No Update FOR Audyssey?

[Matthew J Poes]

Sorry for the old thread but couldn't find anything more up to date...

Matthew, if we "give up" on EQ of high frequencies, does Dirac still offer an advantage over Audyssey?
Is there any consumer system (by which I mean not requiring hardware costing thousands of dollars) that has noticeably better ability to EQ bass and/or integrate a subwoofer?

What is the most budget-friendly option to get that done with reasonable results?

Audyssey-based products don't even have an option for any manual EQ - What are my alternatives that allow manual EQ at the Audyssey price range (x3700h price area)?

Assuming the manual EQ is done right, what is the difference in result between good manual EQ and Audyssey? I thought manual EQ will generally be PEQ filters (and seems like Anthem ARC is also PEQ) while Audyssey uses FIR? And Dirac uses a combination of both? Does that make a difference?

Getting back to integration, is there anything that also handles the phase differences between the speakers and the subwoofer and even the speakers themselves? Beyond just setting the delays so that all sounds arrive at the same time, which still leaves significant phase differences in many frequencies.

There is a lot to unpack here.

around the transition zone and below the problems in a room are almost completely minimum phase. PEQ is minimum phase. It doesn’t mean the EQ doesn’t change the phase. It absolutely does. It means the phase change is tied to the frequency change. As it needs to be. So correcting the amplitude response also fixes the phase response.

now let’s say we have other delay issues. Well…those can be addressed with time alignment just fine. Nothing fancy needed.

so if we stick to 250hz and below, the likelihood of there being much difference between IIR and FIR filters is likely nill. Exceptions would be just that. Unusual exceptions.

Audyssey uses only FIR which is very inefficiency at low frequencies. The filters also tend to have pre-ringing problems. Audyssey is fairly good, the issues are minimal. Their newest version also has enough taps to do a good enough job that it’s unlikely PEQ would be any better. The differences between manual and automatic tend to be more about how good a job you did setting each up and how good a job it’s algo did figuring out what to correct or not. I find when you get the mic positions right it isn’t bad. It’s why I tell people who have it to get the app and restrict to bass frequencies.

as for Audyssey vs Dirac. I keep telling folks I plan to do a comparison and the plans keep going to the back burner. As many know, I moved, built a house, and have been in limbo. The house will be move in ready in 3 weeks or so, I hope to restart these projects then. I have Dirac, Audyssey, and ARC available. Maybe can come up with more. My experience in the past was Dirac did a bit better job getting the bass right. The response was often that much smoother.

now…if we begin to compare the bass management and subwoofer alignment that Audyssey XT32 vs Dirac BM does, well now the prize firmly goes to Dirac I believe. I have limited experience with it as I moved soon after receiving my own device with Dirac BM, but my understand and limited experience of the two suggests Dirac is a lot better.

a lot of folks do the mics wrong when they do the setup. If you read the study that Olive did, he found that The sound quality preference was tied to how close together the mic positions are. You should always do multiple measurments about your head area. I argue at least 5, more is better. However, a lot of folks put the mic too far out. For the best sound quality in your primary listening position, these work best when the mic is only placed around that seat and not in any other seat. This helps bias the changes toward that seat while still minimizing errors from individual measurement points.

 

[GalZohar]

Thanks a lot of the reply!

What does Dirac BM do that others don't? Does the single subwoofer version come only with super-expensive receivers (even relative to Anthem)? Relative to Audyssey I suppose there's no real comparison as Audyssey comes with relatively cheap receivers. I'm wonder what Dirac does and if it might be possible to obtain in other (cheaper) means.

Seems like Audyssey (and arc?) just try to time align by measuring delay and EQ each speaker and each subwoofer separately? What is possible to do beyond that other than manually adjusting the subwoofer delay?

 

[Matthew J Poes]

Thanks a lot of the reply!

What does Dirac BM do that others don't? Does the single subwoofer version come only with super-expensive receivers (even relative to Anthem)? Relative to Audyssey I suppose there's no real comparison as Audyssey comes with relatively cheap receivers. I'm wonder what Dirac does and if it might be possible to obtain in other (cheaper) means.

Seems like Audyssey (and arc?) just try to time align by measuring delay and EQ each speaker and each subwoofer separately? What is possible to do beyond that other than manually adjusting the subwoofer delay?

I’ve done a few videos and talks on this. I will try to send links when I find them.

the fundamental problem is more complex than you are making it out to be. If the only go were a flat response, then that is easy. You invert the transfer function of the room. Instant perfect response. But that wouldn't actually fix anything. The goal is to try and use an algorithm that can infer from a set of steady state responses the true errors of the speaker vs room, separate those minimum phase problems that are spatially consistent and audible from those that are stochastic issues and inaudible. Above 500hz that is very hard to do. Most get this wrong. It’s why folks like Toole believe there is no point in even trying.

There are four major zones in a room that need to be acoustically addressed by this software. The stochastic zone is one where EQ is very difficult. Room reflections should not be eqed as the ear filters those out. The transition zone can be eqed to a point but must be done carefully. Reflection related anomalies should usually be left alone but some are appropriate to EQ. If they are spatially robust and minimum phase, EQ is certainly an option. The modal region or multi-modal region can be eqed. However it needs to be done in a spatially robust way. EQ for a single measurement point is often not a great plan. It can do more harm than good. Big peaks being knocked down is fine. But flattening it can be a real problem if the peaks and dips move with position.

the Unimodal region falls below that area, often starts around 35hz (but depends on the degree of LF absorption and dimensions of the room), and can always be eqed. It will always be minimum phase and spatially consistent. It’s the easiest of the problems to fix.

the stochastic region will never be a fully minimum phase problem. Minimum phase EQ automatically fixes the system phase so FIR filters have no value. But since this is not a minimum phase issue, EQ screws up the phase. Imaging would suffer greatly if you eqed each speaker and didn’t match the phase. The simplest means of matching phase would be to zero it out. This is what Audyssey and Dirac do (but not ARC). The trick is in the details. Because it is stochastic, phase is different for every measurement point. So you have to correct the response and phase to match each other on the balance. Dirac is more sophisticated than Audyssey is at doing this. They have a paper that explains how and why floating around. Audyssey is essentially and inversion of the average room transfer function with bandwidth and correction limits imposed and a error corrector that adjusts the correction to minimize the RMSE of phase and amplitude. Dirac does this with a greater focus on a flat target for both phase and amplitude. It leads to a flatter and more even response. It’s just more resource intensive to do.

The modal region has an issue. The problem is minimum phase (So FIR offers no real advantage and a lot of disadvantages), but it’s also spatially Inconsistent. How do you make a robust correction then? EQ can’t and no sustenance on the market fixes that. But what you can do is use multiple subwoofers to create a spatially robust problem and correct that. To do that you must phase align (but not time align) the subwoofers. You can’t time align because at the transition zone of the two the LCR’s have non-trivial group delay. A time aligned system would not be phase aligned at that frequency. You often see most corrections seek to be about one cycle behind to match the group delay.

Audyssey is very basic in how it handles this. Is limited to a pair of subs and works best with true crossovers. Dirac is far more complex now with the ability to handle infinite subwoofers and various filter settings. You can run the mains full range and still align everything with overlapping subs. An approach I favor.

as for cost. Pioneer and Onkyo now sell inexpensive receivers with dirac. I have one for review. It doesn’t have the advanced BM however. I will review in about a month.

 

[GalZohar]

What does the BM do though for the integration of subwoofer and mains? When you have a single subwoofer (or already optimized your multiple subwoofers to work best together), what is there to do other than adjusting delay of the subwoofer channel to best integrate with the mains?

I tried to measure my responses with different (single) subwoofer delay settings and got that the defautl audyssey delay best aligned the subwoofer with the left speaker, but I needed to increase the distance by 2 meters to best align it with the right speaker, which made a significant improvement to the right speaker, and didn't hurt the left speaker much at all, and the overall result is audibly better. Subwoofer is in a corner near the right speaker, and left speaker is very far from left wall as my whole system (and living room / HT section of the open floor plan) is not at the center of the wall.
Is there anything I can actually do better with Audyssey? What would a more advanced system do better about this?

Overall it sounds like if correction is limited to just low frequencies, results should be similar for Audyssey/Arc/Dirac? And if that is correct, then the only thing I don't really understand is what can be done about subwoofer-main integration beyond just adjusting delay (which even with Audyssey is possible in a manual approach).