EQ based on resonances?

[webzeb]

Hello!

First of all, many thanks for that great piece of software that REW is!

I’m completely new to REW and trying to EQ my bedroom listening setup. Indeed, I started by reading tutotials and ebooks here and there.

Since I like to understand things and to optimize all what can be, I tried an approach based on:

• Setting a target curve that should +/- less match the loudspeakers ‘natural’ response in the room minus the resonances/room modes (assuming a low range boosted by a lot of boundary gain). It was very empirical and I’m not sure I did well!
• Identifying the main resonances with the dedicated feature.
• Correcting these resonances with modal filters, copying the RT60 from what REW gave and adjusting the gain until the predicted curve just touch the target curve (that I assume – even it’s probably wrong – to be the ‘natural’ response of the speaker with the ton of boundary gain I have).
• Checking that I got read of the long decays due to resonances on the waterfall.
• Indeed, no boost of nulls.
• Indeed, no correction above 250Hz.

(Please note that I reserved 4 bands of EQ out of 10 for the mid-range/highs based on Spinorama.org proposed EQ parameters. Not sure I will keep it however...)

However:

• I’m not 100% sure it's the best possible approach here (nor it’s a correct approach at all...).
• I’m not 100% sur I applied it well.
• I have strong asymmetry that I don’t know how to approach!
• I have some modes that are just next to some dips/nulls and I don’t know how to treat them (do I correct them at the risk of amplifying the dips or do I leave them alone?)
• I have some modes that are extremely close to each other, and I don’t know what to do (I guess correcting each one with a dedicated modal filter is a bad idea ; plus it steal available EQ bands...).
• I have a lack of energy on one channel between 60-70Hz that does not seem to be due to nulls: how I make sure and, in this exceptional case, is positive gain okay?
• I have an enormous lack of energy in the low mid-range due to nulls on the other speaker, but not quite at the same frequencies.
• I'm lacking psychoacoustics knowledge to make the better choices (I've just order Dr. Toole's book but will only receive it within a month...)/

What is important for me is that the result is the best in terms of psychoacoustics/listening.

May you have advice?

Thanks!

Fred

PS: details below, as well as measurements attached


Listening setup:

• WiiM Ultra with independent L/R 10-band PEQ (f with 0.1Hz res. / G with 0.1dB res. and -12dB~+12dB / Q with 0.01 res. and 0.1~24.00)
• Hypex NC122MP
• Wharfedale Linton Heritage 85th Anniversary

Room:
14sqm / 151sft. Bedroom
Asymmetric (front and rear walls are parallel / left and right walls are not and, in addition, are not the same material)
Listening from the bed, which has a big and thick headboard
Loudspeakers are only 11cm (4’’3/8) from the wall / unfortunately it’s difficult to adjust that without compromising
Loudspeakers are on a sideboard, with 1.63m / 5’4’’ of distance between their centers
They are more or less symmetrically located to the listening position
Listening distance is about 2.60m / 8’6’’

Measuring setup:

• Macbook connected to the WiiM Ultra throught SPDIF
• REW v.5.40Beta 132
• WiiM Ultra with:
  • Balance adjusted by ear (not perfected)
  • EQ disabled
• MiniDSP UMIK-1 with foam cap, facing ceiling (except for a few measurements, see details), and 90deg calibraiton profile
• Separated measurements for each loudspeaker
• 1M Sweep
• -12dBS
• Measured level of about 77dB with one loudspeaker / 80dB with two loudspeakers is my max. listening volume, but I often listen way lower
• 7 measurements per loudspeaker, averaged (RMS+Phase) :
  • Left ear
  • Face
  • Right ear
  • Left ear, slightly (~15cm / 6’’) shifted down and left (mic pointing down to the bed)
  • Right ear, slightly (~15cm / 6’’) shifted down and right (mic pointing down to the bed)
  • Left ear, slightly (~15cm / 6’’) shifted top and left
  • Left ear, slightly (~15cm / 6’’) shifted top and left

 

[webzeb]

P.S.: I may do *very light* room treatment. I was thinking about filing the headboard - which is thick but hollow and using linen that looks pretty 'sound-transparent' - with sound aborbing foam.
But this is a small room that serves as master bedroom, home office and listening room. Hence, the different use cases are already fighting each other so there is very little I can do without compromising too much...

 

[Keith_W]

Nothing you wrote makes any sense. What is a "modal filter"? You can not "EQ resonances", you can only EQ the frequency response. As it turns out, peaks in the frequency response also correspond with longer decay (higher peak = longer decay), so flattening a peak will also reduce the decay. But we don't call that "EQ'ing a resonance", it's simply "EQ'ing a peak". Forget everything you wrote, I have never heard of anything like what you describe. We don't "EQ to the RT60", the RT60 is dealt with room treatment in the frequency range where the RT60 is meaningful (i.e. short wavelengths). If you are proposing a new method of EQ, you will have to test it, show results, and convince everyone else that your novel technique actually works. Otherwise, I suggest you wait until you obtain your copy of Dr. Toole's book. Of note, his book does not tell you how to EQ. You will need other resources for that.

 

[webzeb]

Hello,

Thanks for your reply.
I'm a beginner, so please be indulgent!
I have zero pretention to (re)invent anything; just trying and learning...

My understanding is that the frequency response I measure is a combination of:

• Direct sound coming from the speaker,
• Sound reflected by walls, ceiling, floor, furniture […],
• Among this reflected sound, lower frequencies that contrary to the highs the speaker also emit at the back and are reflected byy the front wall, boosting the bass region (and causing some dips by inteference at some precise frequencies),
• Rooms modes which accumulate energy and can result in peaks and nulls depending on where you measure in the room.

All interacting (can add or subtract depending on the phases).
Is that correct?

I was simply trying to correct for the peaks created by room modes, leaving the rest untouched.
My understanding is that correcting for these peaks also reduce the decay time.
is that correct?

Regarding the modal filter, in the EQ, there is a filter called ‘modal’.
Parameters are:

• Frequency,
• Gain
• RT60

My understanding is that this type of filter aims at targeting precisely peaks in the frequency response cause by room modes, by automatically adjusts the Q factor depending on the resonant mode decay time and the gain.
Is that correct?

 

[webzeb]

PS: I do understand that no EQ will magically get rid of the resonances. There still be there. There still be frequency response difference due to room modes depending on where you stand.
My question was: is the approach correct to correct finely for the peaks room modes creates in the frequency response? (And incidentally reduce the associated decay time below audible threshold.)

 

[Keith_W]

Hello,

Thanks for your reply.
I'm a beginner, so please be indulgent!
I have zero pretention to (re)invent anything; just trying and learning...

My understanding is that the frequency response I measure is a combination of:

• Direct sound coming from the speaker,
• Sound reflected by walls, ceiling, floor, furniture […],
• Among this reflected sound, lower frequencies that contrary to the highs the speaker also emit at the back and are reflected byy the front wall, boosting the bass region (and causing some dips by inteference at some precise frequencies),
• Rooms modes which accumulate energy and can result in peaks and nulls depending on where you measure in the room.

All interacting (can add or subtract depending on the phases).
Is that correct?

I was simply trying to correct for the peaks created by room modes, leaving the rest untouched.
My understanding is that correcting for these peaks also reduce the decay time.
is that correct?

It's probably more accurate to say that you want to correct minimum-phase room modes. That concept can be a bit difficult to understand for a beginner - I suggest you read what John wrote about it here. These minphase regions are more likely to be found at low frequencies, and you need to check that they are stable over a listening area and not a single and very specific position. But yes, what you said is otherwise correct.

Regarding the modal filter, in the EQ, there is a filter called ‘modal’.
Parameters are:

• Frequency,
• Gain
• RT60

Well, it looks like you came across something in REW that I never noticed before I'll let others who know more comment about it. In the meantime i'll read up a bit about it and learn what it actually does.

 

[John Mulcahy]

I tried an approach based on:...

Yes, that's a reasonable approach, and it's why the EQ window has tools to find resonances, show live updates of the predicted waterfall as filters are adjusted and show pole-zero plots for the response and the filters. However, the few studies that have been done on the audibility of low frequency resonant decays conclude that sensitivity to the frequency response effect is much higher than sensitivity to the decay itself, so the primary goal is to even out the response.

 

[janbth]

Hello,

Thanks for your reply.
I'm a beginner, so please be indulgent!
I have zero pretention to (re)invent anything; just trying and learning...

My understanding is that the frequency response I measure is a combination of:

• Direct sound coming from the speaker,
• Sound reflected by walls, ceiling, floor, furniture […],
• Among this reflected sound, lower frequencies that contrary to the highs the speaker also emit at the back and are reflected byy the front wall, boosting the bass region (and causing some dips by inteference at some precise frequencies),
• Rooms modes which accumulate energy and can result in peaks and nulls depending on where you measure in the room.

All interacting (can add or subtract depending on the phases).
Is that correct?

I was simply trying to correct for the peaks created by room modes, leaving the rest untouched.
My understanding is that correcting for these peaks also reduce the decay time.
is that correct?

Regarding the modal filter, in the EQ, there is a filter called ‘modal’.
Parameters are:

• Frequency,
• Gain
• RT60

My understanding is that this type of filter aims at targeting precisely peaks in the frequency response cause by room modes, by automatically adjusts the Q factor depending on the resonant mode decay time and the gain.
Is that correct?

I agree with John and think you are on the right track.

It’s a good choice to order Sound Reproduction by Floyd Toole. I think you will find Chapter 14 («Below the Transition Frequency: Optimizing the Experience») to be relevant.

 

[ddude003]

Mitch Barnett has a lot of useful information here... https://accuratesound.ca/resources-audio-digital-signal-processing/