What dB level is considered "flat"?

[KMA]

I’m having troubling understanding what I’m thinking is probably a basic concept to most.

Have a look at the attached pic.

What dB value would constitute a flat response (or “the middle”)?
In other words, how high are my peaks and how low are my nulls?
Can I change the DB display to the left to indicate what flat would be?

Forgive the newb questions. Thanks for your help.

 

[Kal Rubinson]

Flat, as an absolute, is obvious and not generally achievable. OTOH, compare the left side of your graph with the right side. It is less flat.

 

[skid00]

'Flat' is a never-occuring-in-real-life frequency response where *all* frequencies have the same measured dB level. IE, 20 Hz is at 90 dB, 100 Hz is at 90 dB, 15,000 Hz is at 90 dB.

*However*, speakers in a typical room sound too bright with that response. Most people prefer the low bass frequencies to be about 10 dB louder than the highest frequencies. This varies a bit depending on age, hearing degradation, music being played, etc..

Note that in your image, the bass peaks and troughs are mostly due to room modes, which boost and cut specific frequencies due to echoes reinforcing or damping loudness *at the exact spot the microphone is measuring from*.

 

[DanDan]

Imagine a 'best fit' flat line with equal amounts of energy above and below it. What can we say about it? Well, it is tilted downwards towards HF, normal and desirable. The excursions above and below the flat line are say +/- 5dB at LF but much tighter higher up. Then guessing, the holes in the bass response are typically caused by cancellations, both resonant and non resonant. I would try placing those speakers much closer to the Front Wall, even almost touching it.

 

[AJ Soundfield]

Can I change the DB display to the left to indicate what flat would be?

Any of the horizontal lines on your graph would be "flat". You wouldn't want to see that measured in room. Given your 2db division scaling, your "speaker" response >600 Hz, at the microphone, is pretty "flat". Below 600 Hz, the room/speaker is what you are seeing...and that's not too bad either, given the scale. Unsure how you grabbed that image, but the bottom frequency scale is missing, though guessable.
A very key missing bit of info, is where the mic is relative to speakers, in room. A single graph like this (presumably L & R channels) is just one small snapshot of information.

Forgive the newb questions. Thanks for your help.

We were all newbs at some point. Welcome.

cheers

 

[KMA]

Thanks for your responses, everyone!

I understand that “flat” isn’t achievable. I guess I was just looking for a way to better understand the size of these peaks and nulls.

For instance, I have a peak at 400 and a null right next to it at 300. The difference between them is approximately 7dB. So, do I have a 3dB peak and 4dB null, or is it more like a 5dB peak and a 2dB null? In other words, it seems to me that I would need a rough approximation of a “flat” response to have any idea of how big these resonances or cancellations are.

 

[KMA]

Imagine a 'best fit' flat line with equal amounts of energy above and below it.

That sounds reasonable to me. This particular measurement was made with an SPL calibrated to 75dB, so I wondered if there's any value in considering 75dB to be the non-real-world “flat” response?

Then guessing, the holes in the bass response are typically caused by cancellations, both resonant and non resonant. I would try placing those speakers much closer to the Front Wall, even almost touching it.

My Dynaudios are rear-ported, so I’m already wary of placing them too close to the wall (thereby increasing the bass response), but I will absolutely try them closer and re-measure. This is complicated by the fact that there’s a bay-window on the front wall.

And I should mention that I’m starting to assemble and install my acoustic treatments today, so these measurements are the “before” picture, taken in a bare, totally untreated room. I’m looking forward to comparing these with the measurements post treatment.

 

[KMA]

A very key missing bit of info, is where the mic is relative to speakers, in room.

The mic has been placed so that it forms an equilateral triangle with a point on each speaker between the tweeter and the woofer (my Dynaudios are 2-way).

My mic/mix position is based on the 38% rule - a theoretical compromise between peaks and nulls.

 

[AJ Soundfield]

I understand that “flat” isn’t achievable.

It is achievable. It's also highly misguided, thus not recommended. The very first thing you should decide, is whether you want to make your 2 ears happy, or your 2 eyes and the microphone? Absolutely not the same thing.
Based on you answer, I will either extricate or continue. Your last comment above was an important clue.

For instance, I have a peak at 400 and a null right next to it at 300. The difference between them is approximately 7dB. So, do I have a 3dB peak and 4dB null, or is it more like a 5dB peak and a 2dB null? In other words, it seems to me that I would need a rough approximation of a “flat” response to have any idea of how big these resonances or cancellations are.

If those represented what was actually at your 2 ears, the peak at 400 would be audible, the null at 300 wouldn't.
A single snapshot is better than none, but doesn't tell the story.
Hah, you answered as I typed

 

[AJ Soundfield]

The mic has been placed so that it forms an equilateral triangle with a point on each speaker between the tweeter and the woofer (my Dynaudios are 2-way).

My mic/mix position is based on the 38% rule - a theoretical compromise between peaks and nulls.

How far from speaker to mic?
I have no idea what that latter rule is, link?
Wow, I just realized your room is bare. Why?? Not a living room?

 

[DanDan]

Rear ports do not need much space to work. Do try close to the wall. Even inside the bay window and flush to the two walls if that is possible.

 

[AustinJerry]

What dB value would constitute a flat response (or “the middle”)?
In other words, how high are my peaks and how low are my nulls?
Can I change the DB display to the left to indicate what flat would be?

If you expect useful responses to your questions, you should present the measurement data in a more useful format. When posting a REW measurement, use the small camera icon in the upper left corner of the measurement screen to capture the image. In the image capture options, select "Include Title" and "Include Legend", and add a short description of what the measurement represents, e.g. "Front left speaker, Room Correction on". Including the Legend shows two important pieces of information, the horizontal scale so we can see at which frequency a peak or dip occurs, and whether you have applied smoothing to the measurement (smoothing makes the response curve look much better, when in fact it might not be).

As a general guideline, if a peak and subsequent dip has a range of less than 10dB, it is considered reasonable. So look across the entire measurement to see if any of the dB swings exceeds 10dB. A smoothness of 5dB would be quite good. Set the vertical scale on the measurement screen to read 50dB-100dB in 5dB increments. The scale you are using accentuates the size of the response variations, so it "looks" worse than it is.

Also, it is important to let us know if you are using room correction, e.g. Audyssey. If the measurement is without room correction, we would tell you that you could likely improve the flatness of the response by using room correction, assuming your processor supports it.