Two unresolved issues

[Ethan Winer]

Guys, let's take the emotions down a notch . It is just a conversation and should not raise anyone's angst.

No kidding. Why do people feel they need to be insulting?

--Ethan

 

[Ethan Winer]

I made no comment of the efficacy or lack thereof of null testing, so your question in message #171 - "And why do you think a null test is not proof?" - is yet another example of you putting words in my mouth.

So do you agree that a null test is conclusive or don't you?

So if you don't use something like an Audio Precision to measure amplifiers and digital products, how can you be so sure that your measurements and null tests are good enough to characterize the performance of the devices under test? I'll ask again, Ethan, as perhaps you missed this question. What A/D converters do you use, what is their resolution/noisefloor performance, and what calibration have you performed of that hardware?

I thought you already knew my equipment list. I rarely need to record anything to do the sort of tests I do. If you haven't watched my AES Audio Myths I guess it makes sense that you don't know much about what I do and how I do it.

When I do need to record I use an M-Audio Delta 66 sound card. Yes it's old, but it's clean enough for all but the most critical testing, as proven by the graphs below. Certainly this card has noise plus distortion below what's audible, and the frequency response is flat enough too. The FFT screen caps below show the THD and IMD and spectral noise of an original Wave file, a copy played out and back in through my Delta 66 sound card to include both the A/D and D/A, and again out and in through my $25 SoundBlaster X-Fi sound card. I generated the source file in Sound Forge, so I imagine it's as clean as a digital source can be. I created four frequencies in a row, each five seconds long, so the source file and the recorded copies are all 16 seconds long:

20 Hz
1 KHz
10 KHz
19 + 20 KHz

The record level meters said -1 for both sound cards for all frequencies. The FFT settings are as follows:

FFT Size 65,536
FFT Overlap 99%
Smoothing Window Hamming

--Ethan

PS to Amir: I have listed my equipment dozens and dozens of times in forum posts over the years. I have no reason to hide what gear I use, and as you should know by now I'm not ashamed of anything I use. Not even my $150 Pioneer receiver I mention all the time.

 

[Ethan Winer]

Ethan, I hope you take John's point to heart and that of others here who really want to achieve the highest performance.

I never believed (or said) otherwise. I understand that some people want the best, and that's great if they can afford it. Even if "best" sounds not one whit better than "good enough" with distortion at "only" -80 dB = 0.01 percent. But I started this thread in an attempt to get Mike Lavigne to explain what specific physical properties digital recording cannot capture. And we all know how far that got. In hindsight, I should have kept separate the Meyer & Moran stuff, and started another thread for that to avoid exactly the sort of thread drift we see here now.

BTW, even if one wants and can afford the very best, $6,000 speaker wires and too-small bass traps and vinyl demagnetizers ad nauseum are still a total waste of money.

--Ethan

 

[Ethan Winer]

Ethan, I think the historical record on increasing jitter awareness in transports is a matter of public record. There are posted Harley articles on the Stereophile website.

I don't know what that means, but in modern digital gear - and even ancient digital gear - jitter is not an audible problem. The only possible exception I'm aware of is when Amir pointed out to me that HDMI audio can have jitter artifacts approaching -60 dB, which borders on being audible. I realize that jitter has been demonized for many years now, but the truth is that jitter is benign even in amounts much larger than actually occurs in normally working gear.

This article explains the artifact levels typical for jitter, and lets you hear other artifacts at various levels to better relate that to what is actually audibile:

Artifact Audibility Report

As for Nyquist, I think theory is fine for what it is but in practice there are issues in implementation. I've been able to play with all kinds of hirez and 16/44 recordings and there is not a small difference between the two although the gap has narrowed somewhat as CD playback has matured.

Nyquist is not in dispute, though as I said not all A/D/A implementations are equal. If using a sample rate higher than 44.1 KHz really does change the sound, and that's verified by a proper blind test, the reason will not be due to the higher sample rate per se. It might be filter linearity or ringing etc or something similar. But not because frequencies higher than anyone can hear are present.

--Ethan

 

[Mike Lavigne]

BTW, even if one wants and can afford the very best, $6,000 speaker wires and too-small bass traps and vinyl demagnetizers ad nauseum are still a total waste of money.

--Ethan

why did you need to add this little bon-bon to your post?

what does the cost of something have to do with anything?

'too-small bass traps? huh?

i'm the one who said i make no claims about demag'ing. others bought it up.

after you get picked on is this just a poke in the dark to deflect the flak?

 

[Ethan Winer]

I'll try again Bruce, but if your next reply is insulting you go on permanent ignore.

1. I used the same exact program you claim to use ... Sound Forge. These graphs are from spectral analysis/FFT plugin. Guess you either need to RTFM or stop claiming you use Sound Forge when you obviously don't know how to read the graphs or know what I'm talking about. Spectral analysis can give you lots more information than a normal FFT graph. The freq is on the left, the timeline goes across and the amplitude is in colors!!!

So what is the exact amplitude of this color? I enlarged it from one of your FFTs for clarity.



How many dB down is 24 KHz? How about 30 KHz? Are you really arguing that energy at 30 KHz even makes a difference? This is the problem with the type of FFT you showed. It obscures the data that actually matters - absolute level versus absolute frequency, in exchange for showing the change in spectrum over time. A more usual FFT showing an average of the same five-seconds from each file would have been much more useful. Then for example, if the "poor" SACD players M&M used actually had a deficiency at frequencies that are audible, I'd agree you have a point. Though you still skirted this point I have made several times now: Even if 2/3 of the tests used "poor" players, that still leaves a huge number of tests where nobody heard a difference. And you still haven't listed the player models you used so I can compare with MM's list to verify they are exactly the same model. Nor did you state what SACD music source you see (unless I missed that).

Also, I must have missed this from a few days ago:

It is proof if you can null a digital signal against an analog signal. Doing a null against 2 digital signals proves nothing. It's already been digitized.

There's no reason that null tests must use two digital audio files in a DAW. The earliest Hewlett-Packard distortion analyzers were based on nulling. You can even null the input and output of a power amp to see its distortion. If the amp is inverting all you need is a couple of resistors and a trim pot. If it's non-inverting you also need a small inverting preamplifier of high quality.

--Ethan

 

[KeithR]

couldn't make it past one page on this thread---Mods, can it be moved to Ethan's forum?

i really didn't think this forum would be turned into this.

 

[Ethan Winer]

why did you need to add this little bon-bon to your post?

Because it's the truth.

what does the cost of something have to do with anything?

That's easy: the higher the cost of an ineffectual product, the worse of a scam it is and thus the more it needs to be exposed. I'll point out that using foam coffee cups as speaker cable elevators is a silly waste of time, but it doesn't get my noise out of joint when people do that. Heck, maybe they think it looks cool. But $300 for a set of cable elevators that don't improve the sound even a little?

'too-small bass traps? huh?

I measured these and they did nothing useful:

Cathedral Panels

I haven't had the pleasure of measuring these, but they do nothing useful either:

Acoustic ART

after you get picked on is this just a poke in the dark to deflect the flak?

LOL, I think you're missing something very important.

--Ethan

 

[Stereoeditor]

When I do need to record I use an M-Audio Delta 66 sound card. Yes it's old, but it's clean enough for all but the most critical testing, as proven by the graphs below.

With all due respect, Ethan, your graphs don't prove anything, other than you are not familiar with the limitations of the tools you are using. The noisefloors of your spectra are dominated by the effects of the FFT windowing, and I am surprised you don't realize that.

John Atkinson
Editor, Stereophile

 

[Ethan Winer]

With all due respect, Ethan, your graphs don't prove anything. The noisefloors of your spectra are dominated by the effects of the FFT windowing, and I am surprised you don't realize that.

I also listen John, often at loud levels when needed. My ears confirm that noise and other artifacts from the Delta sound card are enough below the music to not matter. I'm (not really) surprised you don't realize that.

--Ethan

 

[RBFC]

We need to keep the discussion civil. We will be forced to close this thread (and any others that spring up to replace it) if jabs continue to be thrown. State the FACTS that support your arguments and don't worry about the individuals who are posting. Be prepared to back up facts with peer-group acceptable proofs. There have been far too many unsupported claims in some of the posts. There has also been a lack of substantial proof for assertions made in many cases. Come on guys, we can do better!

Lee

 

[Ron Party]

Time for a pause until cooler heads prevail. Thread will be reopened if members wish to remain on topic and take the high road, i.e., discuss the merits, not the person.