Auto EQ With/Without Phase Correction

[AudiocRaver]

It is a well-documented fact that we as humans are quite insensitive to phase shift, at least with fairly simple test tones from a single speaker, and that a HUGE amount of phase shift is necessary before we CAN hear it. To be clear. Here is a quote excerpted from an article on the topic..

"All things being equal, and if one has the option, of course get the phase correct - at least at the one point in space where it can be done!! However, this presents problems for two-eared listeners in multiple seats in reflective rooms (solve this one and a Nobel prize awaits). It is indeed fortunate that humans are so unresponsive to this effect because, if we could hear phase shift, we would go absolutely nuts in everyday life."

For now, I am not going to comment on the silly "we have two ears" remark. We'll have fun with that one in another thread!

For the purpose of this thread, I ask: Is speaker EQ (auto or manual), WITH phase correction, such as Dirac Live, better than that WITHOUT phase correction? And of course if one assumes that Toole's remarks as quoted above are the end-all rule on the subject, then phase correction would make no - or very little - difference. Here's the thing, though: Many make that very assumption, and it is only conditionally correct. Toole is first to say so.

Phase response with beings that have two ears, like us (Floyd said so himself!) have the ability to locate sound sources in 3D space, and with pretty good accuracy. In a 2-channel room, we are able to locate sounds in an immersive soundstage with great accuracy as well.

So here is the question: How sensitive are we as humans to phase differences between signals in the perception of a soundstage? And how does that affect the question of AutoEQ with/without Phase Correction?

 

[Sonnie]

Can we do a blind AB test to see? JUST KIDDING of course, although that might be fun.

I have a hard time grasping phase correction as it relates to equalization, so this should be interesting.

 

[AudiocRaver]

A blind AB test! YES,! We have a miniDSP nanoAVR-DL and a miniDSP nanoAVR-HD. They can be connected in sequence with HDMI. The DL runs Dirac Live. The HD can run filters created by REW. Both accommodate custom target curves (the filters for the HD effectively get their target curve from REW).. The JD will correct with minimum phase filters. The DL will correct with Dirac Live's impulse/phase correction.

Blind AB, here we come!

 

[Tony V.]

It is my understanding that EQ IS in its own way phase correction. You effect the phase when you make adjustments that are different on separate channels. This is one of the reasons that I came up with the thought in my other post about Adding EQ to a system (particularly different frequency boosts/cuts on each channel) may not be the first plan of attack and room acoustics should be addressed first.

Using an EQ that uses a wide Q (so adjusting several frequencies at a time) that are fairly deep say more than 4db that dont copy the other channel can change the phase in that frequency range.

 

[AudiocRaver]

Correct. And if the two speakers are nicely matched and the room is symmetrical, so the same EQ can be applied to both channels, then the phase response that accompanies the EQ will match L and R and the two channels should stay "phase locked" so to speak.

But if they are not matched or there are other variables and the two sides do not match, then the EQ will not be the same and neither will the phase with minimum phase filters. So soundstage and imaging (SS&I) can be affected negatively. A phase-corrected auto eQ approach will take those factors into account and apply correction in such a way that the phase and frequency response are both improved, and SS&I are improved with them.

 

[Sonnie]

A phase-corrected auto eQ approach

Dirac Live

 

[AudiocRaver]

Certainly MY choice.

 

[tesseract]

Blind AB, here we come!

I'd love to take part in that!

 

[tesseract]

Analog EQ'ing can and will skew phase. DSP correction helps offset this mostly undesirable trait. About the only time it is welcome is to correct the speaker, which is another boon of DPS, especially systems like DEQX, which takes 2 measurements, one of the speaker near-field, then one of the room response.

Getting phase right seems to be a focus on some of the better sounding speakers I have experienced and I feel is an important aspect if creating high fidelity sound.

 

[Lanion]

I've been educating myself on this lately and liked this example of EQ effect on phase. Extrapolating from this its clear that the more speakers there are the harder this gets!

 

[Matthew J Poes]

I know this is an old thread, but I thought it would be nice to comment and revive.

I know that Toole, Geddes, and others actually disagree with the notion that any EQ should be used for room correction above a room's transition frequency. They have spoken negatively on numerous occasions of the notion that any eq can usefully be applied above the transition frequency and into the stochastic region.

I believe the comments that some of these folks have made really apply to minimum phase filters more than FIR based such as Dirac's Mixed phase (Of which I was a beta tester). I strongly believe that minimum phase filters cannot usefully be used for room correction within the stochastic region, not only does it cause problems with the speakers phase response which I have found to create problems with the soundstaging, but it also tends to do more harm than good to the response. That isn't to say that careful and judicious use of PEQ for a speakers anechoic response correction isn't good, I think that can be done effectively, but I think the average user cannot do this (and would do more harm than good).

I think that minimum phase filters are fine to be used below the transition frequency where the room behaves in a minimum phase steady state manner anyway. At that point there are a number of benefits to using PEQ over FIR, since they are a far more efficient filter (FIR uses a lot of taps at lf's), no preringing, etc. I would encourage careful and correct use of minimum phase filters in these bass ranges.

The tricky area is the transition region between the minimum phase steady state zone dominated by sparse room modes and the stochastic region dominated by the speakers direct sound. That transition region has high modal density and makes a mess of the midbass flatness, but in many small rooms with lots of damping sound does not behave as minimum phase in this region and I worry that minimum phase filters may again do more harm than good.

I was very impressed with Dirac as a beta tester. At the time I had Focal Utopia Towers as well as my recently completed Gedlee Abbey's. I tested both of them with DIRAC and was floored at how much better it made both speakers sound. What most impressed me was not just how much better the speakers measured, but how much better they imaged. The soundstage was far more palpable. I'm a believer in that kind of room correction. Since then I have played around with FIR filter tools and applied nothing but phase correction to my speakers and found a similar improvement in the soundstaging. I didn't find this change to be subtle either and believe strongly that I would be able to tell a difference in an ABX test (Doesn't everyone, and we are wrong so often I know). This very well could be due to mismatches in the speakers, even small ones. I also wonder if the room could be playing a roll in this as well.

That's my opinion.

 

[AudiocRaver]

I know this is an old thread, but I thought it would be nice to comment and revive.

I know that Toole, Geddes, and others actually disagree with the notion that any EQ should be used for room correction above a room's transition frequency. They have spoken negatively on numerous occasions of the notion that any eq can usefully be applied above the transition frequency and into the stochastic region.

I believe the comments that some of these folks have made really apply to minimum phase filters more than FIR based such as Dirac's Mixed phase (Of which I was a beta tester). I strongly believe that minimum phase filters cannot usefully be used for room correction within the stochastic region, not only does it cause problems with the speakers phase response which I have found to create problems with the soundstaging, but it also tends to do more harm than good to the response. That isn't to say that careful and judicious use of PEQ for a speakers anechoic response correction isn't good, I think that can be done effectively, but I think the average user cannot do this (and would do more harm than good).

I think that minimum phase filters are fine to be used below the transition frequency where the room behaves in a minimum phase steady state manner anyway. At that point there are a number of benefits to using PEQ over FIR, since they are a far more efficient filter (FIR uses a lot of taps at lf's), no preringing, etc. I would encourage careful and correct use of minimum phase filters in these bass ranges.

The tricky area is the transition region between the minimum phase steady state zone dominated by sparse room modes and the stochastic region dominated by the speakers direct sound. That transition region has high modal density and makes a mess of the midbass flatness, but in many small rooms with lots of damping sound does not behave as minimum phase in this region and I worry that minimum phase filters may again do more harm than good.

I was very impressed with Dirac as a beta tester. At the time I had Focal Utopia Towers as well as my recently completed Gedlee Abbey's. I tested both of them with DIRAC and was floored at how much better it made both speakers sound. What most impressed me was not just how much better the speakers measured, but how much better they imaged. The soundstage was far more palpable. I'm a believer in that kind of room correction. Since then I have played around with FIR filter tools and applied nothing but phase correction to my speakers and found a similar improvement in the soundstaging. I didn't find this change to be subtle either and believe strongly that I would be able to tell a difference in an ABX test (Doesn't everyone, and we are wrong so often I know). This very well could be due to mismatches in the speakers, even small ones. I also wonder if the room could be playing a roll in this as well.

That's my opinion.

Well put!

I have played a little with applying minimum phase filters at frequencies above transition, where the speakers and placement were already giving well-matched results, and been pleased with the sound. This is consistent with your findings, I believe. I agree that trying to use minimum phase filters at those frequencies, in general, is not wise, and is probably the cause of consternation for those who have done so. Mixed-phase filters (Dirac LIve) do far better. My BEST hand-tuning setups, including treatment, have ALWAYS been improved by Dirac LIve! Soundstage and imaging went from good to great.

 

[Matthew J Poes]

Well put!

I have played a little with applying minimum phase filters at frequencies above transition, where the speakers and placement were already giving well-matched results, and been pleased with the sound. This is consistent with your findings, I believe. I agree that trying to use minimum phase filters at those frequencies, in general, is not wise, and is probably the cause of consternation for those who have done so. Mixed-phase filters (Dirac LIve) do far better. My BEST hand-tuning setups, including treatment, have ALWAYS been improved by Dirac LIve! Soundstage and imaging went from good to great.

Thank you!

My only negative experience with DIRAC has been with the lite (dead flat is a bad curve in my opinion) version or a bad set of measurements leading to bad corrections. I use a similar approach to DIRAC with PEQ in the bass, I create a weighted average of the listening area, typically composed of 10-15 measurements or more. I’ve experimented with different numbers and find that it makes little difference if the listening area is small. For example 3-4 separate measurements at a single listening seat is adequate if that’s the only listening seat. I find that to be a good scaler, two seats could use 5-7, 3 seats could use 8-9, etc.