Have you set up a system and used room correction software in a room that's known to be considered 'bad' WRT acoustics? What were the net gains in sound quality?
What are the assumptions for the rooms intended for using electronic correction? Are the designers assuming bare floors, carpeted, drapes, all hard surfaces....?
I’ve used correction in a lot of different systems and acoustic spaces. I don’t know if it’s exavtly fair to say that they set up room correction based on treatment assumptions. I think that these algorithms don’t really care that much about the treatment, they are designed to compensate in any environment. I believe that Dirac Live and Audyssey are sold for sure in large acoustic spaces and I believe they are not modified for this.
I know that they normally test them in rooms with different levels of absorption and assume that home will be all over the place. They seek robust correction in a variety of environments.
Room correction can’t change certain things in a system which are conflated with other “things” in a typical room measurement. For example, let’s take a peak in the bass caused by the subwoofer having a non-flat response. That peak may look different in different places around the room, but it’s effect will always be present. Thus as a minimum phase phenomena it can be corrected with eq. A peak caused by a mode is also minimum phase and can be corrected by eq, but modal peaks are position dependent. It’s effect is not felt throughout a room. As such that eq could actually make things worse near by. How near by depends. To know for sure we need many measurements around a varied space to confirm how the peak changes. If the peak remains in most measurements about which you sit, eq is a good fix. If it moved around and goes away or even becomes a dip in that same region of space, eq is a bad idea (and acoustic treatment may be a better idea).
At mid and high frequencies a different and more significant set of conflated issues take over. That is, the in-room measurement is a mix of direct and reflected energy. As such the response tends to follow the early reflection response. In fact an average of the listening window and power gives you something very close. Exactly how closely it matches this depends on the wall absorption. The more absorption the greater the ratio of direct to reflected energy and the closer the response would match the listening window (but don’t forget that research doesn’t support listening in an anechoic chamber). Room correction can only correct what it can measure in the room. Reflections, diffraction, resonates, along with all the linear distortions of the speaker combine in the measurements. However the fix to each of those is different. Eq is not the fix to all, it’s not even a good bandaid for some of them.
So a really robust room correction looks at all the measurements and basically looks for evidence that certain problems are moving around in space. If a peak changes it’s center frequency, that is likely diffraction. Can’t eq that. If a peak or dip changed amplitude in space but not frequency, probably a resonance or reflection. The source of the reflection matters for eq. If it’s in the room, then as long as you can show that it’s spatially robust it may be ok to eq. However that is rarely true. If it’s in the speaker it depends. If the peak or dip is constant with angle, you can also eq. IF it changes with angle (poor offaxis response) you can’t eq. However it’s hard for an automated eq system to manage this.
Then you have another issue. While rooms are minimum phase and speakers are generally minimum phase, stochastic room/speaker interactions (what happens above the transition zone) is generally mixed phase. PEQ is minimum phase so you can’t fix a mixed phase system with a minimum phase filter without introducing more problems. Harman uses minimum phase filters but does so in a safe way. They only use in-room measurements for issues at and below the transition zone. They fully characterize a speaker and apply eq to speaker problems in the stochastic zone based on free space characterization. These problems are still minimum phase and so correctable. Dirac uses a mixed phase filter and can basically fix both minimum phase and linear phase problems. It’s trick is that it has to discern the minimum phase speaker problem and address it in a way that assumes the speaker has a constant response with changing angle. Where it doesn’t (and most don’t) it needs to figure that out and not correct. It does that by taking a lot of measurements over a varied space and looking at differences in each for clues of a speaker that has a poor polar response. This is why single point measurements are a bad idea and further why I don’t recommend doing it with Dirac.
I won’t speak for Toole, but I think his argument is that this automated approach is not needed. That it makes more sense to build Better speakers and apply correction to minimum phase problems. These problems that arise in the stochastic region are likely not audible. Whatever correction is fixing may not matter much and runs the risk of making things worse. I e echoed this by saying that I have found that good speakers in a good room are often not helped even by Dirac. It’s effect is subtle or inaudible, and further testing would be needed to know that it’s subtle effect is good. I’ll have much more to say about this in the article with some citations for the claims I’m making.
Now what is my subjective experience? Subjectively i round it makes a big difference in sound. It’s hard to tell the improvement in the bass from the improvement everywhere else. To account for this I manually improve the bass using PEQ and establish room curves that the auto eq follows. That way the only correction is applied to the mid and high frequencies. I can easily toggle Dirac on and off (as I can with ARC, YPAO, and Audyssey. I find the benefit in the bass is almost always better, but i find that most of these can be bettered by me manually. Dirac being an exception. In the mid and high frequencies my experience has been that Dirac is the only one I found to be innocuous or an improvement. The rest were a mixed bag that I inevitably turned off (I have only used ARC on subs so I don’t know what it’s like). With poor speakers in poor rooms, often massive linear distortions or the response are the most audible detriment. In these cases all of them seem to make things better because they make the linear response more listenable.
But...I can’t help but go back to the fact that those are flawed speakers and rooms, that eq isn’t the solution. Yes it made it more listenable but it still was a bad system. Room correction can’t turn a bad speaker into the equivalent of a good one.
Why do I still use and like room correction? Well room curve and ease of use. It’s handy to have something that can correct the bass quickly and easily. Room curve, I like different curves sometimes. Different recordings may merit a different tonal balance (circle of confusion) and it can be a handy way to do that. Sure tone controls can too, but I like this approach. I also have no room correction other than in the bass on my system right now, so I obviously go back and forth myself.
