurement from Stereophile, http://www.stereophile.com/images/archivesart/805fig3.jpg
The accompanying text indicates that it shows a flat respopnse to 20Khz and a resonant mode at 26Khz which might be a problem with hi rez. I would have thought that this would be quite good for telling the difference between CD and hi rez if one could hear it as the 26Khz progam material should stand out.

I'm not altogether sure why you consider that what matters with hi rez is the difference in the transition band (18-22 khz) , which is above my hearing, rather than the differences above 22Khz which are above pretty much everyone's.

I'm not convinced that the 3dB difference at 20kHz would be all that likely to make all that much difference, especially bearing in mind the likely in-room response, although i can see that it would not be ideal. The only differences at that frequency are probably going to be at a very low level and they will be at frequencies where human hearing has a very low sensitivity (if any) and no pitch discrimination. If we are to be testing the effects of different filters perhaps we should wonder about the phase too.

Taking the same reasoning used in relation to 16/44 -where does the difference between 24/96 and DSD 128 lie? If it's in the transition band of the PCM filter then what speaker do we need?

Either way I think it's quite difficult to say what spec of speaker is required to hear the benefits of hi rez without being clear about exactly what those benefits are, and also without excluding pretty much all kit.

In any event, the benefits of hi rez and the inherent problems of 16/44 must logically be two sides of the same coin. So if anyone can't hear the benefits of hi rez, it follows ineluctably that they can't hear what's wrong with 16/44.

 

[F1eng]

What's the HF unit on the Animas Frank?

Not sure. A Fostex compression driver, I think.
OTOH I tried the super tweeters with my Goldmund Epilogs (and probably had them running for almost 10 years). I couldn't tell wether that were on or off but I reasoned that if my mate could so could some guests...
I disconnected them when I got the Devialet D-Premier 3 years ago since it did not have spare preamp outputs (my Goldmund Mimesis 22 did)
The Epilogs have Scanspeak Revelator tweeters.

 

[F1eng]

Surely all right thinking forum members sit exactly on axis, Frank?

Joking apart I have pretty well never pointed box speakers directly at me, but the adjustable treble horns on my Animas do seem to need to point directly at me for the greatest clarity.
I love them but you need a big room and sit well back for good integration, I am about 5 m away, and the sweet spot ain't that big.

 

[merlin]

Thanks Frank.

The Fostex are typically flat to over 35khz AFAIK although off axis they do drop away really quickly above around 10khz as you say.

Given the distance you have to the speakers though I would imagine getting the tweeter roughly on axis at the listening position is not too difficult?