Computer Optimization of Room Acoustics

[ernestcarl]

anti-room-correction

Never heard it phrased in such a manner, but ok. It sounds "magical" due to the marketing speak often used to sell said product.

There's talk about DRC being able to send out an anti-sound a bit delayed from the original sound, from the same speaker driver, to counteract in-room reflections and speaker reflections.

The latter ("speaker correction" and not "room correction") actually does work and is regularly used to correct/improve some horn speaker designs by reducing back reflections/echoes causing coloration in sound -- but the reflections are very consistent in this application.

Inside a "room" where positions change a lot all of the time... that's where the specific technique in question doesn't work so good or often make things worse. I believe improving the room acoustics -- besides purchasing better speakers -- itself is the goal to aim for.

A speaker system's magnitude and time deviations from linear can be fixed to some degree through digital digital processing, for sure, but you got to know what is you're doing -- and the limitations of your corrections based on the speaker design -- even when using advanced semi-automated software. Speaker system designers should be the ones correctly implementing this right from the start, though.


*Then again, if you are implementing a multi-way system, you are also sort of now becoming the system designer of your own to a lesser degree... most people do not choose this route.

 

[FrantzM]

Found MSO and I am convinced this is the way of the future.

My next project will be to use MSO to integrate (at first) 2 subs with my 8341 mains.

I wonder if it is possible to run MSO in the 0-100Hz band for subs and also run it again for the 100-300Hz band again to correct the mains for room modes.

Again it would be limited to the 2 mains but experimentally I wonder what could be done with 2 Rhythmik FM8 midbass units. I could position those as L and R but with different wall distance for example. Then a program like MSO could optimize the in room response at least to 250Hz due to limitation of the FM8.

Does this sound like a sensible thing to do?

If so I suppose DDRC88 is the only way to go keeping everything in hardware? I’d like the mains to stay digital but the subs and FM8s can use analog out through whatever converter - not particular about that.

Better bass is my issue too and I will , eventually , tackle MSO.
@dallasjustice has done that using the FM8 with his JBL M2 with 2 other subwoofers for a total of 4. He uses audiolense for DRC and crossover. Je no longer post unfortunately here.
I have some free time and will read on MSO. The process so far looks a bit daunting to me. I ha e however all the necessary tools and accessories.

 

[Digital_Thor]

Never heard it phrased in such a manner, but ok. It sounds "magical" due to the marketing speak often used to sell said product.



The latter ("speaker correction" and not "room correction") actually does work and is regularly used to correct/improve some horn speaker designs by reducing back reflections/echoes causing coloration in sound -- but the reflections are very consistent in this application.

Inside a "room" where positions change a lot all of the time... that's where the specific technique in question doesn't work so good or often make things worse. I believe improving the room acoustics -- besides purchasing better speakers -- itself is the goal to aim for.

A speaker system's magnitude and time deviations from linear can be fixed to some degree through digital digital processing, for sure, but you got to know what is you're doing -- and the limitations of your corrections based on the speaker design -- even when using advanced semi-automated software. Speaker system designers should be the ones correctly implementing this right from the start, though.


*Then again, if you are implementing a multi-way system, you are also sort of now becoming the system designer of your own to a lesser degree... most people do not choose this route.

Thank you for your reply
The reason I wrote "anit-room-correction" , is because I've heard so many refer to these auto-system-softwares, as a way to "remove" the room. I know this is wrong, since the room will always be the same... and if you really wanted to do something like this, you might have to go the B&O route like in the Beolab 50 and 90, where multiple drivers co-op to create a "new" soundwave, rather than the fixed one from a normal single driver in a typical speaker box.

I tried to build a speaker with a baffle that co-op with the drivers and then put the speaker on a box in the middle of the room, after I moved the sofa, so that a minimum of reflections was in the way. Then I measured it on multiple angles to find where the filter would work the best, and then EQ'ed the gated response of the tweeter and midrange to work in sync - and finally checked the result to be sure that everything worked out as planned. Then I moved the speakers back in place and added the subwoofers to create coherence between mains and subs. Not that difficult but require a bit of fiddling. Everything now sound really smooth and neutral - both with movies and music. Sometimes when seeing a documentary or talk.... I kinda forget the speakers... the sound just seem to come from the screen.
Every time I hear these auto corrected systems, that are measured in the listening position.... they seem to sound muffled and un-natural. They can sound a bit "smoother" and better cover up imperfections in older recordings.... but still nothing I don't feel that I could imitate by simply playing around a bit with basic EQ.

Fulcrum acoustics might do what you mention about the horn... in this case a compression driver:

 

[schlager]

MSO works like a charm, bit of a learning curve, but once you get the hang of it, I actually find it funny to use, because it does really work really well. Funny to watch your in room future bass response changing, as MSO is crushing the numbers.
There is much debate of digital room correction (DRC) working or not. To my ears it is working very well and my before/after measurements supports that notion , but not all DRC programs are created equal. I use DRC Designer and along with Audiolense and Acourate they are reported to work best. Probably because they give full time domain correction and much more user tweaking possibilities, where other programs only give partly time domain correction or none at all aka only using IIR filters.

Lets get one thing straight, DRC is not supposed to remove the room, but to remove the unwanted artifacts any room will imprint on the sound received in the sweet spot. And correctly done it will in fact improve the sound over a wide listening area. And yes DRC makes anti-pulses to cancel out unwanted 1. reflections from the speaker/room interaction. I will not go into deeper details regarding DRC as it has been covered numerous times elsewhere, for ex. here.

Acourate Digital Room and Loudspeaker Correction Software Walkthrough

In this article, I walk through the steps using Acourate to produce a default or baseline correction that is repeatable. By following the same steps, one should be able to achieve a similar baseline correction. This baseline correction is designed to provide the listener with a perceptually flat...

audiophilestyle.com

If one want to get really nerdy

DRC: Digital Room Correction

a little taste on what DRC does

1. Initial windowing and normalization of the input impulse response.
2. Optional microphone compensation.
3. Initial dip limiting to prevent numerical instabilities during homomorphic deconvolution.
4. Decomposition into minimum phase and excess phase components using homomorphic deconvolution.
5. Prefiltering of the minimum phase component with frequency dependent windowing.
6. Frequency response dip limiting of the minimum phase component to prevent numerical instabilities during the inversion step.
7. Prefiltering of the excess phase component with frequency dependent windowing.
8. Normalization and convolution of the preprocessed minimum phase and excess phase components (optional starting from version 2.0.0).
9. Impulse response inversion through least square techniques or fast deconvolution.
10. Optional computation of a psychoacoustic target response based on the magnitude response envelope of the corrected impulse response.
11. Frequency response peak limiting to prevent speaker and amplification overload.
12. Ringing truncation with frequency dependent windowing to remove any unwanted excessive ringing caused by the inversion stage and the peak limiting stage.
13. Postfiltering to remove uncorrectable (subsonic and ultrasonic) bands and to provide the final target frequency response.
14. Optional generation of a minimum phase version of the correction filter.
15. Final optional test convolution of the correction filter with the input impulse response.

Final thoughts on DRC, I can only say, done correctly, it will be the biggest upgrade you can ever do to your sound and it works perfectly when doing MSO and then run DRC on top.

 

[ernestcarl]

Fulcrum acoustics might do what you mention about the horn... in this case a compression driver:

Although it cannot totally eliminate it. I've noticed this with the Sceptre S8 where you can still see some "echo" in the power cepstrum and even in wavelet measurements -- but it reduces this enough where the horn doesn't really sound so obviously "shouty" and colored like what one might initially expect.

 

[ernestcarl]

DRC makes anti-pulses to cancel out unwanted 1. reflections from the speaker/room interaction.

At what bandwidth and how big of an optimum area are you talking here?

 

[johnp98]

If one want to get really nerdy

DRC: Digital Room Correction

a little taste on what DRC does

Does anyone here have experience with Denis Sbragion free DRC?

As I am about to take the deep dive into it, but just wondering if anyone has also gotten into it.
Will post results when I actually do my own DRC.

 

[ernestcarl]

Does anyone here have experience with Denis Sbragion free DRC?

I have only tried the DRCDesigner.

Personally, haven't found the need for anything more sophisticated than what REW and rephase can already do in stereo and MCH as of the moment. My more "manual" way of doing things was also not exactly an easy learning path -- lots of experimentation, requiring a lot of time. It may be possible to get more or less similar results with other more automated software.

If I were to upgrade to something beyond the basic 5.1 or 7.1 (e.g. Atmos), then I'd use the AVR's DRC and more advanced upmixing since it's already included in the cost of the device anyway.

Will post results when I actually do my own DRC.

I'd be very interested in seeing the actual correction filters it generates -- both phase and magnitude.

 

[schlager]

At what bandwidth and how big of an optimum area are you talking here?

Guess it works at any bandwidth that is in question regarding 1. reflections and it works over a wide area. Don't take my word for it, just do an before-after measurement in REW and look at the filtered IR tab or at the Impulse response tab, you should see a clear improvement, with less spikes in the first 2 to 10 msek.

DRC Designer can do a better job at cleaning up the impulse response, than using REW+Rephase aka the manuel way. But I don't always find the standard filter templates working the best, so I use the custom filters and fiddel around trying different time windows for the filters and the compare them, listening to them. One very important aspect of making a filter to your liking, is having the right target curve, which is set in DRC Designer. When I find a filter I like, I take measurements in REW to confirm that the filters are working properly, looking at frequency response, IR, step response, waterfall plots, spectrogram etc.

It can be a tedious process going forth and back changing and tweaking parameters and then "reviewing" the filter. The filter I'm using now took me about 2 month to arrive at and I probably had about 50 filters through the convolver, but it was well worth the effort and I also learned a lot in the process.

 

[schlager]

ernestcarl, you can drag and drop the DRC correction file into REW for further analysis.

 

[ernestcarl]

Guess it works at any bandwidth that is in question regarding 1. reflections and it works over a wide area. Don't take my word for it, just do an before-after measurement in REW and look at the filtered IR tab or at the Impulse response tab, you should see a clear improvement, with less spikes in the first 2 to 10 msek.

I ask this question because even Mitch has stated the ff:

"One aspect of DRC that may not be so obvious is that we are adjusting the amplitude of the frequency response over time. And usually, just the low-frequency response as some room reflections above the transition frequency is a good thing."

I took multiple measurements at the MLP of my couch and the spikes/reflections in the higher frequencies will move in time as one moves the microphone position even by only a few centimeters:

Taken at the MLP only

How is one supposed to correct this consistently, and effectively over a wide listening area? Hmmmn... I don't think you can, esp. as one goes up higher in frequency.

The filter I'm using now took me about 2 month to arrive at and I probably had about 50 filters through the convolver, but it was well worth the effort and I also learned a lot in the process.

Where did you add-in your filters, or rather, what application(s) did you use to create them?

 

[schlager]

ernestcarl, you could try using a larger time window (FDW) for the DRC program, to treat those reflections.

I use Equalizer APO to handle the correction filter, from DRC Designer.

 

[ernestcarl]

ernestcarl, you could try using a larger time window (FDW) for the DRC program, to treat those reflections.

I use Equalizer APO to handle the correction filter, from DRC Designer.

But which reflections? There are many and the target changes as I sway my body forward and backward and/or lean left to right:

As one moves closer to the side-walls at the edges of the couch, well away from the center MLP, the reflections become stronger.

Nevertheless, the more important energy peak of the direct sound is still far stronger than any of these early and late reflections -- just observe/examine the spectral decay and wavelet graphs.

What's more concerning to me is how all the reflections are contributing to the overall spectral characteristic decay over time. If it's largely even across the bandwidth (esp. the mids), then it means we've got nothing to worry about. I have checked at all positions across the couch and everything is very even, and I found nothing particularly sticking out to be highly concerned about.

 

[schlager]

Logic tells us that the room will still have reflections after DRC, as the room is still there. DRC only treat reflection, within a certain time window, captured by the microphone in your listening position (LP) or wherever you measure from. So to really see the effect of DRC, you should take one measurement in your LP before DRC and another measurement after DRC, without moving the mic. If you want to treat more and later reflections, then acoustic absorption and diffusing is the way to go.

 

[ernestcarl]

ernestcarl, you can drag and drop the DRC correction file into REW for further analysis.

It's been a while since I last used DRC Designer so I performed a quick check on its output (based only in single point measurements).

In a nearfield setup of less than a meter (~0.8m), I find the default ERB and minimal generated filters doing too much.

I have the anechoic curves for my desk monitors (KH120), so I know what the direct on-axis and LW response should look like.

I was surprised how much boost it applied to the low bass (no obvious ability to manually manipulate the target curve in the GUI it seems besides the ready made presets)... but since I'm also trying to preserve the headroom of the speakers this was rather, again, just too much. Adjusting the sliders in custom mode may improve things a bit, but I'm not really too invested with this software -- development has ceased -- so I think I'll just skip it.

I like mine better, as it does the least amount of manipulation.

Just the phase. What the default ERB and minimal filters do to the FR above 1kHz -- well, it's not horrible, but it would be wise to rather just ignore or completely disable...

 

[ernestcarl]

If you want to treat more and later reflections, then acoustic absorption and diffusing is the way to go.

Thanks, but I'm out of space for diffusion (personally prefer it) so the current setup will have to do for the time being.

 

[sarumbear]

As I described in my earlier Widescreen Review Magazine article on low frequency/subwoofer optimization, SFM is a powerful tool in extracting more performance out of multiple subwoofers in arbitrary rooms.

@amirm the link above no longer works. I wonder if there’s a new one that works?

 

[schlager]

seem like you are boosting the lows too much, as the distortion is sky rocking. You can try and lower the maximum correction boost in the custom settings and/or roll off your target curve in the low end to prevent too much boost.

 

[ernestcarl]

seem like you are boosting the lows too much, as the distortion is sky rocking. You can try and lower the maximum correction boost in the custom settings and/or roll off your target curve in the low end to prevent too much boost.

I would prefer it did not apply any in there. I checked with the custom setting value set to 0dB gain and still there is some applied gain and/or significant total loss of headroom. The monitors are positioned in a way that there is significant loss of bass below 100Hz -- this was intentional.

 

[schlager]

I can only advise you to roll off the target curve below 80hz or so in DRC Designer.