New to Understanding Room Correction & Measuring

MalVeauX

MalVeauX

Audioholic
Hello all,

Thank you for taking a moment to take a look at this. I appreciate your time. I am new to measuring and trying to slowly creep into a little better understanding of how measuring works, what it represents and how to use it for room correction and/or a little equalization to correct for the room. I'm practicing in my office on a 2 channel setup, near field without a sub-woofer (I have one, but am not using it for this exercise yet, one step at a time) simply to get some experience doing this, using software and understanding it all. Any guidance or correction would be greatly appreciated! I have already read a few articles and tutorials and watched a few videos to at least try to have a better vocabulary and understand a bit more about what's going on, but I am not an engineer nor do I have any background in sound science other than being a life long audio-enthusiast (without golden ears though). I mainly do critical listening with headphones, have a stereo setup in my office and then my home theater. Most of this exercise is to help me understand what I'm doing when it comes to room correction and equalization when it comes to my home theater as I will be building a new home soon and will have a dedicated room for audio so I'm preparing for that (though it will not be a totally custom room, likely a garage converted to a finished room and then I can treat the inside but likely will not be able to control dimensions beyond a blueprint).

TL;DR: I'm trying to practice measuring my office and doing room correction equalization to understand how to read the graphs, understand what's happening in the room and how to potentially approach correcting the room or the equalization to get a flatter response as an exercise in an imperfect room with imperfect placement.

I have started with:
Polk Monitor 30 II Books x 2 via AudioSource AMP 100 & MaverickAudio TubeMagic D2
MiniDSP UMIK-1 with both calibration files
REW Software
Equilizer APO Software
My office!

Now, my office is by no means treated or even optimal for critical audio application. It's my work office and the placement of everything is limited, so I'm simply trying to use it as a good exercise to measure things, understand what it means, what the room is doing with the frequencies, overall performance in the room, what can be done to correct a little bit for the room, knowing its not optimal. It's also a small room and positioning is not optimal due to the layout of the office. Again, this is an exercise for me to simply understand and practice doing this. Any help understanding or correcting any interpretation would be excellent!

My office is 930 cubic feet in volume, 9'6" x 12'6" x 7'10" with a door and one window. The position of the speakers is not optimal due to layout of office. It's a near field setup basically in the right front corner of the room away from the door and window where my desk is.

Forgive this graph or any values, I'm trying to use the room simulator in REW to get an idea of things after I input values with my measurements:

RoomSim_EQREW.jpg


I used the UMIK-1 to do some measurements in my listening position. I was reading that for room correction it's potentially better to use a 90 degree position with the microphone with the 90 degree calibration file applied in REW to get an idea of the room perhaps a bit better than directly trying to measure the speakers themselves. This is something I'm not entirely clear on. I was trying to read more about best way to measure the output of the speaker frequency response and the room response in general as a whole and whether it's ideal to use the 0 degree or directly at the speaker position or the 90 degree (towards the ceiling) position of the microphone for best examples of what's truly going on and how to interpret that. Any help there is great!

My initial measurements are the following in the room (I used REW and the UMIK1 to set my SPL to 75db via Pink Noise for this):

Office_Pre_Measure.jpg


Trying to understand and interpret this (above) graph, it looks like my speakers are not going to have any meaningful output until around 60hz and rolls off sharply down to 40hz where it blends into ambient room level noise. I have some kind of trough around 110hz but it's only around 3 or 4 db so I assume this is normal and nothing to fuss with? However, there's a weird peak at 150hz give or take a bit, that really spikes high. Is this room reflection or room gain or something like this? The other frequencies going from there to around 1khz all seem to be on the higher side of where reference was and a 5db and 10db difference is pretty significant, so I assume this is something I would want to correct to be flatter in response. From 3khz on, it seems to be appropriate and fairly flat with minimal fluctuation, no more than 2db or so. So the question is, to correct this (for theory sake, knowing this room is not optimal for this) what kind of treatment ideas would one start to consider? Otherwise, beyond equalization, but if I were to think of doing anything physically to the room based on this graph, what approach would one take?

Next, I used this measurement and I applied a House Curve (not something completely flat, the Harman's approach I believe it is?) and then used EQ in REW to generate filters to try and correct for some of the big amplitude changes in the frequencies above. I saved this EQ set of filters and applied them via Equalizer APO. I then measured the response again in REW. The following two graphs are the new room frequency response after equalization and the final image are both overlapped to see the difference.

Office_Post_Measure.jpg


Office_Pre&Post_Measure.jpg


So to interpret the above changes to the new frequency response, it seems that the measurements and equalization made some positive changes towards a flat response in the room, so that would represent the room correction I think? The trough around 110hz is still present but is only around 5db in difference, as I read it was wise not to try to fix troughs by increasing their levels at that frequency (I'm not sure if that's true). The major peaks however are now gone completely and that entire stretch of frequencies from 150hz to 600hz is now much flatter and much closer to reference, with a few peaks that are noteworthy. But overall the response fluctuates much less and is closer to a flat response than it was at least. I am curious what may cause that large spike around 150hz however that was corrected significantly. Also, I'm curious, what would be the ideal way to approach that trough at 110hz, or is it fine like this? Not asking from a sound perception point of view, but more the idea of what I can be looking at in terms of the room to correct for, or equalization to address (if possible) without harming the overall response. Any guidance there would be appreciated!

If you got this far, thank you for taking the time to read through and any comments or help to understand the above would be greatly appreciated!

I would like to use anything learned from this process to better understand how to approach the same thing in a larger room with more control for a home theater (building new home, will have new dedicated room). Any tips on how to use the above to identify room correction or treatment or best ways to equalize, etc, would be awesome!

Very best,
 
WaynePflughaupt

WaynePflughaupt

Audioholic Samurai
I used the UMIK-1 to do some measurements in my listening position. I was reading that for room correction it's potentially better to use a 90 degree position with the microphone with the 90 degree calibration file applied in REW to get an idea of the room perhaps a bit better than directly trying to measure the speakers themselves. This is something I'm not entirely clear on. I was trying to read more about best way to measure the output of the speaker frequency response and the room response in general as a whole and whether it's ideal to use the 0 degree or directly at the speaker position or the 90 degree (towards the ceiling) position of the microphone for best examples of what's truly going on and how to interpret that. Any help there is great!
I prefer 0-degree, but I’m a voice crying in the wilderness on that. I’ve seen cases where 90-degree exaggerated high frequency readings. I expect at least one cause for this would be a ceiling that doesn’t absorb or diffuse the highs (such as you’d have with a textured sheetrock ceiling), like perhaps a smooth, varnished wood ceiling.

That said, in most cases either orientation is fine, as long as you make sure you’re using the appropriate calibration file for the orientation.


My initial measurements are the following in the room (I used REW and the UMIK1 to set my SPL to 75db via Pink Noise for this):

View attachment 36780

Trying to understand and interpret this (above) graph, it looks like my speakers are not going to have any meaningful output until around 60hz and rolls off sharply down to 40hz where it blends into ambient room level noise. I have some kind of trough around 110hz but it's only around 3 or 4 db so I assume this is normal and nothing to fuss with? However, there's a weird peak at 150hz give or take a bit, that really spikes high. Is this room reflection or room gain or something like this? The other frequencies going from there to around 1khz all seem to be on the higher side of where reference was and a 5db and 10db difference is pretty significant, so I assume this is something I would want to correct to be flatter in response. From 3khz on, it seems to be appropriate and fairly flat with minimal fluctuation, no more than 2db or so. So the question is, to correct this (for theory sake, knowing this room is not optimal for this) what kind of treatment ideas would one start to consider? Otherwise, beyond equalization, but if I were to think of doing anything physically to the room based on this graph, what approach would one take?
You have a room mode at 150 Hz. Easily corrected with EQ, as you have seen.

The issue at 110 Hz is not a trough, but a null. Same at 250 and 300 Hz. Nulls are narrow and deep. Troughs are wide, such as the issue between 1-3 kHz (although that is a very large example of the type). Null’s can’t be equalized, as your graphs show. Troughs can be equalized.

The trough between 500-900 Hz should respond to equalization. Above about 300-400 Hz you’re EQing the speaker’s direct sound. Below that point you’re equalizing the effects of the room.

Regards,
Wayne A. Pflughaupt
 
MalVeauX

MalVeauX

Audioholic
I prefer 0-degree, but I’m a voice crying in the wilderness on that. I’ve seen cases where 90-degree exaggerated high frequency readings. I expect at least one cause for this would be a ceiling that doesn’t absorb or diffuse the highs (such as you’d have with a textured sheetrock ceiling), like perhaps a smooth, varnished wood ceiling.

That said, in most cases either orientation is fine, as long as you make sure you’re using the appropriate calibration file for the orientation.

You have a room mode at 150 Hz. Easily corrected with EQ, as you have seen.

The issue at 110 Hz is not a trough, but a null. Same at 250 and 300 Hz. Nulls are narrow and deep. Troughs are wide, such as the issue between 1-3 kHz (although that is a very large example of the type). Null’s can’t be equalized, as your graphs show. Troughs can be equalized.

The trough between 500-900 Hz should respond to equalization. Above about 300-400 Hz you’re EQing the speaker’s direct sound. Below that point you’re equalizing the effects of the room.

Regards,
Wayne A. Pflughaupt
Thanks Wayne,

That's very helpful, I appreciate it. Excellent point about the ranges, 300~400Hz and beyond is the speaker directly while below is the room, that helps a lot to understand what I'm seeing and why. And thanks for helping with the null vs the trough, that helps to understand that a lot more (and why EQ didn't fix one).

I will experiment next with adding a subwoofer to this, to see about how to ideally blend it into this via a graph and measurement too. I was reading that to do this on could start by simply running pink noise to the sub on its own just as I did with the speakers above, and raise its levels to 75db so its the same volume as the speakers and then work on the frequency response and blending from there. My question is whether I can use the UMIK-1 to measure SPL accurately, same as I did the speakers, or if the low frequencies are more difficult for the UMIK-1 to capture. It has a calibration file with sensitivity so it is supposed to be able to be used as a fairly accurate SPL monitor, I think. But I'm far from certain. I need to find a good source on how to blend a sub into the range but for laymen.

Very best,
 
WaynePflughaupt

WaynePflughaupt

Audioholic Samurai
REW’s SPL meter will work as well with subs as mains. The only possible issue is that since pink noise is a random signal, the meter will probably tend to jump around more with the low frequencies vs. what you saw for the main speakers.

No reason why your proposed method won’t work. However, most people seem to prefer their subs to be several dB hotter (higher SPL) than the main speakers. So feel free to increase the level if you aren’t happy with the way things sound at 75 / 75 dB.

Regards,
Wayne A. Pflughaupt
 
MalVeauX

MalVeauX

Audioholic
REW’s SPL meter will work as well with subs as mains. The only possible issue is that since pink noise is a random signal, the meter will probably tend to jump around more with the low frequencies vs. what you saw for the main speakers.

No reason why your proposed method won’t work. However, most people seem to prefer their subs to be several dB hotter (higher SPL) than the main speakers. So feel free to increase the level if you aren’t happy with the way things sound at 75 / 75 dB.

Regards,
Wayne A. Pflughaupt
Thanks,

I admit I too like the subs hotter for movies, but not really for music. I would like to try to generate a pretty flat response if possible just to learn to do it, the process, and then alter them to taste after but just to have the experience doing it in the first place.

I just need to find a good source on how to best blend the sub to the mains so that it achieves that. I'm not sure how that looks in terms of a graph for each individual speaker and then blended as one response, such as the graphs above to know what to look for and how to approach correcting it.

Very best,
 
MalVeauX

MalVeauX

Audioholic
Heya,

So I attempted to measure my room response with the sub on its own, then the bookshelves again, then blended them based on where I saw the cross over physically on the graph between them when measured. To start, I used the SPL meter in REW and Pink Noise with sub-cal selected to bring the SPL of the sub to 75db approximately with the crossover turned up completely (120hz in this case) with no audio from the other speakers. Then I did another measurement of just the bookshelves again (already calibrated and EQ'd from before). With the graph between the two, I looked for the obvious crossover point on the frequency response, it happened to be around 80hz so I just went with that (but I like this, unless this is incorrect methodology) as it confirms why often 80hz is a good rule of thumb to begin with).

The bookshelves are Polk Monitor 30 II's and the sub is an old Polk PSW505 12". My source is the same as above, Maverick TubeMagic D2 which outputs to an Audiosource AMP100 stereo amplifier. I have the passthrough L/R RCA channels going to the sub-woofer's two input channels and set to on. The sub then receives the full signal passthrough regardless of the volume attenuation of the amplifier's volume knob. I set the sub's high pass filter to maximum (120hz, Phase 0 degree). I increase volume on the sub itself to 75db based on the UMIK-1 in REW measuring it. So I'm doing this manually without a digital processor to do crossover like an AVR.

Office_Sub+Books_Crossover_Pre.jpg


Is this the correct way to find real crossover for the room and speaker combinations? I just looked at the fall off of the bookshelf speakers and they're still ok to around 65hz or so, then rapidly drop into ambient. The sub response is picking up around 20hz but steps up near 25~26hz and is good from there forward, holds well without any major room gain and no major nulls or dips and then begins to roll off at 80hz and drops fast from there. So I selected 80hz as the best likely spot for crossover, based on this. I admit I'm a little annoyed that it seems to be 80hz (well, just a bit over 80hz really) since that's such a common value for this. But maybe this confirms why its such a common value? Again, my high pass filter was 120hz (fully open) on the sub to avoid limiting it there. And it didn't seem to have much output after 80hz anwyays. This sub seems to do well from 25hz to 80hz as its effective range.

Next, I did another measure of the sub and the books together as one to see the response as follows:

Office_Sub+Books_Blended.jpg


It seems to be ok. Not as flat as I would like, but for a practice run, this seems fruitful. I can see that if my reference was 75db, the sub picks up around 25~26hz and is there and only wavers +/- 3db or so and the bookshelf blends from there, running what appears to be a little hotter at 80hz and above, being about +4 and +5db louder after that point. But no major nulls and no major room modes that I can identify and no major deltas, everything is within +/-5db of what reference would be expected to be I think?

So now the full frequency range again:

Office_Sub+Books_Combined_EQon.jpg


So now the sub & bookshelves are blended around 80hz with the crossover after both were measured to 75db SPL to be reference. This sweep is 20hz to 20khz. I don't think I see any major nulls or room modes at any point. I have a little bit of a loss of SPL at 40hz but it's within 5db of reference and I have a gain of about 4db at 90hz or so, still close to reference. Looks like I do have one little spike around 220~240hz that goes just a bit beyond +5 db from reference. But from 300hz to 10khz it's pretty flat I think, not going very far from 75db either direction, within a maximum of 3~4db overall change at any point at worst it seems.

That is, if I read the above correctly. While the curve itself is pretty jagged, smoothed at 1/6th, from a change in SPL from reference, its minimal, within +/-5db for the most part. So should I call this a relative success in terms of measuring and correcting for a relatively flat response in this particular room?

Thanks for any input.

Very best,
 
Last edited:
MalVeauX

MalVeauX

Audioholic
Heya,

Ok, I did it again just to do the process again and try to even things out with placement and correction to get a better response. Here's the results:

Here's the sub frequencies blending the sub and the bookshelves together from 10hz to 200hz:

Sub & Bookshelf Blended Flat.jpg

I think overall I got it as flat as it can be without having better placement in the room of everything and treatment, but I don't see any major issues with room mode, nulls, etc. I think it's close.

Next, I did a full sweep of the full frequency range from 20hz to 20khz to see how the rest responded after measuring and correcting the curves:

Sub & Bookshelf All Freq Flat.jpg


I think that looks pretty good, it looks a lot like my House Curve does, which is what it was modeled towards as a goal (Harman's House Curve concepts).

So I have metrics to look at, I'm going to listen to it a while now and see how it actually sounds.

Very best,
 
A

Alastair

Audiophyte
Heya,

Ok, I did it again just to do the process again and try to even things out with placement and correction to get a better response. Here's the results:

Here's the sub frequencies blending the sub and the bookshelves together from 10hz to 200hz:

View attachment 36893
I think overall I got it as flat as it can be without having better placement in the room of everything and treatment, but I don't see any major issues with room mode, nulls, etc. I think it's close.

Next, I did a full sweep of the full frequency range from 20hz to 20khz to see how the rest responded after measuring and correcting the curves:

View attachment 36892

I think that looks pretty good, it looks a lot like my House Curve does, which is what it was modeled towards as a goal (Harman's House Curve concepts).

So I have metrics to look at, I'm going to listen to it a while now and see how it actually sounds.

Very best,
Looks like you had a lot of success on paper. I am wondering how it sounds when listening normally?

I am in the process in sorting out my sub and mains integration for both stereo music and 5.1 film playback.
 
Pogre

Pogre

Audioholic Warlord
Heya,

Ok, I did it again just to do the process again and try to even things out with placement and correction to get a better response. Here's the results:

Here's the sub frequencies blending the sub and the bookshelves together from 10hz to 200hz:

View attachment 36893
I think overall I got it as flat as it can be without having better placement in the room of everything and treatment, but I don't see any major issues with room mode, nulls, etc. I think it's close.

Next, I did a full sweep of the full frequency range from 20hz to 20khz to see how the rest responded after measuring and correcting the curves:

View attachment 36892

I think that looks pretty good, it looks a lot like my House Curve does, which is what it was modeled towards as a goal (Harman's House Curve concepts).

So I have metrics to look at, I'm going to listen to it a while now and see how it actually sounds.

Very best,
What kind of smoothing are you applying to the 10 - 200 hz chart? I'm noticing your full range sweep seems to show some squiggling down there.
 
MalVeauX

MalVeauX

Audioholic
What kind of smoothing are you applying to the 10 - 200 hz chart? I'm noticing your full range sweep seems to show some squiggling down there.
I had done 1/6th smoothing based on some reading, but later realized I should use zero smoothing from 10hz to 400hz approximately to really get an idea of things in the room. From there up, a little smoothing is ok, but 1/6th is probably the limit, overall now I don't use smoothing hardly at all with my latest measurements to see every bit. Smoothing just "looks" nicer but isn't meaningful really.

Very best,
 
MalVeauX

MalVeauX

Audioholic
Looks like you had a lot of success on paper. I am wondering how it sounds when listening normally?

I am in the process in sorting out my sub and mains integration for both stereo music and 5.1 film playback.
The sound is fantastic, balanced. When I lower the overall volume output, it's amazing how dynamic it still sounds even at very low volume, well below listening level. When you use measurements and move a sub around to get ideal output from the room response, it makes a huge difference to flatten out the modes and fill in the nulls. That flat response sounds great even at low level volume. And at listening volume, it's just fantastic and full.

Very best,
 
WaynePflughaupt

WaynePflughaupt

Audioholic Samurai
You should rescale the graphs for 45-105 dB vertical, which is more-or-less the standard. That huge scale you’re using would make the Grand Canyon look like a Kansas prairie...

Regards,
Wayne A. Pflughaupt
 
Pogre

Pogre

Audioholic Warlord
You should rescale the graphs for 45-105 dB vertical, which is more-or-less the standard. That huge scale you’re using would make the Grand Canyon look like a Kansas prairie...

Regards,
Wayne A. Pflughaupt
Yup. That's my point. Between that giant scale and the 1/6 smoothing it looks like an attempt to hide issues instead of fix them. As it is, I don't think that chart is useful at all.
 
MalVeauX

MalVeauX

Audioholic
You should rescale the graphs for 45-105 dB vertical, which is more-or-less the standard. That huge scale you’re using would make the Grand Canyon look like a Kansas prairie...

Regards,
Wayne A. Pflughaupt
Agreed, thanks. I did this in another thread with the same advice. I simply didn't update this thread. That's why I mad this thread though, to learn.

Very best,
 
MalVeauX

MalVeauX

Audioholic
Yup. That's my point. Between that giant scale and the 1/6 smoothing it looks like an attempt to hide issues instead of fix them. As it is, I don't think that chart is useful at all.
Agreed, the graph is not useful at this point; now that I've learned more and understand why (but later than the posting of said graph).

Nothing to hide, but thanks. I made this thread to learn how to do this. In another thread I got better advice at the time, so I continued in that thread and went to two graphs, one for bass (20~400hz) with no smoothing, and the other for 400hz~20khz with no smoothing to light smoothing closer to 1/12th at the most, but frankly I'm ok with zero smoothing now, for the purpose of not losing information that suggests things that could be corrected. Again, nothing to hide. The context of this thread was to learn, which was a while back now and I've moved on to new threads and didn't bother updating this one.

Thanks!

Very best,
 

newsletter
  • RBHsound.com
  • BlueJeansCable.com
  • SVS Sound Subwoofers
  • Experience the Martin Logan Montis
Top