Room EQ, do's and dont's

[andreasmaaan]

If the errors in the frequency response are minimum phase to begin with, IIR filters (being also minimum-phase) will correct the phase at the same time as they correct the amplitude response.

 

[andreasmaaan]

...and in the case of loudspeakers and rooms, the errors tend to be (more or less) minimum-phase.

 

[RayDunzl]

If the errors in the frequency response are minimum phase to begin with, IIR filters (being also minimum-phase) will correct the phase at the same time as they correct the amplitude response.

...and in the case of loudspeakers and rooms, the errors tend to be (more or less) minimum-phase.

Does that include nulls when the waves from two sources (stereo) cancel each other at some point in the room?

That's what I have in the 48Hz area at the listening position with the current setup.

 

[mitchco]

A few approaches:

https://yabb.jriver.com/interact/index.php/topic,87538.0.html

https://www.reddit.com/r/audiophile/wiki/room-correction

https://www.dropbox.com/s/10xdhh83jokzbxv/REW_rePhase_tuto.pdf?dl=0

https://www.diyaudio.com/forums/ful...ectrical-loudspeaker-correction-networks.html

Some guidance on minimum phase: https://www.roomeqwizard.com/help/help_en-GB/html/minimumphase.html

I use commercial DSP software like Audiolense and Acourate.

IMO, there is no better audible upgrade for the time and effort put in than audio DSP.

Have fun!

 

[andreasmaaan]

Does that include nulls when the waves from two sources (stereo) cancel each other at some point in the room?

That's what I have in the 48Hz area at the listening position with the current setup.

That kind of null will not be fixable with min-phase EQ, no. How are you dealing with it atm?

 

[MRC01]

Room treatment (tube traps, bass traps, acoustic foam, etc.) could take a bit out of a null like that, by weakening room modes. My suggestion, and general philosophy, is to do as much as possible with room treatment and use EQ only to fix whatever issues remain.

 

[Sir Sanders Zingmore]

I use commercial DSP software like Audiolense and Acourate.

Does Audiolense work on a mac?

 

[RayDunzl]

That kind of null will not be fixable with min-phase EQ, no. How are you dealing with it atm?

Since the 48Hz area is not localizable, I notched that range out of one channel as an experiment.

https://www.audiosciencereview.com/.../spl-for-critical-listening.6891/#post-153182

then

https://www.audiosciencereview.com/...or-critical-listening.6891/page-3#post-153284

 

[mitchco]

Does Audiolense work on a mac?

No, both are Windows programs, but I do believe some members run the Designer under Mac's virtual machine, and then use a Mac native convolver to host the FIR correction filters for music playback. Bernt on the Audiolense forum can confirm...

 

[Krunok]

Exactly! Phase distortion is related to the slope of the filter. Use gentle slopes (numerically small Q, wide frequency range) minimizes it.

With FIR filters you can independetly correct phase of your filter so high Qs can be used in the 20-300Hz region assuming your measurement is precise enough. With IIR filers you can't correct phase but I think you can still use high Qs as I don't think phase shift is really that audible in bass region.

Question: When the room boosts or cuts certain frequencies, it also induces phase distortion. Could it be possible that the phase distortion of a parametric EQ opposes this, so the filter corrects phase as well as amplitude? If so, a parametric EQ's "side effect" of shifting phase would actually be beneficial!

Room can cause signals from left and right speakers to be out of phase on certain frequencies. You will see such things mainfest as a modestly narrow dip which can be more than 10dB deep with 1/12 smoothing. In such cases you can modify phase of one loudspeaker until they are both in phase or you can lower amplitude of the probleamtic speaker and let the other handle the bass response. In the 20-300Hz region you should aim for flat response when both speakers are playing, after 300Hz each speaker should have linear response so they should be treated independently.

 

[Krunok]

That kind of null will not be fixable with min-phase EQ, no. How are you dealing with it atm?

I had similar problem and I solved it with shifting the phase of the speaker with problematic positioning by 90 degs at that frequency.

 

[Krunok]

As I said, my experince with room EQ is limited, as is my knowledge, yet I cannot stress more the importance of accurate measurement as a basis for the room EQ. For that reason I will quote this text as my learning curve would be faster if I have had it on hand when I was starting:

Averaged measurements

The idea behind averaged measurements in REW is to get rid of over-corrections which
validity is limited to the exact place where the microphone has been setup in case of a
single measurement. There is a wide corpus of technical information pertaining to
interference of sound waves and various physical phenomena which affect the perception
of the sound within very short distances from one point to the other. In order to make a
correction which takes into consideration the main interaction between the loudspeakers
and the room, it seems appropriate to make multiple measurements and to average them in
order to extract the significant signal from the averages and to avoid the unnecessary
noise.

This has been remarkably illustrated by Floyd E. Toole (@Floyd Toole ), Ph.D. then Vice President
Acoustical Engineering Harman International Industries, Inc. in a series of lectures named
“The Science of Audio”.

The procedure is simple: you take the same equipment which would be necessary to make
a single measurement and you repeat the procedure of measurement at the edges of a
parallelepiped, located around the heads of the listeners at the listening position. You
number the measures L1..L9 for the left channels in case you make 9 measurements, and
R1..R9 for the right channels (8 measurements at the edges and a 9th measurement, at the
center of the parallelepiped).

Hopefully tomorrow, if I catch some time, I will make a comparison between average of 9 sine sweeps vs RTA measurement with mic moving along these 9 points. If done properly I expect the results to be pretty consistent, but let's see..

 

[mitchco]

Some of the procedures in post 24 show how to average measurements as well, including this tutorial https://www.dropbox.com/s/10xdhh83jokzbxv/REW_rePhase_tuto.pdf?dl=0 which describes the same procedure above.

However, some of the top DSP software packages take a slightly different approach so as to achieve similar results without having to average. The issue is most software simply inverts the magnitude response and is indeed constrained to a single point. Audiolense (which also does multi-seat averaging) and Acourate use a slight different analysis technique to avoid overcorrection:

If one studies the chart closely, one can see the envelope response is following the peaks but not the dips. Note the marker at just over 100 Hz., one can see a very narrow dip, but the new analysis does not follow that narrow dip. This new frequency response can be considered as an upper envelop of the original magnitudes. The spectral envelope is used as the basis for further calculations. It even supersedes the need for multi-location measurements in a typical small listening environment like a living room. I have verified this in my book where I took 14 REW measurements across a 6ft x 2ft grid area (where my couch is) and the low frequency response is virtually the same at all 14 points, based on a single analysis measurement used to calculate the correction filter.

Audiolense, pretty much the same, even though it does have an option for multi-seat measurements...

 

[Krunok]

Some of the procedures in post 24 show how to average measurements as well, including this tutorial https://www.dropbox.com/s/10xdhh83jokzbxv/REW_rePhase_tuto.pdf?dl=0 which describes the same procedure above.

However, some of the top DSP software packages take a slightly different approach so as to achieve similar results without having to average. The issue is most software simply inverts the magnitude response and is indeed constrained to a single point. Audiolense (which also does multi-seat averaging) and Acourate use a slight different analysis technique to avoid overcorrection:

View attachment 22692

If one studies the chart closely, one can see the envelope response is following the peaks but not the dips. Note the marker at just over 100 Hz., one can see a very narrow dip, but the new analysis does not follow that narrow dip. This new frequency response can be considered as an upper envelop of the original magnitudes. The spectral envelope is used as the basis for further calculations. It even supersedes the need for multi-location measurements in a typical small listening environment like a living room. I have verified this in my book where I took 14 REW measurements across a 6ft x 2ft grid area (where my couch is) and the low frequency response is virtually the same at all 14 points, based on a single analysis measurement used to calculate the correction filter.

Audiolense, pretty much the same, even though it does have an option for multi-seat measurements...

Are you saying that a single sweep when "enveloped" like you have shown above can replace multipoint averaged measurement yielding the same results?

 

[mitchco]

Are you saying that a single sweep when "enveloped" like you have shown above can replace multipoint averaged measurement yielding the same results?

Yes. But it is specific to Acourate and Audiolense as the envelope/psychoacoustic filtering is proprietary for both software packages...

 

[Krunok]

Yes. But it is specific to Acourate and Audiolense as the envelope/psychoacoustic filtering is proprietary for both software packages...

Aha. So I can't take a single sine sweep from the center of LP and hope to get the same result if I "envelope" it? It's probably a little bit more complicated than that..

 

[mitchco]

Yes, proprietary psychoacoustic analysis filtering...

 

[Olli]

No, both are Windows programs, but I do believe some members run the Designer under Mac's virtual machine, and then use a Mac native convolver to host the FIR correction filters for music playback. Bernt on the Audiolense forum can confirm...

I am using a Mac System. You can design the filters with AL and a Windows Laptop and then upload them into either JRiver MAC version or Roon‘s convolution engine.

 

[mitchco]

Your system is really impressive and obviously extremely well tuned! Still, I can't find info how you measured the repsones you've shown, so can you please clarify?

I have seen many discussions on this topic, some in favor of measuring bth speakers at the same time to evaluate response up to 300Hz, some saying it's not neccessary. I would very much appreciate if you can find the link and post it , preferrably in the "Room EQ, do's and dont's thread".

See "Set up to take measurements" in this article provides the details. Scroll down a bit... Plus, I try and follow these guidelines as best as possible, except mine are floor standers and using an updated target frequency response. It has worked out pretty well, as I fall within most if not all of the guidelines, but with a crappy room ratio and at the top end of the recommended room decay time at 600ms broadband - so room is lively.

I think @JohnPM the author of REW (awesome software and thank you for the latest fix!) maybe be able to answer the question about measuring both speakers at the same time...

Edit: PS. I use the default values in REW (500ms right window), sweep mode from 10 Hz to 24 kHz and 1/12 oct smoothing. I also use Acourate's microphone alignment tool for final mic positioning. I use a multi-client ASIO driver with the Lynx Hilo, so I can run REW and loopback the sweep signal into JRiver's ASIO line input and passes it through the convolution engine with the correction filters, digital XO, etc., in the circuit...

 

[MRC01]

I made the same measurement, this time using the built-in mics on my Zoom H4 recorder. Neither seems accurate above 600 Hz, and I'm less concerned about those higher freqs because the room doesn't affect them much and the speakers have pretty flat response anyway.
Here's how the NT1As (grey) and the H4 mics (black) record the same thing.

It's a much bigger difference than I expected. Mics matter!
Here's the EQed response (red) based on the H4 measurements. In the range where the measurements seem reliable, this stays within about 2 dB of my target curve.

The EQ to achieve this has nothing narrower than 1/2 octave nor more than 6 dB.
Here's the group delay L and R, with minimum phase in black. If I understand this graph correctly, the response is nicely linear. I wonder if that is typical of planar magnetics.

I really should get a calibrated microphone made for this kind of work. They're not expensive.