“Gaming” measurements

[idiot_savant]

@March Audio - The beginning of the thread, there was agreement that most people don’t understand/ trust measurements. 
 

As for speculation, I have shown you a datasheet that word for word matches my thought experiment. As a representative of an audio company, who makes DACs with sabres in them, maybe you could demonstrate the performance benefits of DRE in the real world?

 

your friendly neighbourhood idiot 

[idiot_savant]

And back to the objective thing for the QB9, I’m afraid this kind of thing crosses though boundaries - like I say, they didn’t deliberately add “euphoric” distortion, they must have felt that the cost in removing the alias outweighed the benefit ( ringing, phase distortion ).

 
I’m all for giving users the ability to choose when there may not be a “100% correct” approach. 
 

I’d have no problem with DRE if, for example a light came on so measurements could have an asterisk, or just be able to turn it off somehow

 

your friendly neighbourhood idiot 

[March Audio]

18 minutes ago, idiot_savant said:

@March Audio - The beginning of the thread, there was agreement that most people don’t understand/ trust measurements. 
 

As for speculation, I have shown you a datasheet that word for word matches my thought experiment. As a representative of an audio company, who makes DACs with sabres in them, maybe you could demonstrate the performance benefits of DRE in the real world?

 

your friendly neighbourhood idiot 

The conversation really didn't go that way, but yes I pointed out that understanding the test parameters is key.  This is a little different to your earlier assertion that we agreed "cheating is widespread".

 

Considering its your thought experiment, don't you think it's more appropriate for you to demonstrate this "gaming" in action?

 

Demonstrate where some ESS snr measurements are incongruous or look excessive.

 

Only a small sample but I have 3 ess based dacs here we can test and look for this.

[idiot_savant]

Well,

 

I don’t have any suitable test equipment or a DAC that might do it. I can think of a few ways to illustrate it if I had such things - it’s more difficult than you might think as the test equipment gain ranges as well. 
 

And it’d be a pretty poor effort from the manufacturer if I could find a magazine measurement that showed it in action - since it’s meant to be a cheat, after all, and @The Computer Audiophile has made it clear these forums are not to be used to “catch out” manufacturers - so I can think of a DAC that looks like it might be doing this, but that would be speculation on my part and could cause problems for that manufacturer if I were wrong

 

your friendly neighbourhood idiot 

[idiot_savant]

On 5/27/2021 at 1:52 AM, The Computer Audiophile said:

I think many people have learned that measurements and specs can be used in nefarious ways, in every hobby, business, and endeavor. As such, they take the numbers as a single data point that doesn’t usually mean too much. 

 

[March Audio]

37 minutes ago, idiot_savant said:

*pedant alert*
The QB-9 doesn’t add distortion in “listen” mode - it doesn’t remove as many aliasing products as in “measure” - all of which are guaranteed to be above 22.05kHz. This trade off is done to minimise ringing, a bit like ….

 

 

MQA

 

 

 

your friendly neighbourhood idiot 

 

Ok, but that's not how it was described to me earlier.

 

Now where is that article on filter ringing....

 

 

[idiot_savant]

Well - maybe you should do some research before jumping in? 😉

 

your friendly neighbourhood idiot 

[March Audio]

16 minutes ago, idiot_savant said:

Well - maybe you should do some research before jumping in? 😉

 

your friendly neighbourhood idiot 

Sorry I made the mistake of trusting what I was told in this thread 😉

 

That the engineers wanted/implemented  some "euphonic distortion".  That implies in band and audible.  That's what I based my response on.

 

 

 

Here is the link.  Saves me running through it.

 

https://troll-audio.com/articles/filter-ringing/

 

[idiot_savant]

No worries - rule #1 - what marketing says is happening != what the engineers would say is happening over a beer

 

your friendly neighbourhood idiot 

[The Computer Audiophile]

28 minutes ago, idiot_savant said:

these forums are not to be used to “catch out” manufacturers - so I can think of a DAC that looks like it might be doing this, but that would be speculation on my part and could cause problems for that manufacturer if I were wrong

Thank you for being careful. It’s hard to unring a bell. 

[March Audio]

1 hour ago, idiot_savant said:

 and @The Computer Audiophile has made it clear these forums are not to be used to “catch out” manufacturers -  

Can we exclude MQA from that prohibition? 😉

[semente]

This is from the Ayre DX-5. Inaudible?

 

Ayre Acoustics DX-5, frequency response at –12dBFS into 100k ohms from balanced outputs with data sampled at 44.1 and 96kHz with Measure reconstruction filter (left channel cyan, right magenta), and with Listen filter and DSD data (left blue, right red). (1dB/vertical div.)

https://www.stereophile.com/content/ayre-acoustics-dx-5-universal-disc-player-measurements

[semente]

From the same model (BTW Ayre doesn't use feedback):

 

Ayre Acoustics DX-5, spectrum of 50Hz sinewave, DC–10kHz, at 0dBFS into 600 ohms (left channel blue, right red; linear frequency scale).

 

 

Ayre Acoustics DX-5, Listen, HF intermodulation spectrum, DC–24kHz, 19+20kHz at 0dBFS into 100k ohms (linear frequency scale).

 

 

 

Ayre Acoustics DX-5, Measure, HF intermodulation spectrum, DC–24kHz, 19+20kHz at 0dBFS into 100k ohms (linear frequency scale).

[Jud]

Regarding DAC filtering: The sharper the filter (and thus the less frequency-based distortion), the greater the Gibbs effect (ringing, time-based distortion). Thus a balance has to be struck either at maximum (though not perfect) sharpness and low frequency-based distortion, maximum (though not perfect) freedom from ringing, or somewhere in between.  Frequency-based distortions (IMD and HD) are usually considered much more audible, though opinions differ to some extent, and those are usually considered more important to measure.  However, many people seem to prefer the sound of filters that are less sharp, so fine - up to them. (By the way, even more extensive user filter selectability than Ayre was available in Resonessence DACs, built by the people who designed the ESS chip.)

 

With regard to whether people may or ought to like the sound of distortion, you're welcome never to listen to Hendrix feedback, Zappa wah-wah or Neil Young fuzztone, but I like my distortion just fine, thank you.

[idiot_savant]

@semente - thanks for the frequency response charts, my apologies - the “listen” filter appears to have a more NOS respond than I remember, but the option to choose is one I would defend. 
 

As for the distortion measurements, I believe the Ayre response to having too much feedback is to have none… I personally may not agree, but they’ve always said this is the trade-off they are making, not disagreed with the measurements so as I’ve always said, if that’s what you like, go for it!

*If* Ayre had claimed to have zero feedback *and* superlative distortion performance, that’s a different thing

 

your friendly neighbourhood idiot

 

[Ryan Berry]

The Listen/Measure switch was put in for pretty simple reasons (as most of the decision we make tend to be): we largely prefer the roll-off and the effect it has on the sound in the "Listen" position we ship all units by default in, but we needed another filter to help reviewers make sense of the numbers they were looking at.  Left to our own devices, we'd simply have used the listen position's MP filter and likely never put in a Measure switch at all.  However, the Measure position switch was put in as a result of the feedback we were getting from reviewers who include measurements in their reviews being frustrated that they couldn't really compare the Ayre to anything else as we were the only ones really using a Minimum Phase filter exclusively at the time.  I know variations of this filter are more popular now with many manufacturers having some sort of Minimum Phase filter available as at least an option, but this wasn't really the case in the 90's.  So ultimately, the option went in and we didn't want to confuse people into thinking we really preferred this filter, so we labeled them "Listen" and "Measure" to indicate that one is for...well, listening, while the other's purpose was specifically designed to help reviewers compare measurements of our DACs vs. the rest of the worlds without getting thrown off by what the filter was doing.  Outside of that filter, there's no DAC settings adjusted whatsoever between the two options.

 

As with all things, people of course play with the switches and try listening to both modes just to see if we were right.  For the majority, I believe this is the case.  In my experience, I'd estimate about 98% of the people I hear back from prefer our listen setting over the measure setting.  In any case, enough to feel that the MP filter was the right design decision for us, especially with how other manufacturers have followed suit since and added a version of the filter as an option.  On the MQA point...well, one thing I typically ask the people that think they hear an improvement in a MQA track is if they think they're hearing the stated improvements in the encoding of the file, or are they comparing on a DAC that normally isn't using a MP filter...suffice it to say that I have enough feedback there to have a pretty strong position on the matter.  That's an entirely different topic of conversation that I can't really go into the depth I'd like, so I'll leave it at that.

 

Largely agreed on the DRE settings and how it was designed to be used as a trick to improve numbers in a particular region years ago.  It's an example of how frustrating as a manufacturers comparing measurements can be.  There's a lot of ways that published specifications are taken that can make a product look like it measures a lot better than it really does from DACs to amplifiers -- which then makes it extremely hard for anyone to really make true comparisons by reading a spec sheet.  Of course, our experience is that numbers have a lot less to do with what we find we prefer in sound than I think any of us wished were true (excluding some pretty extreme cases we've seen over the years).   How easy would it be just to hit the ultimate low distortion numbers and do all of our listening tests on an AP?  Product development cycles would be cut down to a quarter of normal!  Ultimately, each manufacturer needs to make their own decision on what sounds the best to them and hope the market agrees.  I try to be pretty careful when I talk about these things, as I don't like black and white statements like, "this is better because..." and get into a debate about how that can't be true because of some other experience a person had testing this DAC back in 2008, etc.  Instead, I try to state that in our experience, we've found this to be true with our products in the tests we have done.  There's many people in the world and they have all their own preferences.  I may not always agree with what they prefer, but I always try to respect their feelings on the matter.  Interesting topic!

[idiot_savant]

Hi @Ryan Berry - May I be the first to thank you for posting on here? I’ve already said it, but it is genuinely refreshing for a manufacturer to say “we’ve taken decision x, it might measure worse y because z”. 
 

As I’m sure I’ve said, I’m all for giving people the choice! I’ve heard lots of people saying Ayre stuff sounds good, which, at the end of the day is what it’s all about?

 

I have more questions, but worth a thank-you

 

your friendly neighbourhood idiot 

[idiot_savant]

I’m not sure if this continues in the “gaming measurements” thread?

 

your friendly neighbourhood idiot 

[March Audio]

8 hours ago, Jud said:

Regarding DAC filtering: The sharper the filter (and thus the less frequency-based distortion), the greater the Gibbs effect (ringing, time-based distortion). Thus a balance has to be struck either at maximum (though not perfect) sharpness and low frequency-based distortion, maximum (though not perfect) freedom from ringing, or somewhere in between.  Frequency-based distortions (IMD and HD) are usually considered much more audible, though opinions differ to some extent, and those are usually considered more important to measure.  However, many people seem to prefer the sound of filters that are less sharp, so fine - up to them. (By the way, even more extensive user filter selectability than Ayre was available in Resonessence DACs, built by the people who designed the ESS chip.)

 

With regard to whether people may or ought to like the sound of distortion, you're welcome never to listen to Hendrix feedback, Zappa wah-wah or Neil Young fuzztone, but I like my distortion just fine, thank you.

The ringing only occurs with stimuli close to nyquist and is close to nyquist in frequency.

 

Here is a real measurement of a dirac pulse I made just now (from an ADI2 pro FSR)  with sharp filter  sample rate is 44kHz

 

 

Note the period between the ringing pulses.  46uS.  This shows the ringing is at 21.7kHz.

 

This will only appear when there is a music signal close to nyquist.  Music signals are very low in amplitude above 20kHz so any ringing is going to be correspondingly lower.

 

So even with 44kHz audio its above your range of hearing and at very low level.

 

Filter ringing has been a big red herring in audio over the past few years.

 

see here for more info.

 

Filter ringing – Troll Audio (troll-audio.com)

[Jud]

21 minutes ago, March Audio said:

The ringing only occurs with stimuli close to nyquist and is close to nyquist in frequency.

 

Here is a real measurement of a dirac pulse I made just now (from an ADI2 pro FSR)  with sharp filter  sample rate is 44kHz

 

 

Note the period between the ringing pulses.  46uS.  This shows the ringing is at 21.7kHz.

 

This will only appear when there is a music signal close to nyquist.  Music signals are very low in amplitude above 20kHz so any ringing is going to be correspondingly lower.

 

So even with 44kHz audio its above your range of hearing and at very low level.

 

Filter ringing has been a big red herring in audio over the past few years.

 

see here for more info.

 

Filter ringing – Troll Audio (troll-audio.com)

 

I don't think you're disagreeing with anything I've said here.  Mans' position represents the majority of what I've read on the subject. But as I noted, some well respected filter designers (including at least one who makes pro products used by award winning classical producers, and I'm certainly not talking about MQA here) have written that ringing is audibly harmful and ought to be avoided to the extent possible while balancing other factors like frequency-based distortion.

[opus101]

41 minutes ago, March Audio said:

Filter ringing has been a big red herring in audio over the past few years.

 

Certainly DAC filter ringing is without considering ADC anti-aliasing filter ringing too. Every music signal that goes to a DAC has passed through an ADC AA filter. Except perhaps for DSD.

[March Audio]

8 hours ago, Jud said:

 

With regard to whether people may or ought to like the sound of distortion, you're welcome never to listen to Hendrix feedback, Zappa wah-wah or Neil Young fuzztone, but I like my distortion just fine, thank you.

Aahh but that is the art that was created, subtle difference 😉

[March Audio]

14 minutes ago, Jud said:

 

I don't think you're disagreeing with anything I've said here.  Mans' position represents the majority of what I've read on the subject. But as I noted, some well respected filter designers (including at least one who makes pro products used by award winning classical producers, and I'm certainly not talking about MQA here) have written that ringing is audibly harmful and ought to be avoided to the extent possible while balancing other factors like frequency-based distortion.

For sure, I wasn't disagreeing just adding some further info.

 

These impulse plots rarely have any additional info with them and lose context.

 

My experience FWIW, is that the major difference in sound is created by slow filters that roll off in the audio band.

[March Audio]

6 minutes ago, opus101 said:

 

Certainly DAC filter ringing is without considering ADC anti-aliasing filter ringing too. Every music signal that goes to a DAC has passed through an ADC AA filter. Except perhaps for DSD.

This is true. Of the multitracks some may have been recorded through different ADCs with different filters.

[Jud]

4 minutes ago, March Audio said:

Aahh but that is the art that was created, subtle difference 😉

 

Sure.  But the larger point is that distortion isn't necessarily something that makes people recoil. Some folks quite like the sound of filters that leak somewhat (to the extent they can tell the difference, also a subject of back-and-forth discussion).  I'm pretty catholic in my tastes - I like the sound from the Pono, which uses a version of the Ayre filter, but with my usual software player I use quite a sharp filter.