Ongoing Beta Beta Testing: Independent L/R Room Correction with Subwoofer

[slartibartfast]

@slartibartfast Is the response in red the summed L+R channel response measurement?

I'm asking because if you EQ left and right channels each individually to the B&K target, it is expected that the summed L+R response will be about 3dB higher in the bass range.

In my experience the two channels will sum as incoherent sound sources at high frequencies (i.e. +3dB compared to individual channel level) and as coherent sound sources at low frequencies (i.e. +6dB compared to individual channel level).
The difference is usually the biggest below about 200Hz, which to me looks exactly like what your measurements seem to show.

So if you use the "stereo" RC variant I would expect the summed L+R measurement to match the B&K curve well, but if you use the individual channel RC variant I'd expect the summed L+R measurement to overshoot the B&K curve by about 3dB below 200Hz or so.

This would IMHO be the expected behavior if the new RC algorithm works correctly.

One way to address this would be by introducing a new variant of the B&K target curve with approx. 3dB less bass, which could be used by those who want the summed L+R response to match the B&K target when using individual channel correction.

Personally I prefer to have each channel match the original (unmodified) B&K curve, and have the +3dB bass level in the summed response, and I wouldn't want to lose this capability.


With the original Beta release of the individual channel room correction with subwoofer enabled I also saw a similar dip between 30-80Hz with the newly calculated RC filters (this was IMHO indeed a bug):
View attachment 17953
The RC function didn't predict this dip:
View attachment 17954

But it did measure it when running result assessment - reinforcing the fact that this was not working as expected:
View attachment 17955

Note that I haven't yet tried to re-do RC with the updated version of the algorithm mentioned by @WiiM Team in post #28. Based on your (@slartibartfast) results from the previous post it seems to me the new version solved the issue, and I'm looking forward to retesting with the new algorithm!

Both of the responses are the result of L/R measurements and both show the response with both speakers driven. Red is the new algorithm and Blue is the old algorithm that you used. It will be interesting to see your results with the new algorithm.

 

[dominikz]

Both of the responses are the result of L/R measurements and both show the response with both speakers driven. Red is the new algorithm and Blue is the old algorithm that you used. It will be interesting to see your results with the new algorithm.

Managed to do a few tests today. First, I tested to see if my own manual PEQ saved config were negatively affected by re-measuring the response in REW and comparing to saved measurements of the same PEW config from January.
They match really well when SPL is aligned, so it seems to me in my case the saved PEQs were not negatively affected:


Next I also re-ran the individual channel RC with subwoofer active, and got the following results (20Hz-350Hz correction range):

As we can see, the 30-80Hz dip is now gone.

I've validated the results in REW and the individual channel response matches pretty well against the B&K target:


Lastly, to illustrate the point from my previous post about the expected +3dB gain in the low frequencies when measuring both L+R channels together (caused by bass frequencies summing like coherent sound sources):

So to me this all looks to be working well after the latest update of the beta RC algorithm.

Well done @WiiM Team!

P.S. One part of the current implementation of WiiM Room Correction functionality I really dislike is the inability to limit how much PEQ boost is applied.
As you can see in the above measurements, WiiM RC tries to boost a null around 200Hz in the right channel of my system with several positive PEQ filters with up to +6dB gain:

This is a serious waste of amplifier power and, given that there's no auto pre-gain, something which could result in digital clipping.

@WiiM Team Is there any chance that we get separate Max Gain controls for positive-gain (boost) vs negative-gain (cut) filters? I'd like to allow -12dB gain for cuts while at the same time completely block boosts (0dB) - and I'm pretty sure I'm not the only one.
At the moment a single control ("Max Gain") controls the allowed amount of both positive and negative gain range in calculated PEQ filters.
Thanks!

 

[EddNog]

...and I'm pretty sure I'm not the only one.
At the moment a single control ("Max Gain") controls the allowed amount of both positive and negative gain range in calculated PEQ filters.
Thanks!

I've asked like four times for this already.

I'm also glad to see that I'm not the only one with a strange and difficult to explain null around ~185Hz

What speakers and what subwoofer are you using? What's your crossover point?

I have a pair of LS50 Metas crossed over at 80Hz with an SVS SB-3000 and get that null between 150 and 200Hz.

-Ed

 

[slartibartfast]

Managed to do a few tests today. First, I tested to see if my own manual PEQ saved config were negatively affected by re-measuring the response in REW and comparing to saved measurements of the same PEW config from January.
They match really well when SPL is aligned, so it seems to me in my case the saved PEQs were not negatively affected:
View attachment 17971

Next I also re-ran the individual channel RC with subwoofer active, and got the following results (20Hz-350Hz correction range):
View attachment 17972View attachment 17973
As we can see, the 30-80Hz dip is now gone.

I've validated the results in REW and the individual channel response matches pretty well against the B&K target:
View attachment 17974

Lastly, to illustrate the point from my previous post about the expected +3dB gain in the low frequencies when measuring both L+R channels together (caused by bass frequencies summing like coherent sound sources):View attachment 17975

So to me this all looks to be working well after the latest update of the beta RC algorithm.

Well done @WiiM Team!

P.S. One part of the current implementation of WiiM Room Correction functionality I really dislike is the inability to limit how much PEQ boost is applied.
As you can see in the above measurements, WiiM RC tries to boost a null around 200Hz in the right channel of my system with several positive PEQ filters with up to +6dB gain:
View attachment 17976
This is a serious waste of amplifier power and, given that there's no auto pre-gain, something which could result in digital clipping.

@WiiM Team Is there any chance that we get separate Max Gain controls for positive-gain (boost) vs negative-gain (cut) filters? I'd like to allow -12dB gain for cuts while at the same time completely block boosts (0dB) - and I'm pretty sure I'm not the only one.
At the moment a single control ("Max Gain") controls the allowed amount of both positive and negative gain range in calculated PEQ filters.
Thanks!

I haven't tried remeasuring my saved Stereo PEQ filter yet with the new algorithm. Did you check yours with the previous algorithm and find they were affected? WiiM haven't stayed as much but I assume the extra 3 dB at subwoofer frequencies is the reason L/R correction with sub has been delayed for so long. In your final evaluation it still looks too high to me compared to the target curve and is higher than it was with the previous algorithm. It seems that both the previously saved Stereo responses and the L/R responses have been boosted at subwoofer frequencies compared to the last algorithm. We might find that a flat target curve gives us a B&K result for L/R with sub.
Surely the target curves should be modified to take the 3dB bass boost into account.

 

[dominikz]

I've asked like four times for this already.

I'm also glad to see that I'm not the only one with a strange and difficult to explain null around ~185Hz

What speakers and what subwoofer are you using? What's your crossover point?

I have a pair of LS50 Metas crossed over at 80Hz with an SVS SB-3000 and get that null between 150 and 200Hz.

-Ed

The nulls around 200Hz in my system are almost certainly caused mainly by SBIR.

Front baffle of my speakers is a little less than 40cm from the wall behind them which puts the main SBIR null at about 220Hz. BTW, I often use this handy online calculator to find SBIR null/peak frequencies.

I'm using Revel M16 + SBS SB-1000 (the older non-Pro model) in this system, crossed over at 80Hz.

 

[dominikz]

Did you check yours with the previous algorithm and find they were affected?

No I haven't, sorry!

In your final evaluation it still looks too high to me compared to the target curve and is higher than it was with the previous algorithm. It seems that both the previously saved Stereo responses and the L/R responses have been boosted at subwoofer frequencies compared to the last algorithm.

I have a different view on this.
The response match I get now with WiiM RC is IME comparable to the type of results I get with any other EQ or RC tool I used when doing individual channel correction. If you match individual channels to any target curve in any SW, the summed L+R response will overshoot the same target in the bass frequencies by +3dB.
As I explained previously, this is simply because bass frequencies emitted by multiple loudspeakers sum as coherent sound sources, while higher frequencies sum progressively more like incoherent sources. So this is IMHO perfectly normal.

Further, how would you then adapt the target curve to multi-channel systems, depending on the channel count? You'll get the same +3dB additional bass level for each additional channel playing simultaneously.
This is one of the reasons it is reasonable to align each channel independently to the same target, and accept the way different frequencies sum.

Surely the target curves should be modified to take the 3dB bass boost into account.

Well that depends on how much bass you want, I guess. I personally like matching the individual channels to B&K curve and have the extra +3dB in the summed response.

I guess ideal would be to have the ability to import custom target curves so people can use what they prefer - or at least to have two variants of the B&K target curve - one that gives you a good match to B&K on individual channels (which we have today), and another that gives you a good match on the summed L+R response. Though one way to achieve the latter today is to simply use the "stereo" RC variant.

 

[slartibartfast]

No I haven't, sorry!


I have a different view on this.
The response match I get now with WiiM RC is IME comparable to the type of results I get with any other EQ or RC tool I used when doing individual channel correction. If you match individual channels to any target curve in any SW, the summed L+R response will overshoot the same target in the bass frequencies by +3dB.
As I explained previously, this is simply because bass frequencies emitted by multiple loudspeakers sum as coherent sound sources, while higher frequencies sum progressively more like incoherent sources. So this is IMHO perfectly normal.

Further, how would you then adapt the target curve to multi-channel systems, depending on the channel count? You'll get the same +3dB additional bass level for each additional channel playing simultaneously.
This is one of the reasons it is reasonable to align each channel independently to the same target, and accept the way different frequencies sum.


Well that depends on how much bass you want, I guess. I personally like matching the individual channels to B&K curve and have the extra +3dB in the summed response.

I guess ideal would be to have the ability to import custom target curves so people can use what they prefer - or at least to have two variants of the B&K target curve - one that gives you a good match to B&K on individual channels (which we have today), and another that gives you a good match on the summed L+R response.

Ideally the results from a stereo room correction should have a similar frequency response to the L/R + sub results but that 3dB boost prevents it. That 3dB boost was there when WiiM decided to remove the ability to perform Room Correction for L/R + sub so if that boost is still there I don't see why it took so long to reinstate it for this beta as nothing has really changed.
I'll carry on with measurements but I can only do them first thing in the morning or last thing at night when the TV isn't on

 

[EddNog]

The nulls around 200Hz in my system are almost certainly caused mainly by SBIR.

Front baffle of my speakers is a little less than 40cm from the wall behind them which puts the main SBIR null at about 220Hz. BTW, I often use this handy online calculator to find SBIR null/peak frequencies.

I'm using Revel M16 + SBS SB-1000 (the older non-Pro model) in this system, crossed over at 80Hz.

Thanks; this is really wonderful advice. Now I’m having a hard time figuring out just how to best resolve this SBIR room null. I know that in the past, I had the speakers placed differently and that null either wasn’t there or was much less. Do I have to move them even farther away from that back wall? This would severely intrude into the room in an uncomfortable manner and affect viewing angles for the TV. I wonder if pushing the speakers back would make it worse, or mess something else up. I know I came to settle on this speaker position for a reason, but it has also been a long time since I did it and I’ve forgotten the details of my logic.

-Ed

 

[slartibartfast]

Thanks; this is really wonderful advice. Now I’m having a hard time figuring out just how to best resolve this SBIR room null. I know that in the past, I had the speakers placed differently and that null either wasn’t there or was much less. Do I have to move them even farther away from that back wall? This would severely intrude into the room in an uncomfortable manner and affect viewing angles for the TV. I wonder if pushing the speakers back would make it worse, or mess something else up. I know I came to settle on this speaker position for a reason, but it has also been a long time since I did it and I’ve forgotten the details of my logic.

-Ed

Side walls can affect the nulls as well. I have seen nulls appear with both speakers driven which weren't there with either speaker driven independently.

 

[EddNog]

Side walls can affect the nulls as well. I have seen nulls appear with both speakers driven which weren't there with either speaker driven independently.

This is just insane.

 

[Mr Ee]

I’ve forgotten the details of my logic.

The better half is usually the main constraint for speaker placement

 

[dominikz]

Thanks; this is really wonderful advice. Now I’m having a hard time figuring out just how to best resolve this SBIR room null. I know that in the past, I had the speakers placed differently and that null either wasn’t there or was much less. Do I have to move them even farther away from that back wall? This would severely intrude into the room in an uncomfortable manner and affect viewing angles for the TV. I wonder if pushing the speakers back would make it worse, or mess something else up. I know I came to settle on this speaker position for a reason, but it has also been a long time since I did it and I’ve forgotten the details of my logic.

-Ed

I usually recommend to read this really nice illustrated loudspeaker/subwoofer placement guide by Genelec.

In short, ideally you'd want to have the loudspeakers about 1,1m away from any walls (as this moves the SBIR dips <80Hz) with the sub right against the wall or in the room corner (which pushes sub's SBIR dips >80Hz). In this way the nulls are out of the operating range of each part of the integrated loudspeaker+sub system and you are ready to use EQ knock down the peaks.

If this is not possible, a good alternative is to push loudspeakers close to the back wall - this moves the main SBIR null to 200Hz or higher, where it is less audibly offending and/or easier to treat acoustically.

As @slartibartfast said, every boundary in the room (walls, floor, ceiling) will cause their own SBIR + and you also get interference caused by room modes. Lastly, the low frequency peaks and dips will shift a bit depending on whether you open or close windows and doors.

Physics is really not making it easy for us...

 

[EddNog]

I usually recommend to read this really nice illustrated loudspeaker/subwoofer placement guide by Genelec.

In short, ideally you'd want to have the loudspeakers about 1,1m away from any walls (as this moves the SBIR dips <80Hz) with the sub right against the wall or in the room corner (which pushes sub's SBIR dips >80Hz). In this way the nulls are out of the operating range of each part of the integrated loudspeaker+sub system and you are ready to use EQ knock down the peaks.

If this is not possible, a good alternative is to push loudspeakers close to the back wall - this moves the main SBIR null to 200Hz or higher, where it is less audibly offending and/or easier to treat acoustically.

As @slartibartfast said, every boundary in the room (walls, floor, ceiling) will cause their own SBIR + and you also get interference caused by room modes. Lastly, the low frequency peaks and dips will shift a bit depending on whether you open or close windows and doors.

Physics is really not making it easy for us...

Thank you so much! I will try both pieces of advice (subwoofer back against the wall, speakers at least 1.3m from the wall) and rerun room correction to recalibrate, and hope that this does it!

-Ed

 

[dominikz]

Ideally the results from a stereo room correction should have a similar frequency response to the L/R + sub results but that 3dB boost prevents it.

The apparent 3dB bass boost in combined L+R response compared to individual channel response is simply a result of physics of multiple full-range sound source summation.

To be more precise, the bass range of the combined response will be +6dB higher above the individual channel responses, and the high frequencies of the combined response will be +3dB above the individual channel responses - this is expected.

You will see the same effect if you measure with MMM individual in-room responses of any two loudspeakers and compare it to the in-room response measured when both are playing at the same time - even without any EQ applied at all, with no subwoofers or WiiM devices involved.

So the only way to make stereo RC and individual channel RC to produce the same shape of the combined L+R response post-correction would be to use an appropriately modified target with each RC type (i.e. 3dB more bass in the target with the stereo channel RC compared to individual channel RC).

However, I'd actually argue that the current stereo RC results in 3dB too little bass, as IMHO the individual channel bass should be well-matched to the B&K target.

 

[dominikz]

Thank you so much! I will try both pieces of advice (subwoofer back against the wall, speakers at least 1.3m from the wall) and rerun room correction to recalibrate, and hope that this does it!

-Ed

No problem at all! I'd be interested to read how it went. Good luck!

 

[slartibartfast]

The apparent 3dB bass boost in combined L+R response compared to individual channel response is simply a result of physics of multiple full-range sound source summation.

To be more precise, the bass range of the combined response will be +6dB higher above the individual channel responses, and the high frequencies of the combined response will be +3dB above the individual channel responses - this is expected.

You will see the same effect if you measure with MMM individual in-room responses of any two loudspeakers and compare it to the in-room response measured when both are playing at the same time - even without any EQ applied at all, with no subwoofers or WiiM devices involved.

So the only way to make stereo RC and individual channel RC to produce the same shape of the combined L+R response post-correction would be to use an appropriately modified target with each RC type (i.e. 3dB more bass in the target with the stereo channel RC compared to individual channel RC).

However, I'd actually argue that the current stereo RC results in 3dB too little bass, as IMHO the individual channel bass should be well-matched to the B&K target.

Isn't the target curve meant to be the target for the listening position with both speakers playing though?
Anyway I have another weird issue now. I change the selected PEQ filter then make a new measurement with REW and frequently the resulting response is virtually the same as as the response with a different filter selected. I have no idea what is going on. I never noticed this happening before this beta started.

 

[slartibartfast]

It seems that whenever I select a L/R PEQ filter set I get the same response for all of them even though they are visibly different , some Harman and others B&K. Selecting Stereo filters I seem to get the correct response. This is driving me insane.

 

[Smartplug]

It seems that whenever I select a L/R PEQ filter set I get the same response for all of them even though they are visibly different , some Harman and others B&K. Selecting Stereo filters I seem to get the correct response. This is driving me insane.

The more you run the more it get complicated. I would probably run multiple test and save them with various names. From there pick what sound best to your ear.

 

[slartibartfast]

The more you run the more it get complicated. I would probably run multiple test and save them with various names. From there pick what sound best to your ear.

There seems to be an issue where all the L/R PEQ filters sound and measure the same though. Nothing to do with my preference. It seems like the selected filter isn't being applied.

 

[WiiM Team]

doManaged to do a few tests today. First, I tested to see if my own manual PEQ saved config were negatively affected by re-measuring the response in REW and comparing to saved measurements of the same PEW config from January.
They match really well when SPL is aligned, so it seems to me in my case the saved PEQs were not negatively affected:
View attachment 17971

Next I also re-ran the individual channel RC with subwoofer active, and got the following results (20Hz-350Hz correction range):
View attachment 17972View attachment 17973
As we can see, the 30-80Hz dip is now gone.

I've validated the results in REW and the individual channel response matches pretty well against the B&K target:
View attachment 17974

Lastly, to illustrate the point from my previous post about the expected +3dB gain in the low frequencies when measuring both L+R channels together (caused by bass frequencies summing like coherent sound sources):View attachment 17975

So to me this all looks to be working well after the latest update of the beta RC algorithm.

Well done @WiiM Team!

P.S. One part of the current implementation of WiiM Room Correction functionality I really dislike is the inability to limit how much PEQ boost is applied.
As you can see in the above measurements, WiiM RC tries to boost a null around 200Hz in the right channel of my system with several positive PEQ filters with up to +6dB gain:
View attachment 17976
This is a serious waste of amplifier power and, given that there's no auto pre-gain, something which could result in digital clipping.

@WiiM Team Is there any chance that we get separate Max Gain controls for positive-gain (boost) vs negative-gain (cut) filters? I'd like to allow -12dB gain for cuts while at the same time completely block boosts (0dB) - and I'm pretty sure I'm not the only one.
At the moment a single control ("Max Gain") controls the allowed amount of both positive and negative gain range in calculated PEQ filters.
Thanks!

Hi dominikz,

Thank you for your suggestions. I’ll review these Max Gain control improvements with our engineering team and aim to implement them promptly.