First try at REW - feedback requested

[Hazenhart]

Hello patient and helpful ASR members.
I have just completed my first set of REW measurements and an accompanying set of basic room correction filters. I would appreciate any feedback you have in regard to the apparent quality of the measurement itself, any issues to focus on and strategies to address them, and general feedback on my first attempt at filters.
.mdat files are attached (file type changed to .txt), as are txt files of my resulting filters.

System:

• 2 channel audio-only system
• Stand-mounted KEF R3 Metas fed from a Wiim Pro Plus driving an AB amp (soon to be replaced by a Buckeye)

Room:

• This is our primary living area, dedicated to many things other than active listening, so speaker positioning, seating, and room treatments are not variables I can do much more with. That said:
• ~260 ft2 room with 9-foot ceilings, albeit part of a somewhat complicated open floor plan that totals closer to 500 ft2
• Room is quite live and untreated, other than a large thick rug in the listening area
• Seating is effectively against the rear wall (sofa ~1 ft off), speakers are near broken, partial "pony" walls
• Speakers are ~11 feet apart, forming an isosceles triangle with the sofa ~13 feet away

Measurement:

• Taken using a UMIK-1 mounted on a proper mic stand, at the primary listening seat, vertical mic position, using the 90 degree calibration file
• Windows laptop feeding the line-in on the Wiim Pro Plus via the headphone jack.
• Level checks in REW were done using the Tone Generator and SPL Meter. I could not reach the target 75db SPL without the Tone Generator giving me a digital clipping warning. I got to ~68 db. This was basically full output from my laptop headphone jack at 90% volume on the Pro Plus.
• I took a left, a right, and a L+R measurement.

I have been listening to the R3 Metas for the past week or so, and subjectively, I think they sound fantastic. There is nothing weird or bothersome to my ear, but of course I would like to better understand their performance in my room objectively, and adjust the Wiim PEQ to optimize things where possible.
Thanks for your time and wisdom.

Measurement results (L,R) at 1/12 octave smoothing:

From here I used the EQ function to create a House Curve between 40 and 500 hz, "calculated target level from response," "matched response to target," and generated the filters.

I then entered the filter values into the Pro Plus PEQ and re-ran measurements:

 

[Keith_W]

I am not quite sure why you labelled your measurements "ART" when you are not using ART. But whatever.

First thing to see is this bizarre train of ghost copies of the main impulse spaced at almost perfect 2ms intervals. At first, I thought it might be a similar situation to this but since this is only one sweep, the only conclusion is that these are very early and very loud reflections. But it's very strange to get a series of reflections like this spaced so perfectly.

If this is real, and not a measurement artefact, it would sound really smeared for sure. Would you be able to take a photo of your speakers?

You weren't kidding when you said that your room was reflective. This is the RT60 Topt compared to the targets calculated for your room volume (500ft2, 9ft ceiling = 4500ft3) and target of 500ms. As you can see, it's really high between 200Hz - 2kHz.

Your other main issue is your lack of bass. But i'm sure you knew that already.

If you want to learn more about REW measurements, read the book in my signature. It's free.

 

[Hazenhart]

I am not quite sure why you labelled your measurements "ART" when you are not using ART. But whatever.

View attachment 529994

First thing to see is this bizarre train of ghost copies of the main impulse spaced at almost perfect 2ms intervals. At first, I thought it might be a similar situation to this but since this is only one sweep, the only conclusion is that these are very early and very loud reflections. But it's very strange to get a series of reflections like this spaced so perfectly.

If this is real, and not a measurement artefact, it would sound really smeared for sure. Would you be able to take a photo of your speakers?

View attachment 529996

You weren't kidding when you said that your room was reflective. This is the RT60 Topt compared to the targets calculated for your room volume (500ft2, 9ft ceiling = 4500ft3) and target of 500ms. As you can see, it's really high between 200Hz - 2kHz.

Your other main issue is your lack of bass. But i'm sure you knew that already.

If you want to learn more about REW measurements, read the book in my signature. It's free.

Thanks for your follow up.

ART is a reference to the amplifier, soon to be changed so I noted in the saved file.

Here is a picture taken from the listening position:

 

[Hazenhart]

I will certainly continue to read up on REW and room correction, but could you give me a quick summary of the meaning of your RT60 plot? Thanks again for your time and your patience. I am new to acoustic measurements.

 

[Keith_W]

I will certainly continue to read up on REW and room correction, but could you give me a quick summary of the meaning of your RT60 plot? Thanks again for your time and your patience. I am new to acoustic measurements.

Take a look at that book and read from page 55 onwards. That's a beautiful room BTW, I am a bit surprised that the reverb should be so high. All those early reflections in the impulse response may be caused by that sofa. Where did you place the mic? Did you remove the sofa?

 

[Hazenhart]

Take a look at that book and read from page 55 onwards. That's a beautiful room BTW, I am a bit surprised that the reverb should be so high. All those early reflections in the impulse response may be caused by that sofa. Where did you place the mic? Did you remove the sofa?

Thank you.
No furniture was moved for the measurements. I placed the mic at ear location just over the back of the sofa, using long boom on the mic stand. Sofa is basically right against (~7 inch standoff) the back wall.

I can't see your signature. Maybe it's because I'm browsing from my phone. I will check from my PC tomorrow. Thanks again.

 

[Paul7777x]

Book you say… I’ll also check it out tomorrow. I can just about use the basic REW for a quick perk up, but I’d like to know more.

 

[Hazenhart]

Assuming I appropriately manage headroom in the Pro Plus, is it reasonable or even advisable to apply a broad, low q boost to bring the bottom end up a bit? This screen cap shows filters that mostly squash peaks, but also include a couple broad ~5db boosts to bring the level up a bit.

 

[Keith_W]

Q: "Is it OK to use EQ to boost the bottom end?"

A: take a look at your distortion graph. Without getting too nerdy, your distortion graph is difficult to interpret because we don't know what SPL your measurement was taken at. Your measurement indicates 70dB, but the actual volume of the loudspeaker depends on the measurement distance. Also, the measurement is contaminated by ambient noise. So let's take a look at a known reliable measurement. This one is from Erin, and shows distortion at 86dB and 96dB:

It looks as if THD is up to -40dB at 50Hz, and -50dB at 100Hz. This rises by -10dB in the 96dB measurement, so it's quite significant. In fact, distortion appears to equal loudspeaker output at 30Hz.

Conclusion: your R3 Meta's probably don't have the headroom for you to implement a bass shelf without encountering significant distortion. Furthermore, this is assuming that your amp has enough guts to power your speakers with the EQ in place. There is a reasonable chance you may drive your amp into clipping as well.

Of course, all this depends on your listening volume and how sensitive you are to distortion. If you don't listen too loud, and you listen to the kind of music where distortion doesn't matter too much (or if you are insensitive to it as many of us are!!) you may be able to get away with it. But if you like turning it up, it won't work too well.

 

[Hazenhart]

Understood.
So for the low end, should I just focus on suppressing some of the larger peaks as I did with the first pass at filters? How aggressively should they be filtered? Is generating filters at 1/12 smoothing appropriate? Or should I be using something like psychoacoustic smoothing, resulting in one or two filters only.
Thanks again.

 

[Keith_W]

I wouldn't use any PEQ's at all. If you look at your frequency response (below) it's pretty much what you would expect, without any troublesome peaks that might ruin the experience. In this case I would just keep it simple.

What you do need are subwoofers. This system would sound a bit thin and lack authority. Whether it's a problem for you depends on what music you listen to, but IMO any system that aspires to the goal of fidelity NEEDS to be able to reproduce bass. And once you have subs, you might need your PEQ's.

The only issue with your frequency response are the two dips I highlighted. You can see they are symmetrical (appears in both left and right) and occurs at about 130Hz and 350Hz. One of them is likely SBIR and the other ... who knows. You may be able to improve the situation by pushing the speakers closer to the wall. But they are small and shallow enough that they probably wouldn't be much of a problem.

To answer your question about DSP correction for bass: you want to correct the unsmoothed summed response of L and R. As for correcting dips, this requires a bit of nuance. Yes, you can do something about it with DSP. But that is going to be a very long topic of discussion.

 

[Hazenhart]

Thanks for this. This is the insight that I find really valuable. I see peaks and dips but am unsure what is and is not considered problematic. Your assessment matches my subjective listening experience -- nothing bothersome or unpleasant -- but yes, a little thin. For the record I had very low expectations given the size of the speakers and the room, but the perceived bass is actually more satisfying than I anticipated. Yeah the sustained dip around 350 Hz was the one I was thinking about eq-ing up. No dramas though. I may try pushing the speakers closer to the wall as an easy experiment. As it is they are only ~20 cms out but there is no reason not to try. Adding subwoofer(s) is a project for the future, once the bank account recovers from this spending spree.

Thanks again for your time and attention. You keep odd hours if you're in Melbourne...

 

[staticV3]

In-room frequency response measurements should ideally be taken using the Moving Microphone Method, with both channels playing together.

Windows laptop feeding the line-in on the Wiim Pro Plus via the headphone jack.

This I find a bit suspect.
Who knows which APOs the laptop may be applying to its headphone jack, or if the laptop's internal headphone circuitry has flat frequency response.

The better method would be to open REW's Generator, choose "Save to file" at the bottom, and then export your test signal (sweep or Pink noise) as .wav

Load the Wav file into WiiM Home or any other playback software that can stream to the WiiM and you have removed variables/uncertainties from your signal chain.

For Sweep measurements, choose "Playback: From file" in the Measure dialog.

For Moving Microphone Method, set playback in WiiM Home/etc. to repeat and do the measurement while the Pink noise file loops away in the background.

 

[StephanB]

The purpose of room equalisation is not to produce a ruler flat frequency response from 20Hz to 20kHz.

It is to:
1. Compensate for room resonances in the bass region, also called room modes. We want to remove large peaks in those frequencies whilst keeping the dips (the nulls). This region of the frequency response is called "below the Shroeder frequency". In a room your size the cutoff will be around 120 to 200 Hz.
2. Above the Shroeder frequency we don't correct the many irregularities due to room reflections. If there is a gross defect in tonality, we will correct it with wide filters (low Q value)
3. There are 2 methods to measure the room: a. using frequency sweeps, measure 8 to 12 locations around the listening position and average them, or b. using the white noise generator and the RTA analyser, move the microphone slowly around the listening area and integrate the measurements. This is called "moving microphone measurement".

May I suggest that you read about these topics: room modes, Shroeder frequency, average measurements. This will lead to a better result with around 4 to 8 filters.
I hope this helps.

 

[Hazenhart]

In-room frequency response measurements should ideally be taken using the Moving Microphone Method, with both channels playing together.


This I find a bit suspect.
Who knows which APOs the laptop may be applying to its headphone jack, or if the laptop's internal headphone circuitry has flat frequency response.

The better method would be to open REW's Generator, choose "Save to file" at the bottom, and then export your test signal (sweep or Pink noise) as .wav

Load the Wav file into WiiM Home or any other playback software that can stream to the WiiM and you have removed variables/uncertainties from your signal chain.

For Sweep measurements, choose "Playback: From file" in the Measure dialog.

For Moving Microphone Method, set playback in WiiM Home/etc. to repeat and do the measurement while the Pink noise file loops away in the background.

Regarding using the direct feed from the laptop, that is what was described on the REW website:

Getting Started with REW

I do follow what you're saying with downloading the sweep file and running it natively from the streamer. That would make everything easier as well.
I just finished taking measurements with the speakers closer to the walls. This seems to have supported some additional bass in the ~80-100Hz region, and appears to have filled in some of the 350hz dip. Stereo imaging seems a bit different but still good. I will run as-is for a while. After I receive and connect my new amplifier, I will run some more measurements -- using a downloaded sweep and see about tackling the moving mic method.

Thanks.

 

[Keith_W]

Thanks again for your time and attention. You keep odd hours if you're in Melbourne...

Well I did notice that you work for Rio Tinto And that style of architecture (weatherboard) is not common in Perth, which means you are probably in the Eastern states. So you keep odd hours too. And I did also notice the shoe-free household, which usually means Asian. But these days a lot of Aussies have shoe-free households as well.

 

[Hazenhart]

Well I did notice that you work for Rio Tinto And that style of architecture (weatherboard) is not common in Perth, which means you are probably in the Eastern states. So you keep odd hours too. And I did also notice the shoe-free household, which usually means Asian. But these days a lot of Aussies have shoe-free households as well.

Rio Tinto owns the mine I work at in Utah, USA. So western states . I work with lots and lots of Aussies and take frequent calls with Brisbane based folks in my evenings, but I am American born and raised, for whatever that means these days.

 

[Hazenhart]

This I find a bit suspect.
Who knows which APOs the laptop may be applying to its headphone jack, or if the laptop's internal headphone circuitry has flat frequency response.

The better method would be to open REW's Generator, choose "Save to file" at the bottom, and then export your test signal (sweep or Pink noise) as .wav

Load the Wav file into WiiM Home or any other playback software that can stream to the WiiM and you have removed variables/uncertainties from your signal chain.

This proved to be true. After playing back and measuring the sweep from a .wav file saved to my NAS, the measured response was quite different compared to playing the sweep through the laptop headphone jack. Much more low frequency content and smoother around 9k-10k hz. (maybe confirmation bias, but this better matches my seat-of-the-pants experience as well).

I also took three measurements across the sofa, and averaged them:

The suckout around 150 hz seems to be persistent. Don't know if I hear it or not. Played around with partial port bungs a little bit, but think I prefer the ports totally open. Bass sounds good not boomy to my ear. Given the updated (more accurate) measurements, do you see anything in the low frequency region that you think needs PEQ?

Thanks for your time.

 

[StephanB]

The suckout around 150 hz seems to be persistent. Don't know if I hear it or not. Played around with partial port bungs a little bit, but think I prefer the ports totally open. Bass sounds good not boomy to my ear. Given the updated (more accurate) measurements, do you see anything in the low frequency region that you think needs PEQ?

It looks good for a full-range system in a domestic setup.

If I were to EQ it, I would

• start with the "Var smoothing" in REW,
• Target settings: full-range speaker, add your preferred house curve
• Filter Tasks: 30Hz to 12kHz, Individual Max boost: 6dB, Flatness target: 3dB, Allow narrow filters below 200Hz.

What I expect to get:

• 2 narrow peak filters to knock down the room modes at 50 and 85 or 90 Hz,
• 2 broad filters to boost the 100 to 200 Hz region,
• possibly a broad correction around 300 Hz.

Btw. I don't believe that messing with the bass ports in a bass-reflex enclosure does any good. The speaker designer has carefully calculated the port to match the volume of the enclosure and the electro-mechanical properties of the bass transducer. If you want to modify the bass region, use PEQ or a bass enhancer (exciter) for a bass-limited system.

 

[janbth]

I would measure L+R as the basis for EQ in the bass region. (Don’t forget Var smoothing!) I also think that your use of a spatial average for EQing is a good idea.