What is the cause of "digital glare"?

[gvl]

.

We just need @Blumlein 88 to run a ESSDAC->ADC->ESSDAC loop.

https://www.audiosciencereview.com/...8th-generation-digital-copy.6827/#post-151814

Very timely. I'm on it

 

[March Audio]

I was of course using a March Audio DAC with an ESS 9038. And most comments on hearing a difference (I'm mum on whether they were correct) have been of the variety that the 8th generation copy was softer and took away sparkle, air and dulled the edge of transients. Hmmmm, doesn't sound like a description of 800 % elevated glare does it? Removing sighted bias makes a heck of a difference.

Or to paraphrase Rick James, "sighted listening is a helluva drug!"

How did I miss this thread?

I will take a listen

 

[JustIntonation]

For what it's worth. I think I may have picked up on the "ESS glare" with the mobile DAC version of the chip in the SMSL M8A v2 (or was it v3 can't remember).
And I did not pick it up with the two full ESS DACs I heard, the Anedio D2 and a Melokin DA1.
For me I think it came across as a frequency thing mostly, a slight lack of mid-bass and a bit of forwardness in the mid-treble to high treble. (edit: subjectively, I have no doubt it measures flat)
I later though that maybe it was due to the "ESS IMD hump" that's most pronounced on the mobile DAC chip versions but I don't really know.. Atleast it's something that shows up on the measurements.

 

[andreasmaaan]

For what it's worth. I think I may have picked up on the "ESS glare" with the mobile DAC version of the chip in the SMSL M8A v2 (or was it v3 can't remember).
And I did not pick it up with the two full ESS DACs I heard, the Anedio D2 and a Melokin DA1.
For me I think it came across as a frequency thing mostly, a slight lack of mid-bass and a bit of forwardness in the mid-treble to high treble. (edit: subjectively, I have no doubt it measures flat)
I later though that maybe it was due to the "ESS IMD hump" that's most pronounced on the mobile DAC chip versions but I don't really know.. Atleast it's something that shows up on the measurements.

You could also try @Blumlein 88’s test. That’s on a DAC using the mobile version of the 9038.

 

[gvl]

Turns out I have an ADC, it is a bit crappy but should do (Edirol UA-1EX). I'm not set up for a blind test with an assistant switching DACs, so the plan is to make a couple of level-matched recordings from different DACs, then randomize file names saving off the original->random name mapping in a text file, try to guess which DAC recording is playing and check against the saved off info. Sounds reasonable? Any suggestions for recording software on Windows? Audacity? It doesn't support ASIO out of the box, will recording through the Windows input device be as good? I obviously don't care about latency,

 

[SIY]

Sorting tests are a great way to do this. Level matched, as you mentioned. I use Adobe Audition- there's a lot of reported issues with Audacity (like recording 16 bit data when you think it's 24 bit...). Goldwave also works well.

Nice thing about both of those is that if levels aren't perfectly matched, you can do that post-recording.

 

[gvl]

Sorting tests are a great way to do this. Level matched, as you mentioned. I use Adobe Audition- there's a lot of reported issues with Audacity (like recording 16 bit data when you think it's 24 bit...). Goldwave also works well.

Nice thing about both of those is that if levels aren't perfectly matched, you can do that post-recording.

Thanks, I'll check those packages. I will be playing CD-quality and use JRiver's 64-bit volume control to level-match outputting to 24 bits with TPDF dither. Or should I use a passive attenuator after the DAC?

 

[levimax]

Turns out I have an ADC, it is a bit crappy but should do (Edirol UA-1EX). I'm not set up for a blind test with an assistant switching DACs, so the plan is to make a couple of level-matched recordings from different DACs, then randomize file names saving off the original->random name mapping in a text file, try to guess which DAC recording is playing and check against the saved off info. Sounds reasonable? Any suggestions for recording software on Windows? Audacity? It doesn't support ASIO out of the box, will recording through the Windows input device be as good? I obviously don't care about latency,

Recording using Audacity is very flexible and easy, don't worry about exact level match or file names just make sure samples are same length and starting point in song.
Playback recordings using foobar2000 ABX plug which does the rest: level match, randomizes, keeps score, generates "tamper proof" report.
All for free

 

[JustIntonation]

You could also try @Blumlein 88’s test. That’s on a DAC using the mobile version of the 9038.

Aah didn't realize this was the mobile version of the chip. Thanks.
Gave it a quick listen but on my current temporary setup I can't hear a difference.
But this is a very uncritical setup. Old mission speakers (which I got for free) 10 feet from listening position in an untreated living room. I'm practically DAC deaf here at the moment. But I've saved the files and will do a proper blind test once I have my new studio setup/room, will report back here again then.

 

[gvl]

Recording using Audacity is very flexible and easy, don't worry about exact level match or file names just make sure samples are same length and starting point in song.
Playback recordings using foobar2000 ABX plug which does the rest: level match, randomizes, keeps score, generates "tamper proof" report.
All for free

I've been purposely avoiding foobar2000, I guess it is one reason to install it.

 

[levimax]

I've been purposely avoiding foobar2000, I guess it is one reason to install it.

One nice thing about foobar2000 is that unlike so many programs that try to take over your computer when you install them, foobar has good manners, it just installs and if you don't like it uninstalls.

 

[LightninBoy]

Indeed, the audiophile press seems to be in collusion with the small uncompetitive manufacturers, be it of vinyl decks or valve amplifers. Each time I attend live performances of (mostly larger scale) classical music, I am struck by the sometimes harsh glare of how music really sounds.

I still remember the first time I played with a live drummer. I was standing next to the kit and he hit that snare and hoooooolllllllllly baaaaaalllsssssssss was that loud and harsh.

 

[Blumlein 88]

Sorting tests are a great way to do this. Level matched, as you mentioned. I use Adobe Audition- there's a lot of reported issues with Audacity (like recording 16 bit data when you think it's 24 bit...). Goldwave also works well.

Nice thing about both of those is that if levels aren't perfectly matched, you can do that post-recording.

Yes, on Audacity, to my knowledge there is no way to record 24 bit in Windows. You can manipulate and do all sorts of other things, but recording into it is 16 bit only. You can record no problem 24 bit on a Mac. Same issue playing back as well.

On Windows I play back with Foobar with ASIO or WASAPI. I use Reaper to record which will do 24 bit in Windows. Sometimes use it for playback as well.

 

[gvl]

I still remember the first time I played with a live drummer. I was standing next to the kit and he hit that snare and hoooooolllllllllly baaaaaalllsssssssss was that loud and harsh.

I was thinking about these same lines, invite a Jazz band to your home and then think if this is glare or not. In that sense I feel the Tone Board is closer to the real thing than others.

 

[levimax]

Yes, on Audacity, to my knowledge there is no way to record 24 bit in Windows. You can manipulate and do all sorts of other things, but recording into it is 16 bit only.

It looks like Audacity in Windows will record to 32 bit float depending on device https://forum.audacityteam.org/viewtopic.php?t=100593

 

[andreasmaaan]

Yes, on Audacity, to my knowledge there is no way to record 24 bit in Windows. You can manipulate and do all sorts of other things, but recording into it is 16 bit only.

Isn't the default setting 32-bit floating point? Or is there something going on here that's not visible?

I was thinking about these same lines, invite a Jazz band to your home and then think if this is glare or not. In that sense I feel the Tone Board is closer to the real thing than others.

But wouldn't the glare then clearly be audible after 8 generations (i.e. Blumlein's test)?

 

[SIY]

Isn't the default setting 32-bit floating point? Or is there something going on here that's not visible?

There may be, yes. I haven't done it myself, but Scott Wurcer looked at a bunch of supposed high res files generated in Audacity and found they were actually 16 bit- I think zero-stuffed, but I don't remember. I stopped using it, and my 24 bit Audition files are truly 24 bit.

 

[Blumlein 88]

It looks like Audacity in Windows will record to 32 bit float depending on device https://forum.audacityteam.org/viewtopic.php?t=100593

Depending upon the device.

I've only had hands on a handful of recording interfaces. None of them will record using WASAPI in Audacity. Some will playback with WASAPI some won't.

What Audacity needs is ASIO support, but ASIO is not open source so it isn't happening.

 

[Blumlein 88]

Isn't the default setting 32-bit floating point? Or is there something going on here that's not visible?

Default is to open into 32 bit float. You can set it in preferences to be what you wish. 16 bit, 24 bit or 32 bit float. But it usually won't access anything other than 16 bit in Windows for most devices. Even though the tab is showing 32 bit float. In a Mac it'll work fine. In Linux you also have to jump thru some hoops to get 24 bit in and out.

I guess I'm too much an Audacity Cowboy.

And too cheap for software as a service Adobe. Plus I'm still mad they didn't honor my Cooledit license when Adobe purchased Cooledit and turned it into Audition.

 

[andreasmaaan]

There may be, yes. I haven't done it myself, but Scott Wurcer looked at a bunch of supposed high res files generated in Audacity and found they were actually 16 bit- I think zero-stuffed, but I don't remember. I stopped using it, and my 24 bit Audition files are truly 24 bit.

Default is to open into 32 bit float. You can set it in preferences to be what you wish. 16 bit, 24 bit or 32 bit float. But it usually won't access anything other than 16 bit in Windows for most devices. Even though the tab is showing 32 bit float. In a Mac it'll work fine. In Linux you also have to jump thru some hoops to get 24 bit in and out.

I see. Hmm, not good. Glad it hasn't been my recording software of choice thus far...