Studio monitor calibration - some questions

[ColdComfort]

Hi All-

I'm new to REW, and have some questions. I'm helping a friend build out a mixing area around his DAW. He has a home studio that he periodically records bands performing in, and wants to improve his mixing by using studio monitors rather than headphones. I'd like for us to calibrate the monitors using REW, so characterizing them with a calibration mic with the ultimate goal of a monitor-appropriate target curve enabled by a PEQ programmed into his monitors' interface.

By way of background, we're using a Dayton UMM-6 USB mic with 90 degree and 0 degree calibration files, and the monitors are true 2.0 monitors, no sub. The monitors are driven by an interface that we have calibrated via a loopback.

I've been devouring REW guides, forum posts, and videos, and have some questions.

1. I intend to take monitor measurements at the MLP in triplicate, of the individual and then combined (L+R) speakers. For measuring each speaker individually, I understand that the standard is to point an omnidirectional calibration mic on-axis (0 deg) directly at the speaker. This supposedly reduces the impact of room modes and reflections. But since I will also measure the speakers during combined (L+R) output, how should I orient the mic? On-axis but pointed between them? 90 degrees with the mic pointed to the ceiling? Neither strike me as good, as I'm fundamentally changing the mic orientation with regards to the speakers for the combined L+R measurement. The best option seems to be to take all measurements, even individual speakers, with the mic oriented at 90 degrees while accepting that the room will be convoluted into the measurements (the room is heavily treated as-is, though).
2. Is 'flat' the best target curve for a PEQ for studio mixing? I understand that there are house target curves that are more pleasant to listen to such as the Harman curve, but is there anything like an ANSI/ISO/Toole standard target curve for mixing monitors?
3. Related to the above, is there a best practices for the filter task settings when generating a PEQ to achieve a target curve for monitors? If so, what are they?
4. Also related to 2, does the SPL smoothing impact the PEQ generated? If so, I assume I should use 1/3 octave smoothing?

I know there's a whole other part of this: room treatments. The room is heavily treated already, as it is a studio space. The efficacy of those treatments is TBD So please set that aside. I expect in reality this will go: calibrate monitors, realize there's room issues, work on room issues, then re-calibrate monitors. For now, our MVP is calibrated monitors.

Thanks!

edit: bonus question. I read that sine sweeps when played through multiple monitors simultaneously can cause comb filtering issues, and that periodic pink noise is better. Similar to question 1, I hate to characterize the speakers individually with sweep tones and then together with pink noise. Should I use pink noise for all measurements? Am I making things too complicated by taking measurements of the combined L+R output? Not doing that feels like flying a bit blind, as that's the operational mode they're going to be in when used. Unfortunately it introduces these additional variables of mic orientation and test tones.

 

[moedra]

Response to #1:
• If the intent is to reduce the impact of room modes and reflections, the way to go about it is to treat the problem areas with absorption and diffusion. Pointing a mic in any direction will only measure the room slightly differently, giving you slightly different views of the same underlying problems. The room modes and reflections will still be there and they will affect the listener regardless unless you treat them.
• Measure each speaker independently at 90° to the floor and ceiling. Keep the mic in the same place, untouched, for each speaker. There is no need to measure the two speakers together.

Response to #2:
• With REW, use a linear target. You can add this into the EQ window by dialing in a room curve with the following settings:
LF Rise Start (Hz): 500
LF Rise End (Hz): 20 or lower
LF Rise Slope (dB/octave): 0.4
HF Fall Start (Hz): 500
HF Fall Slope (dB/octave): 0.4
Be sure to use the same target level for both speakers.

Response to #3:
• Set the target level such that the target rests across the lower dips of the response, left or right, that has the lowest valleys. It's alright if a low frequency null (or two) drops below the target. Nulls like that can't be corrected with EQ anyway.
• Use a range from just below the lowest measured frequency up to 20kHz.
• Individual and Overall Max Boosts: 3dB
• Flatness Target: 1 or 2
• Allow narrow filters below 200 Hz

Response to #4:
• Yes it does
• Use 1/6 or 1/12 smoothing for the match function to give you adequate detail. I wish there were a 1/8 smoothing option...

Response to bonus question:
• This is why, in my response to question #1, I mentioned that there is no need to measure both speakers together. It won't give you the response you need in any way whatsoever.
• Yes, you are over-complicating the matter. Use sine sweep. Measure left speaker, measure right speaker, make filter.

 

[ColdComfort]

Thank you for the thorough responses @moedra.

If the intent is to reduce the impact of room modes and reflections, the way to go about it is to treat the problem areas with absorption and diffusion.

No, the intent is to flatten the response of these monitors. The room is well treated but has never been characterized. As I flatten the monitors’ response with PEQ, I will also be hunting for room modal resonances or nulls to ensure I’m actually flattening the speakers’ response and not fighting the room. After this first pass on the monitors, I’ll assess the room, make changes as necessary, and then return to the monitors for a final tuning. I’ve read about the MMM method for room evaluation, so that’s my planned method as of right now.

Response to #2:

I’m curious, what is the origin of your recommended tent pole at 500 Hz gentle downwards slope for the target curve rather than a totally flat response? I’m ignorant of anything outside of the “monitors should be flat” mantra.

Response to #3:
• Set the target level such that the target rests across the lower dips of the response, left or right, that has the lowest valleys. It's alright if a low frequency null (or two) drops below the target. Nulls like that can't be corrected with EQ anyway.

Thanks, nice to hear that what we stumbled on is actually the preferred method. Although when doing a trial run with a non-calibration condenser mic, REW consistently popped up with a warning that setting the target as you describe would result in low SPL b/c most of the measured response fell above the target. I’m paraphrasing, but it’s safe to ignore that? From the documentation, it does appear that REW’s author wants users to be very clear on the fact that EQ should not be used to adjust overall levels. Perhaps the warning is another manifestation of that?

 

[moedra]

Thank you for the thorough responses @moedra.



No, the intent is to flatten the response of these monitors. The room is well treated but has never been characterized. As I flatten the monitors’ response with PEQ, I will also be hunting for room modal resonances or nulls to ensure I’m actually flattening the speakers’ response and not fighting the room. After this first pass on the monitors, I’ll assess the room, make changes as necessary, and then return to the monitors for a final tuning. I’ve read about the MMM method for room evaluation, so that’s my planned method as of right now.



I’m curious, what is the origin of your recommended tent pole at 500 Hz gentle downwards slope for the target curve rather than a totally flat response? I’m ignorant of anything outside of the “monitors should be flat” mantra.



Thanks, nice to hear that what we stumbled on is actually the preferred method. Although when doing a trial run with a non-calibration condenser mic, REW consistently popped up with a warning that setting the target as you describe would result in low SPL b/c most of the measured response fell above the target. I’m paraphrasing, but it’s safe to ignore that? From the documentation, it does appear that REW’s author wants users to be very clear on the fact that EQ should not be used to adjust overall levels. Perhaps the warning is another manifestation of that?

What do you mean by "characterized"?

I think you're missing the point slightly, unless I'm completely misunderstanding you. Nulls caused by room mode cancellation cannot be corrected with equalization, so if any exist in your measured responses due to boundary interference, the only way to flatten them out is to treat the problematic areas with proper absorption and diffusion. You need to get rid of the reflections causing those nulls, and no amount of corrective EQ can fix that. You can't flatten the speaker response unless any nulls you find are treated. After that, you can completely correct your speaker response.

---

As for MMM, I have never done that, nor have I ever felt like I needed to. Can't be of any help there. All I've ever needed to do is measure each speaker with a sine sweep, analyze the data to track down nulls, treat the room to absorb reflections and reduce decay times, and remeasure.

In my book, "flat" means "linear". Flat doesn't mean absolutely horizontal. A horizontal target will yield a filter that sounds slightly too bright and off balance with a weak-sounding low end. It always sounds more natural to tilt the target slightly toward pink noise. The settings I gave you produce a result that is more acoustically neutral, and matches the tone from Dirac Live's default target pretty closely. Dirac's default target is also sloped like that.

---

Yes, REW throws a warning. You can ignore it and it will still work fine as long as the top of the measurement (20kHz) crosses the target or at least gets close to it. Remember to ignore deep dips below 400Hz or so, as these are nulls that will not be correctible by any filter.

 

[ColdComfort]

I think you're missing the point slightly, unless I'm completely misunderstanding you. Nulls caused by room mode cancellation cannot be corrected with equalization, so if any exist in your measured responses due to boundary interference, the only way to flatten them out is to treat the problematic areas with proper absorption and diffusion. You need to get rid of the reflections causing those nulls, and no amount of corrective EQ can fix that. You can't flatten the speaker response unless any nulls you find are treated. After that, you can completely correct your speaker response.

We're on the same page here, thanks again for the responses.

In case this thread is helpful to others, while digging into some of moedra's answers, I found these resources:

htt.ps://audiophilestyle.com/forums/topic/56359-help-needed-with-choosing-the-optimal-dirac-curve/
htt.ps://www.stereophile.com/content/katzs-corner-episode-17-perfecting-perfection
htt.ps://www.youtube.com/watch?v=rs96rzVWjYg&t=1863s
htt.ps://www.discogs.com/release/6364268-Various-Best-Of-Chesky-Jazz-And-More-Audiophile-Tests-Volume-2

my account is not allowed to post links due to its low post count, so remove the [.] in the https