This is possibly one of my usual silly questions but I was wondering if there is a “correct” volume for doing measurements for room correction?

I can understand that running louder sweeps would give better signal to noise ratio. But too loud could risk driving speaker into compression and so you might end up correcting for issues that aren’t there when listening at your normal volume.

The other thing I was thinking is that I think it’s not uncommon to set target curves such that nulls are corrected by dragging other frequencies down (rather than trying to fill in the null by boosting it).

My question on that is that if you listen louder than you ran your sweeps, won’t the nulls just reappear?

Let’s say you created a nice smooth curve at 80dB by dragging peaks down to meet nulls. As you crank up your Megadeth, the nulls won’t increase in volume but everything else will. Does that mean your target curve is no longer smooth?

And there’s more, do equal loudness contours come into play when deciding on what level to sweep? I’m not sure if this one is relevant and I haven’t thought about it too much so it may well not be.

Yes , its where the volume control has no influence over the source component capability You want no adverse loading for volumes purpose. You will then achieve correct volume for doing measurements for room correction.

I’m honestly not sure what you mean but from what I can discern it seems like you’re answering a different question.

On 12/02/2026 at 6:06 PM, sir sanders zingmore said:

The other thing I was thinking is that I think it’s not uncommon to set target curves such that nulls are corrected by dragging other frequencies down (rather than trying to fill in the null by boosting it).

My question on that is that if you listen louder than you ran your sweeps, won’t the nulls just reappear?

Let’s say you created a nice smooth curve at 80dB by dragging peaks down to meet nulls. As you crank up your Megadeth, the nulls won’t increase in volume but everything else will. Does that mean your target curve is no longer smooth?

Great Qu, Trev. 👍

Surely (with your 80dB example) when you drag down peaks to meet the nulls ... the nulls aren't at zero - they're, say 10dB down from the peaks. So when you reduce the peaks by 10dB (to get a flat line) when you repeat the track at 90dB ... every frequ has gone up by 10dB - even the null - so you still have a flat line?

But I guess the only way to work out what actually happens is to experiement:

1. do a first run at 80dB

2. pull down the peaks appropriately, to get your flat line

3. then do a second run at 90dB ... and see what REW says?

4 minutes ago, andyr said:

Great Qu, Trev. 👍

Surely (with your 80dB example) when you drag down peaks to meet the nulls ... the nulls aren't at zero - they're, say 10dB down from the peaks. So when you reduce the peaks by 10dB (to get a flat line) when you repeat the track at 90dB ... every frequ has gone up by 10dB - even the null - so you still have a flat line?

Of course!

I told you it was as silly question.

And the way to confirm this is as you suggest below

(trouble is, it's almost impossible for me to run sweeps at home. I have to wait for the rare occasion when no-one else is around (everyone hates them)

4 minutes ago, andyr said:

But I guess the only way to work out what actually happens is to experiement:

1. do a first run at 80dB

2. pull down the peaks appropriately, to get your flat line

3. then do a second run at 90dB ... and see what REW says?

MM sticky question ,

Good practice ,not to add something that is not there ,

At any given volume your room will react I certain way

But at less volume there will be less room effect ,

And visa versa ,

To assess at normal listening levels ,and adjust accordingly ,

I would use that as my benchmark ,

If you have a home alone volume ,take that ,

And remember the variation

Then late night volume the same !

So I would have a benchmark ,one above and one below

That would satisfy my needs ,

Any other vol variation I would live with ,

And not my real world listening

The deqx I had ,allowed for preset curves I think 3 ,then even with those you could + or - any frequency on the run ,to adjust for bad sounding cd or lp

13 hours ago, sir sanders zingmore said:

I’m honestly not sure what you mean but from what I can discern it seems like you’re answering a different question.

You just lack awareness of what your question asked, ....that;s OK, take your time.

16 minutes ago, stereo coffee said:

You just lack awareness of what your question asked, ....that;s OK, take your time.

It's quite possible that I lack awareness of what my question asked (Indeed, my first sentence hinted at that by saying it might be one of my usual silly questions). But when I said that I didn't know what you meant, it was a request for you to please clarify your comment.

This shouldn't be an exercise in zen koans.

Your question asked "if there is a “correct” volume for doing measurements for room correction" ....That's a very high bar being set using the word "correct" and volume, in the one sentence, and entails to me certainly, it is very specific what you need to the urge of achieving "correct"... requiring of its answer.

We see @colinm1 refers to "given volume" and "less volume" but not "correct" volume

8 hours ago, colinm1 said:

At any given volume your room will react I certain way

But at less volume there will be less room effect ,

where your question asked. .

On 12/02/2026 at 8:06 PM, sir sanders zingmore said:

I can understand that running louder sweeps would give better signal to noise ratio

On 12/02/2026 at 8:06 PM, sir sanders zingmore said:

But too loud could risk driving speaker into compression and so you might end up correcting for issues that aren’t there when listening at your normal volume.

Your first quote is partially correct , but guesses that it is "louder sweeps". Achieving the "correct" volume, in fact where the volume control, has no influence over the source component, is where, as you asked "correct volume" is achieved, ...as I answered.

17 hours ago, stereo coffee said:

Yes , its where the volume control has no influence over the source component capability You want no adverse loading for volumes purpose. You will then achieve correct volume for doing measurements for room correction.

In the circumstance , relieving how specific the need of "correct" volume, the question asked. You should look for recommendations from the room measuring equipment manufacturer, particularly if they divulge any mention of a L pad, and recommended series and shunt resistances you should use.

5 hours ago, stereo coffee said:

You should look for recommendations from the room measuring equipment manufacturer, particularly if they divulge any mention of a L pad, and recommended series and shunt resistances you should use.

I’m measuring with a miniDSP UMIK-1 But I don’t understand why I’d ask that. Perhaps you could ask them on my behalf and let me know what they say? Thanks 👍

Dirac seems to suggest at least 20dB above the noise floor of the room.

On 12/2/2026 at 5:36 PM, sir sanders zingmore said:

This is possibly one of my usual silly questions

Hi Trevor,

I will have a go at this one, and will do my best to avoid koans, philosophy or word puzzles.

On 12/2/2026 at 5:36 PM, sir sanders zingmore said:

but I was wondering if there is a “correct” volume for doing measurements for room correction?

75 dB(C) with pink noise. Don't ask me to justify why it has to be exactly this and nothing else. It doesn't. But this is commonly used by tools like REW and Audyssey. It's a 'safe bet'.

On 12/2/2026 at 5:36 PM, sir sanders zingmore said:

I can understand that running louder sweeps would give better signal to noise ratio. But too loud could risk driving speaker into compression and so you might end up correcting for issues that aren’t there when listening at your normal volume.

That's right. And if your room is anything like mine, you might easily find yourself exciting room components like window panes or frames, bookshelves touching the wall, etc. These are much better fixed at the source.

In fact, I have played a sine wave quite loudly and manually swept it through the bass range in order to find any such audible rattling, then walk over to the rattling object while playing the exciting frequency and try to eliminate it there and then.

On 12/2/2026 at 5:36 PM, sir sanders zingmore said:

The other thing I was thinking is that I think it’s not uncommon to set target curves such that nulls are corrected by dragging other frequencies down (rather than trying to fill in the null by boosting it).

That doesn't sound like a good idea. My understanding of best practice (given the limitations of the general technique) is to ignore nulls. They are usually so narrow that you can't hear them (as 'something missing') anyway.

Better to establish your average level through the bass range, by eye is fine, then (1) trim any sharp peaks, and (2) apply broad smooth trims and boosts to broad lifts and swales in the measured response.

On 12/2/2026 at 5:36 PM, sir sanders zingmore said:

My question on that is that if you listen louder than you ran your sweeps, won’t the nulls just reappear?

The nulls aren't 100% nulls anyway. Yes, sitting half way between the side walls is common, and can result in a lateral mode null, but other room modes are still in play. Using the method I described above will make this question irrelevant, IMHO.

On 12/2/2026 at 5:36 PM, sir sanders zingmore said:

Let’s say you created a nice smooth curve at 80dB by dragging peaks down to meet nulls. As you crank up your Megadeth, the nulls won’t increase in volume but everything else will. Does that mean your target curve is no longer smooth?

That sort of null is insanely narrow and not audibly important.

On 12/2/2026 at 5:36 PM, sir sanders zingmore said:

And there’s more, do equal loudness contours come into play when deciding on what level to sweep? I’m not sure if this one is relevant and I haven’t thought about it too much so it may well not be.

No, ignore them. It's doesn't matter what you hear at the test SPL, say 75 dB. You are making a 'reference' correction which will be right when played back at sufficiently loud SPL that the subjective perception of loss of bass at low volumes is irrelevant. I think we don't need that sort of adjustment once the SPL reaches the low 80s dB, or more.

If you want to make that sort of adjustment for quieter listening, do it separately and save it as a separate setting to apply when desired.

cheers,

Grant

On 16/2/2026 at 1:08 PM, Grant Slack said:

Hi Trevor,

I will have a go at this one, and will do my best to avoid koans, philosophy or word puzzles.

75 dB(C) with pink noise. Don't ask me to justify why it has to be exactly this and nothing else. It doesn't. But this is commonly used by tools like REW and Audyssey. It's a 'safe bet'.

That's right. And if your room is anything like mine, you might easily find yourself exciting room components like window panes or frames, bookshelves touching the wall, etc. These are much better fixed at the source.

In fact, I have played a sine wave quite loudly and manually swept it through the bass range in order to find any such audible rattling, then walk over to the rattling object while playing the exciting frequency and try to eliminate it there and then.

That doesn't sound like a good idea. My understanding of best practice (given the limitations of the general technique) is to ignore nulls. They are usually so narrow that you can't hear them (as 'something missing') anyway.

Better to establish your average level through the bass range, by eye is fine, then (1) trim any sharp peaks, and (2) apply broad smooth trims and boosts to broad lifts and swales in the measured response.

The nulls aren't 100% nulls anyway. Yes, sitting half way between the side walls is common, and can result in a lateral mode null, but other room modes are still in play. Using the method I described above will make this question irrelevant, IMHO.

That sort of null is insanely narrow and not audibly important.

No, ignore them. It's doesn't matter what you hear at the test SPL, say 75 dB. You are making a 'reference' correction which will be right when played back at sufficiently loud SPL that the subjective perception of loss of bass at low volumes is irrelevant. I think we don't need that sort of adjustment once the SPL reaches the low 80s dB, or more.

Thanks for your comprehensive (and comprehensible) answer.

On 16/2/2026 at 1:08 PM, Grant Slack said:

If you want to make that sort of adjustment for quieter listening, do it separately and save it as a separate setting to apply when desired.

cheers,

Grant

Regarding this, I’ve finally hooked up

An RME ADI dac that I bought ages ago. It has a dynamic loudness feature which is really really nice!

I had one of these units too and it was a comprehensive and easy tool to use with plenty of program memory storage - a better power supply made a noticeable difference - I wasn't all able to compensate for the rather "hard" sound unfortunately, but understand since then that there are ways to do this (preamps, line trannie, etc) - I do regret selling it.

Like most 'programmable dacs', they don't tolerate trying to correct close peaks/dips (as mentioned above, many dips can be ignored) but apart from that, the little thing has a far superior sound than any 'mini des' units (IMO here!)

1 hour ago, HdB said:

I had one of these units too and it was a comprehensive and easy tool to use with plenty of program memory storage - a better power supply made a noticeable difference - I wasn't all able to compensate for the rather "hard" sound unfortunately, but understand since then that there are ways to do this (preamps, line trannie, etc) - I do regret selling it.

Like most 'programmable dacs', they don't tolerate trying to correct close peaks/dips (as mentioned above, many dips can be ignored) but apart from that, the little thing has a far superior sound than any 'mini des' units (IMO here!)

At this stage I’m not actually using the RME for room correction (that’s run in a miniDSP SHD unit). I’m only using it for preamp and loudness duties

As for the “hard” sound and “improved”power supplies…. Nah I’m not your man when it comes to that. To my ears it’s completely transparent as a dac and preamp, no more to be done.

Sad Bastard 3 should ask Sad bastards 1 and 2.. Just saying.

On 12/2/2026 at 6:06 PM, sir sanders zingmore said:

Let’s say you created a nice smooth curve at 80dB by dragging peaks down to meet nulls. As you crank up your Megadeth, the nulls won’t increase in volume but everything else will. Does that mean your target curve is no longer smooth?

And there’s more, do equal loudness contours come into play when deciding on what level to sweep? I’m not sure if this one is relevant and I haven’t thought about it too much so it may well not be.

Basically no .... but it depends, on precisely what you're thinking of.

If you mean (literal) null. It's a true "cancellation", which does (at least at one very specific frequency) go to zero..... and it goes to zero no matter what you have done with EQ (because it cannot be resolved with EQ).

The confusion arrises by looking at smoothed data (and/or data which has a finite resolution on the frequency axis, say 1 data point every 2 Hz, or what have you).
I hope that made some sense (these things are tricky to communicate).

If it's the frequencies that are located on the "shoulder" of the "null". Then no, these will increase in line with the overall SPL increase, so you don't get the issue you're talking about.

If the "dip" is caused by something else..... then it might suffer from the issue you've described (but likely not).

On 16/2/2026 at 1:08 PM, Grant Slack said:

That sort of null is insanely narrow and not audibly important.

^^

Edited March 5Mar 5 by davewantsmoore

I'd want to know how the speaker measured out of the room, first and in increments of 5 Db from 80dB to 100dB.

Knowing this would make a baseline of reference and would most likely find the point where speaker cone compression or distortion becomes a detraction from the speaker output.

Another measurement that can be easily overlooked is of the background noise in room, as in create a REW spectrogram with no speaker/s operating. Often (through experience) there will be something of constant magnitude ie, fridge compressor, PC or amp fan or snoring pet that can give polluted in room readings., with the speakers. It is quite surprising sometimes how loud a fridge can be !

Beyond these points of reference, whatever else is a room interaction, meaning combing, first reflections, standing waves or any other room based anomaly in effect caused by the speaker, showing the in room REW Measurements, at the listening position.

This does not really answer directly "what is the correct volume for room correction ?" but goes a long way to an introduction into what may be causing anomalies within the in room REW Spectrogram, and what action to rectify may come next.

Being able to interpret the REW Spectrogram is an actual thing :) but having a cross reference to baseline measurements does make it easier.