Ongoing Beta Beta Test: Precision Room Correction with FDW (Frequency-Dependent Windowing)

[dominikz]

Just did some listening test and validation with REW using the new "precision" (FDW-based) RC in my living room system.

First how the new "Precision" RC results look in WiiM Home App for a single channel compared to the "standard" RC (i.e. "Precision" RC disabled) for the same channel:

Note that except for the "Precision" RC toggle all other settings in RC were identical between the two trials.

Again we see that with "Precision" RC (FDW) enabled the filters are less sharp - this is expected seeing how FDW with the predefined cycle count smooths the response more at low frequencies. But otherwise it looks reasonably similar in both cases.

Listening to both corrections it is IMO apparent right away that the "Precision" RC has significantly less bass - and IMHO not in a good way. Kick drums and bass lines lose a lot of their impact, and consequently the tonality turns bright-ish.

Next I measured the in-room response in REW with either correction applied and got the following responses:

As we can see, the "standard" RC response matches the B&K curve very well, while the "precision" RC undershoots the target <100Hz by 5-6dB. I guess some might perceive this as improved clarity, but to me it sounds bright and bass-deficient; so much so that I suspect there is a bug in the new RC algorithm. The difference is as if I used the "flat" curve instead of "B&K"!

The "standard" RC is IMHO definitely better in bass level, but due to the indiscriminate use of positive-gain filters it is IMO also not ideal. To be honest I still prefer my own REW-generated corrections to both of these algorithms - though if I had to rely on WiiM RC I'd personally keep the "Precision" RC off and just use the standard algorithm.

@WiiM Team As you can see above, the bass response with "Precision" RC enabled is significantly below the target. Could this be a bug in the new RC algorithm? Could it be related to the fact that I use a subwoofer/bass management in this system, since there was a similar issue when individual channel correction with subwoofers was introduced?

 

[dominikz]

One thing that wasn't clear to me was if the evaluation response is also using FDW. In REW if I enable FDW then the level of the higher frequencies is reduced so it makes sense that bass frequencies are reduced more after applying room correction.
I am not sure how to use REW to check the result of the WiiM correction with FDW as it seems to me that a simple measurement will not take into account the frequency dependent window.

You don't need to apply FDW in REW to validate the results of WiiM RC - regardless which method was used to generate the correction you should use the same smoothing in REW to compare different corrections.
In REW the smoothing type that should be the closest representation of human hearing is the one called "Psychoacoustic", so that one is a good choice for comparisons. I also like to use 1/12 when I want to see more details.

 

[Smartplug]

I wanted to try out this new RC function myself.
These are my settings :
48/4000KHz - Default Gain & Q - Multiple Pass - Dayton with imported calibration file - No Subwoofer. I haven’t it. - 1/12 octave smoothing.
These are my personal impressions :
FDW
1) To my ear, the bass has been reduced even too much
2) I had the impression that the sound recedes and shrinks in the centre
3) The sound becomes a bit cold
NORMAL RC
1) The bass is more present and, compared to the sound without RC, is more solid.
2) The sound is more forward projected and wider
3) The sound, on the whole, seems warmer to me
These are my impressions from my hearing.
I must add one thing. After listening to several tracks the feeling of poor bass has subsided.
It does come back, however, if I switch on the fly the RC’s type during a track.
I'm curious to hear more impressions about sound comparing the two RC.
Even if....100 people, 100 different ears and 100 different ways of perceiving sound.

Indeed, dynamic and loudness has been reduced because of the bass reduction. When bass is reduce significantly it will give the perception of increase of clarity.

 

[Joseespona]

I think that having multiple ways of executing room correction is great. But for me what would be crucial for all of this to be less clunky would be to allow us to use the post room correction measurement so we can measure at any time

 

[EddNog]

I think that having multiple ways of executing room correction is great. But for me what would be crucial for all of this to be less clunky would be to allow us to use the post room correction measurement so we can measure at any time

I agree. Would love this ability to just run a sweep to check FR of the system as-is.

-Ed

 

[WiiM Team]

Just did some listening test and validation with REW using the new "precision" (FDW-based) RC in my living room system.

First how the new "Precision" RC results look in WiiM Home App for a single channel compared to the "standard" RC (i.e. "Precision" RC disabled) for the same channel:
View attachment 18921
Note that except for the "Precision" RC toggle all other settings in RC were identical between the two trials.

Again we see that with "Precision" RC (FDW) enabled the filters are less sharp - this is expected seeing how FDW with the predefined cycle count smooths the response more at low frequencies. But otherwise it looks reasonably similar in both cases.

Listening to both corrections it is IMO apparent right away that the "Precision" RC has significantly less bass - and IMHO not in a good way. Kick drums and bass lines lose a lot of their impact, and consequently the tonality turns bright-ish.

Next I measured the in-room response in REW with either correction applied and got the following responses:
View attachment 18922
As we can see, the "standard" RC response matches the B&K curve very well, while the "precision" RC undershoots the target <100Hz by 5-6dB. I guess some might perceive this as improved clarity, but to me it sounds bright and bass-deficient; so much so that I suspect there is a bug in the new RC algorithm. The difference is as if I used the "flat" curve instead of "B&K"!

The "standard" RC is IMHO definitely better in bass level, but due to the indiscriminate use of positive-gain filters it is IMO also not ideal. To be honest I still prefer my own REW-generated corrections to both of these algorithms - though if I had to rely on WiiM RC I'd personally keep the "Precision" RC off and just use the standard algorithm.

@WiiM Team As you can see above, the bass response with "Precision" RC enabled is significantly below the target. Could this be a bug in the new RC algorithm? Could it be related to the fact that I use a subwoofer/bass management in this system, since there was a similar issue when individual channel correction with subwoofers was introduced?

Hi dominike,

Thanks very much for your testing. From your in-room response in REW, the result of 'FDW' RC is abnormal. Our engineers will check this today.

 

[Vignus]

I haven't tried/bothered testing it yet, as I currently leave the WiiM RC/PEQ disabled because I found the room correction from my DSPeaker Anti-Mode X2D sounds way better in direct comparison to the WiiM's RC (when compared independently against one another). When I attempted to, "stack," the WiiM correction on top of the Anti-Mode, it definitely sounded...less good, to put it politely.

I also still stick with DSPeaker, it still does a better, more effective job in getting rid of bass resonances. However, I tried the new RC and I think the end result, at least in my case, is better then with standard RC.
There's only only thing I think Wiim RC does slightly better then DSPeaker: it makes the sound a bit less dry, warmer, and is more flexible when it comes to making adjustments to the sound.

 

[EddNog]

I also still stick with DSPeaker, it still does a better, more effective job in getting rid of bass resonances. However, I tried the new RC and I think the end result, at least in my case, is better then with standard RC.
There's only only thing I think Wiim RC does slightly better then DSPeaker: it makes the sound a bit less dry, warmer, and is more flexible when it comes to making adjustments to the sound.

Which model DSPeaker are you using? Do you also use it with a subwoofer? Have to tried changing the crossover setting?

I haven’t adjusted mine from the default 80Hz.

-Ed

 

[Vignus]

Which model DSPeaker are you using? Do you also use it with a subwoofer? Have to tried changing the crossover setting?

I haven’t adjusted mine from the default 80Hz.

-Ed

I have a DX2 with analog out, which is the one I currently use, and an old anti-mode 2 (the one with the pathetic remote) that I don't use anymore. I have a sub ( Yamaha YST-FSW150) but I hardly use it, it's very chaep and not really good sounding. I have always used 80Hz.

 

[slartibartfast]

Just did some listening test and validation with REW using the new "precision" (FDW-based) RC in my living room system.

First how the new "Precision" RC results look in WiiM Home App for a single channel compared to the "standard" RC (i.e. "Precision" RC disabled) for the same channel:
View attachment 18921
Note that except for the "Precision" RC toggle all other settings in RC were identical between the two trials.

Again we see that with "Precision" RC (FDW) enabled the filters are less sharp - this is expected seeing how FDW with the predefined cycle count smooths the response more at low frequencies. But otherwise it looks reasonably similar in both cases.

Listening to both corrections it is IMO apparent right away that the "Precision" RC has significantly less bass - and IMHO not in a good way. Kick drums and bass lines lose a lot of their impact, and consequently the tonality turns bright-ish.

Next I measured the in-room response in REW with either correction applied and got the following responses:
View attachment 18922
As we can see, the "standard" RC response matches the B&K curve very well, while the "precision" RC undershoots the target <100Hz by 5-6dB. I guess some might perceive this as improved clarity, but to me it sounds bright and bass-deficient; so much so that I suspect there is a bug in the new RC algorithm. The difference is as if I used the "flat" curve instead of "B&K"!

The "standard" RC is IMHO definitely better in bass level, but due to the indiscriminate use of positive-gain filters it is IMO also not ideal. To be honest I still prefer my own REW-generated corrections to both of these algorithms - though if I had to rely on WiiM RC I'd personally keep the "Precision" RC off and just use the standard algorithm.

@WiiM Team As you can see above, the bass response with "Precision" RC enabled is significantly below the target. Could this be a bug in the new RC algorithm? Could it be related to the fact that I use a subwoofer/bass management in this system, since there was a similar issue when individual channel correction with subwoofers was introduced?

This is mine. Similar story

 

[pjorgenunes]

This is mine. Similar story

View attachment 18953

I don’t have a graphic but just by ear I can tell that the “precision room eq” does, when compared to my previous eq:
- much less bass: became a thin sound and with no body;
- to much focus on the Center (like if it was a mono sound)
- due to the last point I have much less stereo image and soundstage.

 

[slartibartfast]

W

I don’t have a graphic but just by ear I can tell that the “precision room eq” does, when compared to my previous eq:
- much less bass: became a thin sound and with no body;
- to much focus on the Center (like if it was a mono sound)
- due to the last point I have much less stereo image and soundstage.

What worries me is I was starting to convince myself I preferred it

 

[Joseespona]

W

What worries me is I was starting to convince myself I preferred it

I am on the same boat. Yesterday I had to much wine and I liked the bass

 

[dominikz]

W

What worries me is I was starting to convince myself I preferred it

That is perfectly normal and happens to every human being. I've absolutely had my share of similar experiences.

It's just that some people never seem to admit it happens to them too (though it does).

So kudos to you!

 

[canard]

That is perfectly normal and happens to every human being. I've absolutely had my share of similar experiences.

It's just that some people never seem to admit it happens to them too (though it does).

So kudos to you!

who knows...maybe it will end up without correction or just on a subwoofer ;-)

 

[pjorgenunes]

W

What worries me is I was starting to convince myself I preferred it

I'm always for the side that what one likes the most is for sure the best for him

 

[MCG555]

W

What worries me is I was starting to convince myself I preferred it

It confirms again all the thories about psycho acoustics…

I stopped RC - EXCEPT for manually correcting a room-mode peak at 47 Hz via parametric EQ…

 

[jiaxin.li]

Just did some listening test and validation with REW using the new "precision" (FDW-based) RC in my living room system.

First how the new "Precision" RC results look in WiiM Home App for a single channel compared to the "standard" RC (i.e. "Precision" RC disabled) for the same channel:
View attachment 18921
Note that except for the "Precision" RC toggle all other settings in RC were identical between the two trials.

Again we see that with "Precision" RC (FDW) enabled the filters are less sharp - this is expected seeing how FDW with the predefined cycle count smooths the response more at low frequencies. But otherwise it looks reasonably similar in both cases.

Listening to both corrections it is IMO apparent right away that the "Precision" RC has significantly less bass - and IMHO not in a good way. Kick drums and bass lines lose a lot of their impact, and consequently the tonality turns bright-ish.

Next I measured the in-room response in REW with either correction applied and got the following responses:
View attachment 18922
As we can see, the "standard" RC response matches the B&K curve very well, while the "precision" RC undershoots the target <100Hz by 5-6dB. I guess some might perceive this as improved clarity, but to me it sounds bright and bass-deficient; so much so that I suspect there is a bug in the new RC algorithm. The difference is as if I used the "flat" curve instead of "B&K"!

The "standard" RC is IMHO definitely better in bass level, but due to the indiscriminate use of positive-gain filters it is IMO also not ideal. To be honest I still prefer my own REW-generated corrections to both of these algorithms - though if I had to rely on WiiM RC I'd personally keep the "Precision" RC off and just use the standard algorithm.

@WiiM Team As you can see above, the bass response with "Precision" RC enabled is significantly below the target. Could this be a bug in the new RC algorithm? Could it be related to the fact that I use a subwoofer/bass management in this system, since there was a similar issue when individual channel correction with subwoofers was introduced?

Users can enable FDW in the IR Windows option of REW to verify WiiM's FDW processing.
If the goal is to correct only the low frequencies (e.g., below 200Hz), FDW does not need to be enabled, as this range is mainly affected by room modes. Minimum-phase EQ is sufficient for correction, without relying on FDW for time window adjustments.
FDW is primarily used to reduce the impact of reverberation in high-frequency measurements, as reverberation has a more significant effect on mid and high frequencies. Therefore, FDW is more useful when measuring higher frequencies.

 

[slartibartfast]

I tried precision RC 20-20000Hz and this is the evaluation result. I assume this must be using FDW as well judging by the big dips at higher frequencies.

 

[dominikz]

If the goal is to correct only the low frequencies (e.g., below 200Hz), FDW does not need to be enabled, as this range is mainly affected by room modes. Minimum-phase EQ is sufficient for correction, without relying on FDW for time window adjustments.

Absolutely agree with this! Actually, I already wrote about this at some length in this post and and in this one.

FDW is primarily used to reduce the impact of reverberation in high-frequency measurements, as reverberation has a more significant effect on mid and high frequencies. Therefore, FDW is more useful when measuring higher frequencies.

As mentioned in the posts linked above, while I agree in principle with what you are saying, it is perhaps worth reiterating the following:

• EQ corrections in mid and high frequencies are actually loudspeaker response corrections, not "room correction" any more.
• FDW attempts to filter out room reflections from mid and high frequency part of the in-room response, to get only the direct loudspeaker response to apply the correction to.
• However, to achieve this FDW cycle length would need to be tuned to the specific room geometry and placement.

As you can see in my own measurements with and without FDW RC shared here (WiiM Mini with Neumann KH120A in a desktop system) and here (WiiM Amp Pro with Revel M16 + sub in a living room system), the measured mid/high frequency response when FDW is enabled is quite irregular.

However if you compare that to full polar quasi-anechoic measurements I made of both these loudspeakers (which shows much cleaner direct sound responses), you will see the default FDW settings in WiiM RC don't seem to be well-tuned for either of my two layouts. Therefore WiiM RC with FDW doesn't correctly extract the loudspeaker direct sound at mid/high frequencies at the listening location when compared to anechoic data.

Links to quasi-anechoic loudspeaker measurements:

• Neumann KH120A
• Revel M16

As a result, I would never use WiiM RC above the room transition frequency (neither with FDW enabled nor disabled), as that would surely degrade the otherwise very well-measuring direct response of my loudspeakers.
And as you said yourself, below the room transition frequency FDW is not especially beneficial and regular smoothing can be used instead.

Note that this is not so much a critique specific to the WiiM implementation of FDW, as I had similar experiences with other RC tools using FDW too - which is exactly why I said early on that:

IMHO FDW is an interesting concept to experiment with, but I've personally always found it less effective and more complicated than simply using MMM with variable smoothing in REW to correct resonances below the room transition frequency (usually below 400Hz).
Corrections above the room transition frequency are not "room correction" anymore, it becomes "loudspeaker correction" instead - and this better done based on anechoic measurements rather than FDW.
Note that a lot of EQ presets for various loudspeaker models based on anechoic measurements can be found on spinorama.org.

Lastly, while I'm sure the WiiM RC with FDW will work well in some situations, it definitely appears unsuitable for use in either or my systems.
To be perfectly honest, unless some very smart way to automatically optimize low-level FDW configuration for each system is introduced, I fear it might not be suitable for use in most systems.