Any correlation between measurements and perceived sound quality?

[j_j]

Those must have been nasty drivers. I've made many thousands of loudspeaker driver measures over hundreds of drivers and only one didnt have phase completely predictable by Hilbert transform, and that driver was damaged. Multi driver systems of course are a different entity but the original poster mentioned transducer.

I'm responding to "driver in box". Rarely do we use dipoles, yes?

 

[bennetng]

You might not believe the bizarre lengths they'll go to in explaining why your test prevented their golden hearing from having a chance.

Sounds like you'll like this test by Archimago: INTERNET BLIND TEST: Do digital audio players sound different? (Playing 16/44.1 music.)

Well I just read Miska talked about how "flawed" Archimago's listening test is designed by showing a screenshot up to 2.5MHz but only a couple of pixels below 20kHz.

But the funny thing is in Archimago's another article, higher DSD rates actually showed higher noise floor below 20kHz.
http://archimago.blogspot.com/2019/01/measurements-oppo-udp-205-dsd-playback.html

So it could be the Oppo performed poorer in higher DSD rates or the RME's ADC is adversely affected by the Oppo's output. In either case it is not a good thing.

 

[pkane]

Well I just read Miska talked about how "flawed" Archimago's listening test is designed by showing a screenshot up to 2.5MHz but only a couple of pixels below 20kHz.

Miska is very focused on wide-band measurements and how various filters behave well above the audible range. That's part of the appeal of his HQPlayer software (I use it, too ) It offers various filters with different stop-band rejection, slope, and ripple, as well as upsampling and conversion between PCM and DSD.

He's right that an ADC anti-alias filter can have a measurable effect in the audible range. I doubt whether it is audible in Archimago's test, but Miska thinks he can hear it. Even if it is audible, the effect is extremely minor.

 

[Blumlein 88]

Well I just read Miska talked about how "flawed" Archimago's listening test is designed by showing a screenshot up to 2.5MHz but only a couple of pixels below 20kHz.

But the funny thing is in Archimago's another article, higher DSD rates actually showed higher noise floor below 20kHz.
http://archimago.blogspot.com/2019/01/measurements-oppo-udp-205-dsd-playback.html

So it could be the Oppo performed poorer in higher DSD rates or the RME's ADC is adversely affected by the Oppo's output. In either case it is not a good thing.

Did you read the link to the post I had here on aliasing. Yes it is a real thing. I think Miska makes too much of it. I don't believe it would be audible. He thinks it is.

As Pkane said HQ player is superlative software. I've been meaning to try it again with DSD now that I have a DAC that does DSD. Maybe I'll redo the aliasing test using DSD to see what shows up. But as Miska posted you should be able to tell looking at chip datasheets, and I don't see why I'd expect anything audible.

 

[j_j]

Why is 2.5MHz interesting? One can simply look at tweeter response, air transmission loss, and the eardrum and ear canal impedance and ask that question very simply.

https://www.mne.psu.edu/lamancusa/me458/10_osp.pdf

http://www.sengpielaudio.com/calculator-air.htm

Sorry, physics works.

 

[bennetng]

Did you read the link to the post I had here on aliasing. Yes it is a real thing. I think Miska makes too much of it. I don't believe it would be audible. He thinks it is.

As Pkane said HQ player is superlative software. I've been meaning to try it again with DSD now that I have a DAC that does DSD. Maybe I'll redo the aliasing test using DSD to see what shows up. But as Miska posted you should be able to tell looking at chip datasheets, and I don't see why I'd expect anything audible.

If your DAC can do up to DSD512 or higher I would like to like to see how it affects <20kHz measurements as well. AFAIK Miska likes to upsample to the highest DSD rate that the DAC supports.

Audible or not, if an ADC can be affected then other equipment in the signal chain can be affected as well. Only showing a 2.5MHz screenshot without showing <20kHz stuff in detail is misleading.

 

[JustIntonation]

I'm beginning to think though.
On my current very uncritical living room setup (temporary listening spot for me now) I can not tell the difference between the digital players in the test linked to before.
But I could tell the difference between the onboard USB input of my Anedio D2 DAC and the external Armature Hecate (Singxer F-1) USB-S/PDIF converter in the high treble on this setup.
This probably means something was going pretty wrong with my D2 USB input.

I was using a 2m long cheap USB cable to the D2, and a 5m cable didn't work it gave me very severe continues dropouts.
Now I'm using a 0.4m USB cable to the Armature Hecate.
I was previously thinking that as long as you don't get dropouts the audio is bit perfect. But a few days ago I read something different on this forum (can't remember the thread).
Can this indeed be the case? No audible dropouts yet degraded soundquality??

If so this may be very worth keeping in mind.
I once read that there are differences between USB output chips on computers with a cheaper variety not always up to spec. Perhaps combined with a long cable this can lead to being enough out of spec to give trouble before dropouts occur?
And what impact can drivers have on this? Was using very old xmos drivers branded by Anedio before, and now the newest XMOS drivers for the Hecate. Is there error correction involved in these drivers? (also newer generation of chips)

Btw one other unrelated thing to perhaps watch out for.
My Anaview AMS0100_2300 amp has no input stage, it has a variable input impedance between 1.39 kOhm and 12.5kOhm depending on the input.
The sound is therefore very sensitive to the output impedance of the source. I've noticed that the sound changes noticeably with long cables vs short cables. And a DAC output impedance difference will surely have an effect as well.

It seems to me than in practical home setups these kinds of issues can come to play. Where a DAC which measures very good under ideal settings may give audible artifacts in a users system because of USB issues or impedance issues?

 

[Blumlein 88]

Why is 2.5MHz interesting? One can simply look at tweeter response, air transmission loss, and the eardrum and ear canal impedance and ask that question very simply.

https://www.mne.psu.edu/lamancusa/me458/10_osp.pdf

http://www.sengpielaudio.com/calculator-air.htm

Sorry, physics works.

Miska was not saying you heard mhz or that such signals ended up transmitted via the air.

The original question Miska was responding to is recording various DACs using an ADC running 24/96 khz while the DACs are playing 44.1 or 48 khz material. Archimago had recorded players this way to see if people could hear a difference. Miska's opinion is listening to a recording this way wasn't equivalent to listening to the DAC itself due to some ultrasonic content. Not that he thought you heard ultrasonics, but that it could cause audible artifacts in other parts of the system that might occur from IMD back below 20 khz. And your ADC filters would cut that off. Plus ultrasonic artifacts of DACs running at 48 khz could alias down into the audible spectrum because ADC's aren't perfect and alias free. That such aliasing wouldn't be present if you were listening to the actual DAC. So one way it removes ultrasonics plus the ultrasonics could cause aliasing not otherwise present.

Now from that point he shows various imaging in DACs out to megahertz regions. I think it was some imaging at 2.5 mhz at -65 db. I don't think it likely that would alias down into the below 20 khz range at any level that mattered. I've seen a couple ADCs that will have an elevated noise floor by a few db if high level stuff is going on up to 200,000 hz. But we are talking raising the noise floor below 20 khz by 3-6 db and still being -100 db or so. Whether such an ADC responds at all to some low signal level megahertz signal I don't know, but doubt it.

 

[Blumlein 88]

If your DAC can do up to DSD512 or higher I would like to like to see how it affects <20kHz measurements as well. AFAIK Miska likes to upsample to the highest DSD rate that the DAC supports.

Audible or not, if an ADC can be affected then other equipment in the signal chain can be affected as well. Only showing a 2.5MHz screenshot without showing <20kHz stuff in detail is misleading.

If I get a chance I may do some testing like that. The DAC I have will do DSD 256 I think.
Here is the testing I posted earlier about ADC's. You can see they all have their noise floor raised a little by ultrasonics.
https://www.audiosciencereview.com/forum/index.php?threads/comparing-aliasing-in-three-adcs.3272/

I'd assume the IMD that shows up is from the analog part of the circuitry in the DAC.

 

[levimax]

Two things I've been wondering since I've been visiting this forum is what do different types of distortion actually sound like, and since most tube/hybrid amps measured so far only barely meet the audibility threshold are there really such things as "tube sound" and "pleasant distortion"

I seems 2nd order distortion is "good distortion" if there is such a thing. Single ended triode tubes and phono cartridges make lots of it. I have seen it mentioned that 2nd order can "mask" other distortions. Of course there is controversy surrounding this subject. If its true 2nd order is different than others then maybe there needs to be a "weighted distortion" like they use for "noise'. I would be interested in reading some good studies on this.... So far most of what I have read is not what I would consider scientific.

 

[andreasmaaan]

I seems 2nd order distortion is "good distortion" if there is such a thing. Single ended triode tubes and phono cartridges make lots of it. I have seen it mentioned that 2nd order can "mask" other distortions. Of course there is controversy surrounding this subject. If its true 2nd order is different than others then maybe there needs to be a "weighted distortion" like they use for "noise'. I would be interested in reading some good studies on this.... So far most of what I have read is not what I would consider scientific.

I posted this in another thread recently. I'd like to see some more detail than contained in this study, but clearly there is starting to be some significant and quite scientific work done on this.

EDIT: not on the question of whether 2nd order distortion is good/preferred per se, but rather on developing better distortion metrics.

 

[Blumlein 88]

I seems 2nd order distortion is "good distortion" if there is such a thing. Single ended triode tubes and phono cartridges make lots of it. I have seen it mentioned that 2nd order can "mask" other distortions. Of course there is controversy surrounding this subject. If its true 2nd order is different than others then maybe there needs to be a "weighted distortion" like they use for "noise'. I would be interested in reading some good studies on this.... So far most of what I have read is not what I would consider scientific.

Someone posted a study of weighted distortion just yesterday I think.
http://hifisonix.com/wordpress/wp-content/uploads/2017/11/Perceptual-Levels-of-distortion.pdf

 

[levimax]

I posted this in another thread recently. I'd like to see some more detail than contained in this study, but clearly there is starting to be some significant and quite scientific work done on this.

EDIT: not on the question of whether 2nd order distortion is good/preferred per se, but rather on developing better distortion metrics.

Thanks for the link! I don't think any distortion is "good" but both objectively and subjectively there seems to be evidence that not all distortion is perceived the same.

 

[March Audio]

Now from that point he shows various imaging in DACs out to megahertz regions. I think it was some imaging at 2.5 mhz at -65 db. I don't think it likely that would alias down into the below 20 khz range at any level that mattered. I've seen a couple ADCs that will have an elevated noise floor by a few db if high level stuff is going on up to 200,000 hz. But we are talking raising the noise floor below 20 khz by 3-6 db and still being -100 db or so. Whether such an ADC responds at all to some low signal level megahertz signal I don't know, but doubt it.

and this is the point. If we have aliasing within the normal audio band then we can measure it. Any competent ADC should be correctly input filtered, not just for anti aliasing but also for RFI.

 

[andreasmaaan]

Thanks for the link! I don't think any distortion is "good" but both objectively and subjectively there seems to be evidence that not all distortion is perceived the same.

Absolutely, no doubt about it. This is also satisfyingly consistent with studies investigating auditory masking.

There's also some evidence that low (just-audible) levels of distortion may tend to be preferred by some listeners, although that's not something that's been systematically investigated to my knowledge.

 

[levimax]

Someone posted a study of weighted distortion just yesterday I think.
http://hifisonix.com/wordpress/wp-content/uploads/2017/11/Perceptual-Levels-of-distortion.pdf

thanks! andreasmaaan posted same link

 

[andreasmaaan]

thanks! andreasmaaan posted same link

I think it may have been me who posted it yesterday (or day before on my side of the world)

 

[j_j]

and this is the point. If we have aliasing within the normal audio band then we can measure it. Any competent ADC should be correctly input filtered, not just for anti aliasing but also for RFI.

Well, yes, it must be. My goodness. And DAC outputs that are delta-sigma should have something to filter out all that RF before it hits your equipment, too.

Aliasing can indeed be audible to some few people even at 20kHz. We have a couple of listeners (who are young) who can detect aliasing when a redbook is upsampled by 2 without a filter.

Of course, no filtering means that you also make your time resolution worse. I know that sounds odd, but it's a fact.

 

[j_j]

Thanks for the link! I don't think any distortion is "good" but both objectively and subjectively there seems to be evidence that not all distortion is perceived the same.

Well, yes, high-order distortion is bad news. Low order can actually, when you're on the "edge" create a sense of exaggerated dynamic range via tricking the ear into hearing more loudness per input power (since it widens the spectrum).

 

[Blumlein 88]

Oops, guess me posting the link to the paper on distortion wasn't much help then.