Would You Guys Analyze This For Me?

Whoa, how did I miss that? Good catch Dan.

Barry, it's either pink or not. If you had to EQ it to make it flat, the RTA is void. I suggest you take another shot with real pink noise, reduced bandwidth and let the chips fall where they may.

Bill

Ruairi,
Just to clarify, I am NOT in favor of EQing control room monitors, where the product is a recording that is created as a result of referencing the mix through a monitor/room integration. I do however draw a distinction in live audio where the sound coming from the monitors IS the product.
In this case I was pointing out that Barry had an RTA of which about 90% of the data was unnecessary to determine the monitor/room interaction and that no changes to the geometry or surface treatments of the room would actually have much of an affect above 500hz. Having a reduced bandwidth RTA also provides more detail where it is needed to make a judgement.
Here is a small story that might clarify a bit. Most of today's near field monitors are far more linear than the speakers we used back in the day, especially mains speakers which were just huge boxes with re-purposed PA drivers and crossovers. When I installed a mains system in a new control room, I knew the speakers were not as linear as they could be with a bit of White EQ applied. However, EQing IN the control room was just a bad idea. EQing them in an anechoic chamber was needed but I didn't have one. However, I had corn fields and pastures. Placing a speaker in a large grass covered field is a pretty good approximation of an anechoic chamber, especially for us po folks.
So I would run power out fifty yards or so into a field and lay the speakers on their backs with an omni mic (generally an AKG 451 which was my most linear microphone) about three feet up and cranked pink noise through the speakers. I used a Barclay Badap frequency analyzer to pull the curve. My limit with the White was about 3 db, as beyond that, the draw on the power amp was too much and ate up too much headroom. I would tweak the monitors from about 80hz on up to about 15Khz and then install the whole shebang in the control room. Sometimes I would also mark an X curve on the White (black grease pencil for flat and red for X) for those times when we were mixing for theater release.
I say all this to point out that this is NOT EQing monitors to correct for speaker/room interaction. It is EQing to improve the natural frequency response of the speaker or to apply a specific profile for a specific job. Thinking back on this, I believe I was taught this technique by Bob Heil. Thanks Bob!
Bill

-bill_mueller

Great explanation and a great story Bill.  I'm with you on all points.

In SpectraFoo the Spectragraph display has the option to show white noise or pink noise as "flat". I'm pretty sure that's what Barry was referring to in his original post.

Jim's got it right.  I didn't EQ anything.  I used SpectraFoo's display setting to show the pink noise as "flat".

Barry

Constant energy.

Cool, sorry for the diversion.
As it stands my opinion on the graph remains, not bad at all. 
But I'll bet that software is probably capable of some much more informative viewing. 

I suggest measuring one speaker/room response at at time. 
When two are running there will be horrendous comb filtering at the mic unless the two paths to tweeter are identical. 

So one speaker. Spectrum with no fractional octave smoothing. 20-300Hz
You will very likely see nulls caused by reflections from boundaries. 
By moving the mic or speaker you can see the null frequency shift thus identifying the  causal boundary. 
e.g. Move speaker toward the front wall, if a null frequency shifts upwards, you have it on the hook.
Now move the speakers sideways and up/down. If the frequency does not shift you have clearly isolated the Front Wall BIR. 
Save or take a screengrab. Try the other speaker. Differences in the area around the speaker will cause different spectra. 
These nulls cannot be Eq'ed but sometimes one can soften them by taking advantage of peaks caused by modes or position. 
Thomas Barefoot's Wall Bounce Calculator 2D is interesting. 

Eq in the crossover or DSP controller is commonly used to optimise the speaker. 
This can be extended outside the speaker to the Soffit Compensation Filter. 
I also welcome target curves for purpose, e.g. the X Curve.  A 'TV' curve. Or a HF roll off which has historically worked well for mix translation on speakers, still works well for me. 
But then there are  iBuds.....

Eq located within or closely associated with the speaker rather than the room is a given, but rooms 'load' a speaker, particularly at LF. This can be even be visible in impedance changes. 

I have just spent some time with a friend client, trying to optimise his room/speaker response. 
His new upgraded speakers have a peak at 246Hz which I simply cannot find a cause for. 
His older speakers do not show the same peak. Similarly the 2-3K is diminished and the HF accentuated on the new, the old looking pretty flat and smooth. Upgrade?
Selling both sets of speakers has been mentioned.....
Well he is open to that but for now,  he want's to persist with the new, they have their attractions and come highly recommended. 

REW's Excess Group Delay graph indicates the peak is amenable to Eq, and the HF obviously is. 
So I have recommended an Eq in the DAW for the moment. If it works we will make it so using the a MiniDSP.com  module.

DD

Cool, sorry for the diversion.
As it stands my opinion on the graph remains, not bad at all. 
But I'll bet that software is probably capable of some much more informative viewing. 
I suggest measuring one speaker/room response at at time. 
When two are running there will be horrendous comb filtering at the mic unless the two paths to tweeter are identical. 
So one speaker. Spectrum with no fractional octave smoothing. 20-300Hz
You will very likely see nulls caused by reflections from boundaries. 
By moving the mic or speaker you can see the null frequency shift thus identifying the  causal boundary. 
e.g. Move speaker toward the front wall, if a null frequency shifts upwards, you have it on the hook.
Now move the speakers sideways and up/down. If the frequency does not shift you have clearly isolated the Front Wall BIR. 
Save or take a screengrab. Try the other speaker. Differences in the area around the speaker will cause different spectra. 
These nulls cannot be Eq'ed but sometimes one can soften them by taking advantage of peaks caused by modes or position. 
Thomas Barefoot's Wall Bounce Calculator 2D is interesting. 
Eq in the crossover or DSP controller is commonly used to optimise the speaker. 
This can be extended outside the speaker to the Soffit Compensation Filter. 
I also welcome target curves for purpose, e.g. the X Curve.  A 'TV' curve. Or a HF roll off which has historically worked well for mix translation on speakers, still works well for me. 
But then there are  iBuds.....
Eq located within or closely associated with the speaker rather than the room is a given, but rooms 'load' a speaker, particularly at LF. This can be even be visible in impedance changes. 
I have just spent some time with a friend client, trying to optimise his room/speaker response. 
His new upgraded speakers have a peak at 246Hz which I simply cannot find a cause for. 
His older speakers do not show the same peak. Similarly the 2-3K is diminished and the HF accentuated on the new, the old looking pretty flat and smooth. Upgrade?
Selling both sets of speakers has been mentioned.....
Well he is open to that but for now,  he want's to persist with the new, they have their attractions and come highly recommended. 
REW's Excess Group Delay graph indicates the peak is amenable to Eq, and the HF obviously is. 
So I have recommended an Eq in the DAW for the moment. If it works we will make it so using the a MiniDSP.com  module.

DD

-dandan

Thanks Dandan.  You, and everyone else, have given me so much to think about and to do/try.

I'll update this as soon as I can.

My head hurts. 

Great stuff.

Thanks Dandan.  You, and everyone else, have given me so much to think about and to do/try.I'll update this as soon as I can.

-barry_hufker

Barry -- you need to do a proper transfer-function measurement, which shows phase as well as amplitude. Without the phase information, you don't know whether the response anomalies are caused by a reflection or other time-domain weirdness. 

An RTA amplitude-only display is often less than helpful.

I don't know if SpectraFoo offers transfer-function measurements. Smaart does.

-a

Thanks Andy.  I'll work on this.

Barry