Where should the estimated in-room response trend line go with a sub?

[Yorkshire Mouth]

We’re very used to seeing Amir and Erin’s estimated in-room response trend lines, sloping upwards from right to left, usually stopping at around 100Hz, depending on the speaker. And I believe it’s generally thought to be ‘correct’ if the left hand side is c.5dB higher than the right.

But if you were to add a sub, or if you had a truly full-range speaker (20Hz-20kHz) what should the trend line do, further to the left (between 100Hz and 20Hz).

Should it continue along the same slope, which I suppose would make 20Hz around +3dB over where it’s at at 100Hz? Flatten? Drop off slowly?

Or should you just leave the sub un-EQ’d (in respect of speaker FR)?

 

[holdingpants01]

I never liked the steady sloping response with truly full range speakers, too much extreme subs and not enough high end, I much prefer shelves on both sides (additive on the low end and subtractive on the high end) on top of flat in room response

 

[Purité Audio]

Whatever you like, I had a couple of older guys here last week who were just not used to low bass, another guy the day before who was used to it and enjoyed it.
As long as there isn’t resonance.
Keith

 

[StigErik]

It depends somewhat on personal preference I think. I use something like this, although slightly more at the lowest end.

Here's mine:

Hi-Shelf: -4dB, 2500Hz, Q=0.5
Low-shelf: +12dB, 60Hz, Q=0.7

 

[thewas]

I use a low shelf of +3 dB, 6 or more dB are too much bass for me with good recordings and medium to loud SPLs, can work though as a loudness compensation for low listening levels or poorer recordings.

 

[StigErik]

I think that the bass shelf level depends a little on how long the room decay is at low frequencies. Typically living rooms have 500ms or longer decay at low frequencies, and that will increase the subjective "level" of the bass.

I have a DBA, where the decay get extremely short. Thus I need more level...

 

[Yorkshire Mouth]

So, I’m just theorising here/coming at it from a slightly different angle. If you had a loudspeaker which reproduced exactly what had been recorded, then in an anechoic chamber it should be flat 20Hz-20kHz.

We know that speakers which measure flat nearfield on the klippel rig will slope on the estimated in-room response. These speakers tend to start to roll off at c.100Hz or so.

If we took the theoretical ‘flat-from-20Hz-to-20kHz-anechoic’ speaker, what would the estimated in-room do between 20Hz and 100Hz?

I’m probably not wording this very well.

 

[Yorkshire Mouth]

Put yet another way, does this answer my question?

 

[dominikz]

We’re very used to seeing Amir and Erin’s estimated in-room response trend lines, sloping upwards from right to left, usually stopping at around 100Hz, depending on the speaker. And I believe it’s generally thought to be ‘correct’ if the left hand side is c.5dB higher than the right.

But if you were to add a sub, or if you had a truly full-range speaker (20Hz-20kHz) what should the trend line do, further to the left (between 100Hz and 20Hz).

Should it continue along the same slope, which I suppose would make 20Hz around +3dB over where it’s at at 100Hz? Flatten? Drop off slowly?

Or should you just leave the sub un-EQ’d (in respect of speaker FR)?

This is a very good question, and one that I've thought about several times through the years.

First, let us repeat why we see downward-sloping in-room response with well-measuring front-firing loudspeakers: it is because they exhibit increasing directionality with frequency. I.e. they radiate omnidirectionally at very low frequencies, and become more directional as frequencies go up.

E.g. if we look at Neumann KH150 spin we can see it becomes basically omnidirectional around 100Hz:

See how the ER/SP slope ends once the DI crosses zero (or more commonly: when DI crosses the DI offset curve).
Since this is now an omnidirectional radiation part of the response, the natural shape of the PIR curve should also be "flat" in this region - which is indeed what we see:

Note how the response is getting close to flat below 500Hz (where the DI value is already low) and gets quite flat <100Hz

Subwoofers naturally radiate omnidirectionally, so if we integrate a theoretically perfect subwoofer at the typical 80Hz crossover point the natural PIR of a flat on-axis integrated system would remain flat in this region.

This means that you'd end up having a PIR that is flat until about 100-200Hz, and increasingly sloped-down above it - with the slope depending largely on the specific loudspeaker DI shape (and room reflectivity and air absorption factor, of course).

So when listening at reference levels, I'd say the target curve for bass EQ should follow that same shape - flat in the bass and sloping-down above it, to follow the natural trend of the in-room measured response. Note that this is in principle similar to the famous Brüel & Kjær (B&K) room curve.

In practice, however, I find keeping the same slope in the low bass when performing EQ can be quite beneficial; and especially so if you listen at relatively low SPLs. I find the slope compensates a bit for the relatively low sensitivity of human hearing at very low frequencies (and low levels), making low bass more audible. At high SPLs the bass could become too much, so in that case you may prefer a flatter curve in the bass.

So to summarize, I'd say one needs to take the following factors in consideration when choosing the ideal target curve shape for bass EQ:

• Listening distance - less slope the closer you listen
• Room reflectivity - more slope the more reflective the room is
• Loudspeaker DI - more slope the more directional the loudspeaker is; but crossing to flat where DI becomes zero
• Expected listening levels - it can be beneficial to keep the slope in the bass when listening at low SPLs

 

[007Shortz]

I think the following post is important too. In a first approximation the region between 20 and 100hz is more flat or flat depending on the used target curve:

What is your favorite house curve

There is more to it than just a "house curve". Here are a few industry guidelines: In 1974, Bruel and Kjaer instruments wrote an application note on, "Relevant loudspeaker tests in studios in Hi-Fi dealers' demo rooms in the home etc. using 1/3 octave, pink-weighted, random noise." See Figure...

audiosciencereview.com

 

[sigbergaudio]

If a speaker is truly flat anechoically to 20hz I think you would see a rise all the way to 20hz in-room.

What most people who have subs dial in is a pretty flat area from 20hz-100hz, or even 20-200hz, so if something closer to that is the goal, you'd want the speaker start to roll off a bit earlier than 20hz to avoid the potentially massive room gain at the bottom.

The 12dB/octave roll-off of a sealed speaker or sub that starts to roll off gently at 25-30hz matches the room gain in many rooms quite well, ending up flat to 20hz or below.

 

[ZolaIII]

Wouldn't go anywhere, it's determined by your room length primarily and roughly it would dance a bit on average first peak (& related deep) and influenced to following two. It would be at the full swing going 20 - 20 (hz/KHz) that's all. Goal is one or other way (placement, impuls or counter cancellation) to bring it to useful level not to kill it entirely and influence is higher (200 Hz roughly as border line).

 

[holdingpants01]

View attachment 354515Put yet another way, does this answer my question?

Those simulations aren't that accurate, my 8351B + W371A measure anechoic flat to 24Hz or so, but doesn't measure like that in any room I've used them, they're rather flat in the midrange and top with bass bump and dips from the room

 

[Willem]

Subjectively, the impact of uncontrolled room modes will be the dominant factor. Deal with those first ,and worry about your preference for a particular curve later.

 

[dominikz]

Those simulations aren't that accurate, my 8351B + W371A measure anechoic flat to 24Hz or so, but doesn't measure like that in any room I've used them, they're rather flat in the midrange and top with bass bump and dips from the room

From what I've seen in my own tests, as well as measurements done by other people, PIR usually matches really well to in-room measured spatially-averaged responses (e.g. MMM) above approximately 1kHz. Between 300Hz-1kHz the match is usually still pretty good, but room influence is quite visible, and below 300Hz the room completely dominates the measured response.

A few of my own measurements provided as reference:
(source 1)

(source 2)

A few external references (from erinsaudiocorner.com):

How well the results match the PIR depends also on how reflective the room is (e.g. PIR assumes early reflections are not absorbed), and on the listening distance (higher frequencies go down as you move further - as you can see in the red plot above).