Audyssey One - Nexus (Automated hi-res Denon/Marantz calibration with REW/Umik-1/2)

[anphex]

Any idea why Neuron 3.1 goes completely wrong here? I wanted to try real measurements made with measurement mic and REW, using timing reference, naming, sorting, but I always run into this problem. Doesn't happen when I use Audyssey extracted measurements.

Here is the .mdat file: https://anphex.one/index.php/s/wifcjK5NH73CLNk

 

[OCA]

If you are using the latest REW Beta (78) some important API endpoints have changed syntax and will cause errors with Neuron which I no longer update. You should use Acoustica with the latest REW.

 

[anphex]

If you are using the latest REW Beta (78) some important API endpoints have changed syntax and will cause errors with Neuron which I no longer update. You should use Acoustica with the latest REW.

The very same thing happens n Acoustica with Beta 78 with the same measurements. It happens once the first "A times B" appears.

 

[OCA]

The very same thing happens n Acoustica with Beta 78 with the same measurements. It happens once the first "A times B" appears.


View attachment 448439

Something is wrong with your setup, maybe measurement naming since this is the first stage before saving pre-processed mdat but not something that happens to anyone else.

 

[anphex]

@OCA Hmm it's weird, REW seems to enter an unstable behaviour after the first sw250Hzfix is being applied. The measurement, in my case index 44, is created and looks valid but starting from this point REW isn't really usable anymore and has trouble displaying any graphs. Doesn't happen with the Audyssey AVR measurements. The rendering of graphs completely breaks too.

Edit: I just used acoustica with the prior Audyssey odd.wtf measurements and there it doesn't set the speakers to "Small" nor does it enable "LFE + Main". Intended behaviour or quirk again? Oh well, I just set it manuallto "small" and crossovers where okay and it sounds fine!



(index):3521 POST http://localhost:4735/measurements/44/command 400 (Bad Request)

	fetchWithRetry	@	(index):3521
	fetchWithRetry	@	(index):3544
	await in fetchWithRetry
	postNext	@	(index):3565
	fixSubs4REW	@	(index):2794
	await in fixSubs4REW
	main	@	(index):1371
	setTimeout
	progress2	@	(index):1328
	checkRewMeasurements	@	(index):1328

Dunno what breaks here, but it seem some operation inside REW is going horribly wrong.

 

[OCA]

If this is Acoustica, my guess is that "trim Ir to windows" command is failing which indicates something fundamentally wrong with your measurement structure, as no user has reported this issue and hundreds are running the same script daily. Graphs are blocked by the script during optimization and will not get unblocked on uncaught errors. This is what is happening with REW after the error. I’d need your .mdat and .avr files and details of your platform/browser to identify your mistake, but I simply can’t help debug errors additionally on this platform due to time constraints. Check the app and you can find where to report such issues and get support.

 

[middlemarch]

@OCA, you old devil! I'd been under the assumption you were probably in your thirties given your energy level, but come to find out you're 55 years young today! Makes your work output even more amazing! Happy Birthday! And happy to report I'm running Acoustica V5 with my own tweaked Target Curve. I added a slope UP from 1.5K to 10K to help compensate for my old fart hearing loss, and it's working pretty good! I just spent four hours listening to my Mark Knopfler station on Pandora using mostly the Auro3D upmixer and it was Glorious! Very happy! Again, Happy B'day to you, sir!

 

[anphex]

Hi @OCA,

I have a quick question regarding my subwoofer distance after calibrating.

Problem Description:After calibration, my subwoofer is displayed with a distance of 9.99m in the Distances menu. If I open this menu after uploading the calibration data to my AVR (Denon AVR-X2800H), the AVR immediately flags this value as invalid. Consequently, I'm now avoiding opening that menu after uploading the calibration.

My Question/Suspicion:Does this maximum value of 9.99m indicate that my subwoofer's internal signal processing (delay) is very long, and even with the hack tool, Audyssey has reached the maximum possible delay time it can set?

Edit: I forced the lowest crossover search to 60 Hz and now the distance is 9,29m and it sounds just perfect. So it actually IS an incredibly slow processing of the soundbar.

Background on My Setup:I'm using a "repurposed" soundbar (more accurately, a rather large soundbase) as a subwoofer because it goes down to 30 Hz at -3dB. The sound is fine, and all settings from version 5 of Acoustica were uploaded cleanly to my AVR-X2800H.


Additional Consideration:I'm currently unsure whether I should continue using the soundbar as a subwoofer or remove it. My main speakers (KEF R3), located in the same room (bedroom), measure down to a pretty clean 40 Hz due to wall reinforcement. Would it perhaps be more ideal to forgo the soundbar as a subwoofer, or does it still offer advantages despite the 9.99m distance reading?

Thanks in advance for your assessment and help!

 

[OCA]

Hi @OCA,

I have a quick question regarding my subwoofer distance after calibrating.

Problem Description:After calibration, my subwoofer is displayed with a distance of 9.99m in the Distances menu. If I open this menu after uploading the calibration data to my AVR (Denon AVR-X2800H), the AVR immediately flags this value as invalid. Consequently, I'm now avoiding opening that menu after uploading the calibration.

My Question/Suspicion:Does this maximum value of 9.99m indicate that my subwoofer's internal signal processing (delay) is very long, and even with the hack tool, Audyssey has reached the maximum possible delay time it can set?

Edit: I forced the lowest crossover search to 60 Hz and now the distance is 9,29m and it sounds just perfect. So it actually IS an incredibly slow processing of the soundbar.

Background on My Setup:I'm using a "repurposed" soundbar (more accurately, a rather large soundbase) as a subwoofer because it goes down to 30 Hz at -3dB. The sound is fine, and all settings from version 5 of Acoustica were uploaded cleanly to my AVR-X2800H.


Additional Consideration:I'm currently unsure whether I should continue using the soundbar as a subwoofer or remove it. My main speakers (KEF R3), located in the same room (bedroom), measure down to a pretty clean 40 Hz due to wall reinforcement. Would it perhaps be more ideal to forgo the soundbar as a subwoofer, or does it still offer advantages despite the 9.99m distance reading?

Thanks in advance for your assessment and help!

Subwoofer distances are inherently longer than their actual physical distances due to large drivers and active filters in them but Acoustica adjusts sub distance to compensate for dips in speaker responses and will often add up further delays. Generally ,the longer the sub distance, the tighter bass you will end up with.

 

[anphex]

Subwoofer distances are inherently longer than their actual physical distances due to large drivers and active filters in them but Acoustica adjusts sub distance to compensate for dips in speaker responses and will often add up further delays. Generally ,the longer the sub distance, the tighter bass you will end up with.

Thank you! My question was wether 9,99m is the actual peak value you can use with Audyssey and that my sub at 40 Hz might have needed more delay offsert than Audyssey/Acoustica could provide. A warning in the logs in that case would be nice so the user get's an alert when something goes off the given parameters and the calibration is reaching extreme values.


Edit: Another, maybe silly, quetsion: I made the manual REW measurements which caused crashes in the optimization (a few posts above) like this: I didn't use the dolby sweep but used internal REW 256k Sweeps with acoustic timing reference. Was that the culprit?

 

[middlemarch]

@OCA

Question for you if I can pull you away from your GPU work for a moment. I've been trying to use target curves to add some boost to the high end to help with my hearing loss.

However I was informed by someone (LBNL) on the AVS forum that adding positive boost on the high end (I curve up from 2K to 10K ending with plus 6dB) results in a commensurate decrease in the bass end of the curve. He describes it as a see saw effect. Tilting up the highs results in tilting down the lows.

I have not yet done measurements to confirm this, but could you weigh in on this? Is there any way to defeat this behavior? I had been under the impression that a target curve was consistently followed from one end to the other of the spectrum.

 

[Keith_W]

However I was informed by someone (LBNL) on the AVS forum that adding positive boost on the high end (I curve up from 2K to 10K ending with plus 6dB) results in a commensurate decrease in the bass end of the curve. He describes it as a see saw effect. Tilting up the highs results in tilting down the lows.

I have not yet done measurements to confirm this, but could you weigh in on this? Is there any way to defeat this behavior? I had been under the impression that a target curve was consistently followed from one end to the other of the spectrum.

This is a pretty easy question to answer, no need for @OCA for this one Consider the three target curves that I quickly drew for you. On the left is the typical falling response, in the middle is a flat response, on the right is an upward tilt. All the target curves need to be normalised to 0dB - i.e. the loudest part of the target curve is 0dB, and any adjustments are a volume cut relative to 0dB. This is not unique to what OCA did, all target curves work like that.

Now ask yourself how you would increase the top end without getting a bass decrease. The only way to do that is to use a flat target curve, where the entire frequency response is at 0dB. The moment you put any adjustments in, it will be less than 0dB.

A flat target curve (or even a rising target curve) will sound horrible to most people, but if you have hearing loss it might sound better to you.

I hope this makes sense.

 

[ObjectiveSubjectivist]

Did somebody say nexus?

 

[middlemarch]

View attachment 461166

This is a pretty easy question to answer, no need for @OCA for this one Consider the three target curves that I quickly drew for you. On the left is the typical falling response, in the middle is a flat response, on the right is an upward tilt. All the target curves need to be normalised to 0dB - i.e. the loudest part of the target curve is 0dB, and any adjustments are a volume cut relative to 0dB. This is not unique to what OCA did, all target curves work like that.

Now ask yourself how you would increase the top end without getting a bass decrease. The only way to do that is to use a flat target curve, where the entire frequency response is at 0dB. The moment you put any adjustments in, it will be less than 0dB.

A flat target curve (or even a rising target curve) will sound horrible to most people, but if you have hearing loss it might sound better to you.

I hope this makes sense.

Thank you for that. I was not taking into consideration the normalization. In effect, no true "boost" occurs at either end of the spectrum, in actual fact the midrange (say the 1KHz point) is reduced in level. So the overall level is reduced leaving the "boosted" points alone. Got that.

What confused me was the way in which it was described, that there is a "see/saw effect". The resultant spectrum does not pivot about the 1KHz point as this would imply. Instead both the bass and midrange levels are reduced by the amount of "boost" applied to the treble. Is my understanding correct?

I'm well aware of what needs to be done when applying boost during a regular equalization by reducing the overall gain to prevent clipping, just wasn't applying that to the target curve, just saw it as some kind of magic, I guess.

I still struggle with conceptualizing the seeming difference in applying "boost" to bass frequencies vs. treble, in that a 3db increase at 30Hz requires far more power than a 3db increase at say 5KHz, so it would seem from an intuitive standpoint that the two are not equivalent, but in terms of voltage applied they are the same. I just said that, but still have trouble getting my head around it.

 

[masterhw]

The resultant spectrum does not pivot about the 1KHz point as this would imply. Instead both the bass and midrange levels are reduced by the amount of "boost" applied to the treble. Is my understanding correct?

My understanding is OCA is "normalizing" to achieve a target SPL (typically around 75 dB) accounting for human hearing contours (ISO 226). Thus it is more complex in practice than "I want to add +3dB to one end, so I have to take it off the other" as the loudness contours make things non-linear. But accounting for that, this is essentially what does. If you were to only add gain, it would rise above the ~75dB target. The EQ curves are always reweighted when applied to produce approximately equal loudness as the human ear perceives, which does mean taking away from some regions.

Thus when dealing in EQ curves, the absolute levels across the spectrum don't really matter. Only where they are relative to the rest of the spectrum. As any room correction suite worth its name will have its own weighting approach that discards whatever the absolute levels were in the first place, and thus it doesn't matter if the preference curve was dragged down to -10 or dragged up to +10.

 

[middlemarch]

My understanding is OCA is "normalizing" to achieve a target SPL (typically around 75 dB) accounting for human hearing contours (ISO 226). Thus it is more complex in practice than "I want to add +3dB to one end, so I have to take it off the other" as the loudness contours make things non-linear. But accounting for that, this is essentially what does. If you were to only add gain, it would rise above the ~75dB target. The EQ curves are always reweighted when applied to produce approximately equal loudness as the human ear perceives, which does mean taking away from some regions.

Thus when dealing in EQ curves, the absolute levels across the spectrum don't really matter. Only where they are relative to the rest of the spectrum. As any room correction suite worth its name will have its own weighting approach that discards whatever the absolute levels were in the first place, and thus it doesn't matter if the preference curve was dragged down to -10 or dragged up to +10.

I think for me, rather than try to overthink this, I may be better off sticking with "it's magic". Should I give up on efforts to try to compensate a bit for my hearing loss, or is there a way to do this more intelligently?

I have to say I've been pretty happy with my attempts so far to add back a bit of "sparkle" to the high end, but haven't been so happy with the bass, and this helps explain why. Bass was better before trying my tweaking.

 

[Keith_W]

What confused me was the way in which it was described, that there is a "see/saw effect". The resultant spectrum does not pivot about the 1KHz point as this would imply. Instead both the bass and midrange levels are reduced by the amount of "boost" applied to the treble. Is my understanding correct?

Well the see-saw effect is kind of correct, except that the fulcrum (pivot point) isn't in the centre of the curve, it's at the loudest part of the curve.

If you know what specific frequencies you have hearing loss, you may be able to design a target curve that boosts frequencies which you have difficulty hearing, a bit like this:

The left part of the curve up to 1kHz is a normal Harman target curve. The right part of the curve above 1kHz is a high shelf filter that boosts the treble. I drew this as a quick illustration, if your hearing loss occurs above 5kHz then you can apply your high shelf there.

 

[middlemarch]

Thank you, Keith and Masterhw. I've also been doing a series of Acoustica runs with the different target curves. Below is the one I'm currently playing with.
Contrasted with a version of this TC with highs flat to 22K, Acoustica reduces the overall level (including the bass region of course) by about 3dB, possibly because I'm not using a high shelf, plus whatever other magic he's doing.

I can then add back the reduced bass level by adding 3dB of bass fill. So I "think" I can accomplish my goal now that I understand the process a bit better. I can also now better understand why B&W sells as many speakers as they do to the geezer community...

 

[Terplover]

Hello guys, is there any chance of getting this to work without a Sub?

 

[jhwalker]

Hello guys, is there any chance of getting this to work without a Sub?

View attachment 461317

No. A subwoofer is required, as sub integration is a large part of what A1 does.