Peak and null REW reading question/explanation

[kasmira]

Wow, that looks way too amazing to even be real without acoustic treatment. (It very well may be, just, it is a great first reading)

How large is the room, what mic did you test with, and do you have any current treatment right now?

Frequency response shows us very little of the room picture though. Best bet would be looking at the waterfall - a lot of times the frequency response will look good but your mixes will still be turning up muddy or thin because of modal resonances, sbir, and the like.

We want rooms to be close to flat. A little roll off on the high end is sometimes wanted. Sometimes a little bit of boom at 30hz is wanted (depending on the client of course). However, the general consensus is to remove large peaks because those frequencies will sound louder than other ones, so we end up turning them down when we really didn't need to. This can cause some thin bass turning up in mixes. Nulls cause the opposite problem - that is, we tend to boost null areas in our mixes and then our mixes are boomy and flabby as a result. As it stands, your FR looks great, but doesn't tell us much about the way your room sounds (at all).

[Jens Eklund]

Before posting your measurement results

[Swurveman]

Originally Posted by kasmira

Wow, that looks way too amazing to even be real without acoustic treatment. (It very well may be, just, it is a great first reading)

How large is the room, what mic did you test with, and do you have any current treatment right now?

Thanks for your response Kasmira.

It's not without acoustic treatment. It's with 24 panels of absorption of OC703 4" (6" on back wall) in corners/on walls/ on the ceiling.

Originally Posted by kasmira

Frequency response shows us very little of the room picture though. Best bet would be looking at the waterfall - a lot of times the frequency response will look good but your mixes will still be turning up muddy or thin because of modal resonances, sbir, and the like.

Attached below is my waterfall. As you can see I still have modal resonances sticking out around 50 Hz, 100 Hz and 225 Hz. I really don't know how exactly they effect my mixing though. At the end of the ringing they are 40dB below the peaks. So, I'm not sure how that translates to what I'm hearing and the consequent decisions I make in front of near field monitors.

Originally Posted by kasmira

We want rooms to be close to flat. A little roll off on the high end is sometimes wanted. Sometimes a little bit of boom at 30hz is wanted (depending on the client of course). However, the general consensus is to remove large peaks because those frequencies will sound louder than other ones, so we end up turning them down when we really didn't need to. This can cause some thin bass turning up in mixes. Nulls cause the opposite problem - that is, we tend to boost null areas in our mixes and then our mixes are boomy and flabby as a result. As it stands, your FR looks great, but doesn't tell us much about the way your room sounds (at all).

Below is my 1/24 smoothing full spectrum 20-20,000 Hz reading. As you can see I have some nulls in high frequencies. I have been so absorbed with getting my bass response to be good that I have not paid a lot of attention to the higher frequencies in my room. And not a lot of focus is put on the higher frequencies here at Gearslutz. I have nulls around 700 Hz, 2K, 6K, 10K, 14K and 16K. On the other hand, there are very little resonances in those frequencies. So, I don't know how troublesome these nulls are in regards to my mixing in front of near field monitors.

I've read Flat Frequency Not about tailoring the output of monitors to mix with a deliberate +3 dB at 100Hz and -3dB at 10kHz. So, I don't know what my response would look like after that.

Thanks again for your response. Any additional information would be most helpful.

[DanDan]

I have been so absorbed with getting my bass response to be good

And you have done remarkably well. Congrats.
Quite often response is referred to 1Khz. So one might have an SPL of 75dBC at IK, then view the response relative to that.
Or you could try making a straight line fit through the spectrum of interest.
e.g. 20Hz -1Khz or whatever.

In the case of Full Range you might have two references, the LF flat line, and then say a 1K-20K with a downward slope. Typical HF roll off starting points are 2, 4, 8 K.

I am going to assume your Full Range FR is from a measurement using both speakers driven. This would explain the comb filtering. Best to measure and view each speaker separately and combine them in the software. This last piece of advice is by far the singular most common piece of typing I do. It beats me as to why there are recommendations to drive both speakers.
Two speakers or one speaker for ETF,REW etc?
DD

[Swurveman]

Originally Posted by DanDan

And you have done remarkably well. Congrats.
Quite often response is referred to 1Khz. So one might have an SPL of 75dBC at IK, then view the response relative to that.
Or you could try making a straight line fit through the spectrum of interest.
e.g. 20Hz -1Khz or whatever.

Hi Dan. Thanks for your response. I always wondered what the relative reference was for peaks and nulls. so, thanks for clarifying.

Originally Posted by DanDan

In the case of Full Range you might have two references, the LF flat line, and then say a 1K-20K with a downward slope. Typical HF roll off starting points are 2, 4, 8 K.

I am going to assume your Full Range FR is from a measurement using both speakers driven. This would explain the comb filtering. Best to measure and view each speaker separately and combine them in the software. This last piece of advice is by far the singular most common piece of typing I do. It beats me as to why there are recommendations to drive both speakers.
DD

Yep, the FR from 20-20K is both speakers driven.

I've got single speaker results. So, I'll check them out.

I was particularly happy to stumble upon your "Flat Frequency Not" thread. It is easy to get involved in the minutiae of building panels, reading theory - and all the graphs and results from software like REW- and forget that the end goal has speakers involved in the equation of how to get the best translatable mixes. So, once I make sure I'm good to go with my room, I'm going to experiment with the +3 dB at 100Hz and -3dB at 10kHz method.

Thanks again.

[DanDan]

That sloped 'translate' curve is a changing scenario.
My current opinion has become nuanced.
With your spectacularly flat LF including time response, I don't think you will need or want an LF Boost, but by all means try the classic and smaller ones.
I do believe you will still want a HF roll off but it may be worth experimenting with the parameters.
e.g. My current favourite is

(EDIT for clarity)

+1dB from 30Hz to 1K
OdB 1K
-2dB 10K
-3dB 15K
-4dB 20K

(I am Beta Testing Dirac Live, and liking it a lot)

Recently, I noticed a generalised slope of -10dB sloped from 1K-20K on the very best available headphones. (A flat line best fit to actual response measured at the dummy head equivalent of an actual eardrum)

Seems to me the response at nearfield in a very damped room should be somewhat similar to headphones.

There are other historic recommendations for HF Roll Off. e.g. -3dB Octave starting at 4 or 8Khz.

One thing is for sure, no-one seems to suggest HF boost!

DD

[Swurveman]

Originally Posted by DanDan

That sloped 'translate' curve is a changing scenario.
My current opinion has become nuanced.
With your spectacularly flat LF including time response, I don't think you will need or want an LF Boost, but by all means try the classic and smaller ones.
I do believe you will still want a HF roll off but it may be worth experimenting with the parameters.
e.g. My current favourite is
+1dB 30Hz -1K
-2dB 10K
-3dB 15K
-4dB 20K

(I am Beta Testing Dirac Live, and liking it a lot)

Recently, I noticed a generalised slope of -10dB sloped from 1K-20K on the very best available headphones. (A flat line best fit to actual response measured at the dummy head equivalent of an actual eardrum)

Seems to me the response at nearfield in a very damped room should be somewhat similar to headphones.

There are other historic recommendations for HF Roll Off. e.g. -3dB Octave starting at 4 or 8Khz.

One thing is for sure, no-one seems to suggest HF boost!

DD

Dan,

When you are referring to your cuts and boosts are you using a parametric EQ, or is Dirac Live doing it?

This is a whole new area of inquiry for me. So, sorry if I can't tell exactly what you are referring to.

[DanDan]

Dirac Live decides on the Eq filters, as a result of multiple measurements, and a Target Curve. It then generates them in software, there is no hardware. One can chose their recommended Target Curve, or tweak it, or impose whatever Curve one wishes. The Target Curve is created using Waypoints, like in DAW volume automation.

There is much activity on the Room Eq front. Take a look at Audiolense, MiniDSP, Trinnov, the OpenDRC initiative etc. etc.

Let's not forget many active monitors have sufficient Eq onboard to create a HF Roll Off.
Plus, I suspect from your graph that you already have a gentle one in action.

Check the Edit on my previous post, a bit clearer to read I hope.

DD

[Swurveman]

Originally Posted by DanDan

Dirac Live decides on the Eq filters, as a result of multiple measurements, and a Target Curve. It then generates them in software, there is no hardware. One can chose their recommended Target Curve, or tweak it, or impose whatever Curve one wishes. The Target Curve is created using Waypoints, like in DAW volume automation.

There is much activity on the Room Eq front. Take a look at Audiolense, MiniDSP, Trinnov, the OpenDRC initiative etc. etc.

Let's not forget many active monitors have sufficient Eq onboard to create a HF Roll Off.
Plus, I suspect from your graph that you already have a gentle one in action.

Check the Edit on my previous post, a bit clearer to read I hope.

DD

Thanks for the clarification about Dirac Dan.

My monitors do have onboard options to cut and boost. I'll check them out-and along with my averaged single speaker high frequency response from REW- start experimenting. Have to start referencing/comparing my speaker mixes with my headphones as well.

Thanks +2

[Swurveman]

Originally Posted by DanDan

In the case of Full Range you might have two references, the LF flat line, and then say a 1K-20K with a downward slope. Typical HF roll off starting points are 2, 4, 8 K.

I am going to assume your Full Range FR is from a measurement using both speakers driven. This would explain the comb filtering. Best to measure and view each speaker separately and combine them in the software.

Sure enough the averaged L/R 20-20Khz measurement (1/24 smoothing) is a lot better. No comb filtering etc.

Thanks again Dan.

[DanDan]

You are welcome, and congratulations on one of the best responses I have seen.

DD

[Glenn Kuras]

So, I'm not sure how that translates to what I'm hearing and the consequent decisions I make in front of near field monitors.

If your mixes translate well in other systems, meaning there are not major surprises then you are done. It really is that simple.
BTW I agree your tests are very impressive.