"Official" Audyssey thread (FAQ in post #1) - Page 1642

The dynamic aspect of Dynamic EQ takes care of it by mapping the softer (or louder) parts in the input signal onto the right psychoacoustic "curve" since Dynamic EQ looks at the Volyme dial AND the level of the (content) input.
A static approach like THX Loudness Plus would apply the same bass boost curve independant of if it were a soft or loud part in the input signal since it only looks at the Volume dial.

You can test it yourself by running a test sweep in REW or similar and using the same master volume setting but varying the input signal level, such as -10dbFS, -20dbFS and -30dbFS for Dynamic EQ and THX Loudness Plus.
You will see that the curve is shifted 10dB each time with Dynamic EQ and the slope is changed, ie different psychoacoustic curve based on input and volume setting.
But with THX Loudness Plus, it will be the same static curve, just shifted 10db each time.

Because if you want to utilize equal loudness curves to take into account human perception and at the same time preserve the directors intended balance within the content itself...then the dynamic part becomes important.

What other aspects, besides attempting to keep perceived loudness 'in-tact' between all the speakers as MV is lowered, does DEQ offer/achieve/strive for?

What do you mean by frequency response correction? Isn't the frequency response handled by the iteration of Audyssey being used when initial calibration is run and completed?

I'll take a shot at it.

Let's use the example of a train recording (I guess it could also be a marching band!). The top line in each of these diagrams is the train's spectrum as it passes close by. The lower line is the final loudness as the train travels far away. The overall SPL reduces 40 dB during the train recording.



Blue = Playback at Ref volume. When the train is close, we hear all frequencies equally loud. When it is far, the bass is naturally weaker (just like real life), but it is audible.






Red = MV at -30. When the train is close, the bass is weak. When it is far, the bass is much weaker, and becomes inaudible below 60 Hz.






Green = Fixed Loudness Compensation applied. When the train is close, we hear all frequencies equally loud. When it is far, the bass is weak below 100 Hz, and inaudible below 30 Hz.






Grey = Dynamic EQ applied. When the train is close, we hear all frequencies equally loud. When it is far, the bass is weaker, but it is audible. The arrows show where the lower curve had additional bass boost applied in response to the quiet program level. The amount of compensation needed has changed as the train went from loud to soft. Fixed compensation cannot do that.






Blue/Gray = Before /After with Dynamic EQ

Edited by Roger Dressler - 6/30/12 at 1:33am

Roger, that was awesome

Keith -- No offense, but I think you are being a bit too cavalier about declaring Audyssey's tech to have been based an a "false premise" as though it has been factually established as such.

That's what is rankling nathan_h and my dear Feri so much. Roger (with all due respect) has proposed his theories as to why he thinks Dynamic EQ is flawed, based upon his (vast) knowledge and the snippets Chris has provided about their research methodology.... but I think it's awfully presumptuous to assume the book is closed on this discussion and we can simply discard all of the research and testing Audyssey has done -- the vast majority of which we are not privy to.

They did not invent the idea of the surrounds needing to diminish less than the fronts when the volume drops, as noted THX LP incorporates the same idea. Roger's test IMHO establishes what we already know -- at below reference volumes, the surrounds will be louder than the fronts. That doesn't necessarily prove that Audyssey was WRONG in their approach or that the "surround boost" doesn't accomplish its goal of maintaining perceived surround envelopment at below reference volumes with *real world* content.

My dear friend Keith, let me get this straight. After you've created such a huge stir, pages and pages of posts with hypothetical marching bands and whispers, accused Audyssey of a significant scientific blunder in implementation of DEQ-after all that, NOW you're happy with DEQ's effect on real-world content? Now that you've finally done what I told you to do in the first place?
edit: sorry, not piling on my buddy Keith-I'd not seen batpig's post

+1. Totally awesome. And clear too. Thanks, Roger.

Fair enough, batpig. But Chris himself did say in one of his explanations that they dropped some of the old theory because it was based on, IIRC, primitive equipment and was therefore no longer correct. Roger explained that it may no longer be applicable but it was nothing to do with the equipment being primitive. This is from memory - the post is somewhere not too far back in this huge debate! This shows that either Chris wasn't understanding something correctly himself (seems unlikely) or that he didn't want to share something with us (perhaps more likely for commercial reasons).

I am not sure you are right when you say that at below reference volumes the fronts will be louder than the surrounds (unless the surrounds have been deliberately mixed softer - the point of the 'flaw' in the Audyssey method). Roger's test shows that the fronts and the surrounds maintain their equal loudness when the MV is reduced unless DEQ is engaged when the surrounds become relatively louder.

Because it is a fact that the loudness of the fronts and the rears does not change in relation to each other (and BCJ;s test showed this clearly) when the MV is reduced and DEQ is off, then it still seems to me that this is a flaw in the way DEQ was developed.

It's a conceptual depiction, not based on actual loudness curves used by Audyssey or Dolby, nor using their actual frequency selective compensation. Setting aside the details, do you agree with the concept?

LOL! Yep. With the 5dB RLO that you did indeed mention about 40 pages back <<img src="http://www.avsforum.com/avs-vb/images/smilies/smile.gif" border="0" alt="" title="Smilie" class="inlineimg" />> I am happy with what DEQ is doing here. I still think the premise is flawed and BCJ's test, based on Roger's hypothesis, seems to prove it and I still think it is a form of distortion, but I have now tamed it such that the prior issues I had with it are of much less consequence! In my own defence, I would say that this is is the main purpose of this thread - to question, to learn and to adapt accordingly and I have now done all three! I do apologise to all concerned though for any offence caused, any thread derailment and anything else than anyone feels merits an apology - and especially so to you. I have also learned something else: take some freakin' notice of SoM when he tells you what to do!!!!

Next round of beers is on Keith

Wow! Thanks Roger. Perfect explanation. The graphs are a major help in understanding what DEQ is trying to accomplish.

How does DEQ know that the test signals are not real world content? Chris acknowledged that games (real world content wherein the surrounds are often just as loud as the fronts) need a different version of DEQ without the surround boost. That is a tacit admission that such boost is not needed for movies either, because movies can and do have equally loud surrounds from time to time. Doesn't that count?

BTW, the only reason I originally enjoined this topic and explored it, is because it didn't sound quite right in my MultEQ XT calibrated system, with or without RLO. Considering all the research, I expected more. I'm not here to tell anyone how to run their systems. Please decide for yourselves. Base it on what you hear, not what anyone tells you.

Can we agree that Keith may march to the sound of a different drummer in a different marching band?

Did you try THX Loudness Plus when you had the Onkyo?

DEQ attempts to do two things as you lower the volume level: keep the frequency response similar to what you heard at reference level AND keep the balance between the fronts and surrounds similar to what you heard at reference level. My point was that it would be a shame to lose the former just because you didn't like how the latter was handled.
Our human hearing doesn't track volume level changes in a linear fashion. As you lower the volume, bass appears to get softer at a quicker rate than the midrange; so does treble, though to a lesser extent. As such, DEQ starts applying a bass boost and mild treble boost as you lower the volume, in order to give the impression that frequency response is remaining consistent at lower levels. Not perfect, since the compensation is not based on your particular hearing, but still useful.
No. Audyssey MultEQ addresses unwanted contributions of the room, DEQ addresses inconsistencies in our hearing.

So now I have to ask: What results does THX loudness plus strive to achieve? Is it doing the same two things as DEQ (keep reference level response and perceived speaker loudness in-tact as MV decreases)? Does it implement major differences over DEQ or just minor ones?

Would you happen to have a graph of Loudness Plus like the one you had for DEQ, Roger?

Probably. Cannot recall any details about it.


No. I have not studied LP.

To be honest, I agree with you! It does make more sense to just monitor the levels in each channel and adjust based on loudness (i.e. boosting the surrounds when they have softer content, not just applying a global boost to the surrounds), and I have certainly been bothered by this behavior of Dynamic EQ even with movies. With video games it's downright annoying.

I just think "educated enthusiasts" like Keith (and me!) need to be a bit careful before declaring concretely that any professional's research is "wrong" or based on a "false premise". Your argument has a lot of merit and logical appeal, but I am personally not going to be so cavalier about declaring Chris' work to be invalid, especially when he is not here to defend himself and his teams' findings.


agreed

And the bass frequencies of the band I am marching to diminish gradually the further away it gets

LOL. It's warm, flat beer here you know - not sure how well that would go down in sunny New Mexico! But sure, there's a pint in the pump waiting for you

Is your band marching tonight under the Northern Lights?

Sadly I think we are just too far south to see them here. I just went outside to walk my dog before (his) bedtime and no sign of them. People slightly further north have seen some spectacular stuff by all accounts.

Some nice pics here: http://www.bbc.co.uk/news/uk-england-tyne-16679422

Yes, you are right. I'm not intending to try to discredit Chris in any way, nor indeed the sterling work that Audyssey have done in so many ways. I will certainly be happy, for the record, to declare my former assertions as just my opinion. MultEQ has revolutionised the sound I can achieve in my home - nothing else has ever come close to the impact that XT32 has had. I can definitely say I would never buy another AVP without it, unless something unequivocally superior comes along one day. As for Chris, his unfailing politeness and patience is a lesson to us all and I have the utmost respect for his professionalism. This discussion has been fascinating IMO and I have learned a lot from it and from my more learned colleagues - best of all I have made a noticeable, albeit subtle, improvement to the sound in my HT too!

Very annoying in games indeed. Now time to see which I enjoy better for movies lol

I saw the very persuasive hypothesis, ie, not conclusive data. Sorry if I missed it. I wouldn't be surprised.


I don't see how it's a premise. Audyssey indicates it's based on empirical research.