Comparison of DRCs: Dirac Live for Studio, IK Multimedia ARC System 3 and Sonarworks Reference 4 Studio edition

[mitchco]

@dominikz very nice. Can you share the Correction Procedure Designer settings you used for Audiolense? Depending on what settings you used will have a major impact on the (measured and listening) results in the same way one can vary the target curve and hear significant differences. This is the advantage Audiolense and Acourate (and the open source DRC) have over other DRC products that don't have this capability. However, it takes time to understand what the settings control and how that translates into what one hears as it can have a significant impact on the sound quality.

With respect to single point versus multi-point measurements, one should take into consideration how each DRC product is "designed" to work. For example: https://www.audiosciencereview.com/...-altitude-jbl-sdp-75.13095/page-9#post-621360

To illustrate a bit, at least in the frequency domain, here is a direct loopback measurement through JRiver of Dirac's correction filter used in this article on top. On bottom is another DRC product using a FIR correction filter for the same speaker (both left channel). I drew a red line just above the FIR correction filter as I know that is 0 dBFS as I can inspect the FIR filter directly and measure it's peak amplitude (and group delay) in another application.

Note that the Dirac Live correction ontop was using 17 measurement points and the mystery DSP using a single point measurement on the bottom. As I can't inspect the Dirac correction filter directly, or know what is going on in the DiracLiveProcessor (i.e. is it "just" a convolution engine or is there additional processing?) it is difficult to draw any real conclusions. The only conclusion I can draw is that the two filters are going to sound different.

If we turn our attention to the time domain, I see you have some step response comparisons, but your horizontal scale only goes out to about 7ms or so. While that is great to show the time alignment capabilities it is missing the raison d'etre of DRC, which is being able to smooth out the bass response over time. This is what what the low frequency excess phase correction is designed for and one of the "key" differences between these DRC products (i.e. most don't have it). For example, in my system here is the step response, but over a much longer time frame:

Aside from the time alignment, we see no low frequency room reflections interfering with the "ideal" minimum phase response over a long time frame. Looking at the corresponding frequency response:

To me, this is the ultimate test of any DRC product. Drop a mic at your listening position and use REW's default 500ms window with no smoothing and the correction filter in circuit. (PS. and it is just not at one mic location as I have other posts on ASR that show the same results using 14 measurements over a 6ft x 2ft grid area). Let all the reflections in As we can see, below the room's transition frequency, there are no low frequency room reflections getting into the measurement (or technically achieving the ideal minimum phase response). Not only is the low frequency response smooth, but it is crystal clear. This is the major differentiator. Based on the number of systems I have received measurements for, I don't think many folks have really heard what clear bass over time sounds like. This to me is the deciding factor on how good a DRC is at "room eq". Aside from the clarity of the bass, it does not sound "overcorrected" as described in the Trinnov link posted above.

Again, good comparison, but I think you are just scratching the surface wrt tools like Audiolense (and Acourate and like @dasdoing mentions DRC) can achieve. You should be getting results similar to what I show above and in this article on what is accurate sound.

Keep up the good work!

 

[dominikz]

@dominikz very nice. Can you share the Correction Procedure Designer settings you used for Audiolense? Depending on what settings you used will have a major impact on the (measured and listening) results in the same way one can vary the target curve and hear significant differences. This is the advantage Audiolense and Acourate (and the open source DRC) have over other DRC products that don't have this capability. However, it takes time to understand what the settings control and how that translates into what one hears as it can have a significant impact on the sound quality.

Thanks! I just used the 'Default (True Time Domain)' profile:

With respect to single point versus multi-point measurements, one should take into consideration how each DRC product is "designed" to work. For example: https://www.audiosciencereview.com/...-altitude-jbl-sdp-75.13095/page-9#post-621360

Absolutely, didn't mean to say or imply single-point measurements don't have their place, just that:
1) I seem to be getting slightly worse results with single-point in one of my systems (though I fully admit it might just be lack of competence on my side ), and
2) I'm having a hard time understanding how could a very flat measured FR at a single point ever translate to similarly flat measured FR across multiple points - it seems to me one would have to see at least some peaks at other points. The reason why I assume this is that to get a very flat measured FR at a single-point, one has to fill-in dips and reduce peaks that occur at that point in space, but that may not exist at other points - in which case notch filters will generate their own dips and boosts their own peaks. That doesn't mean of course that the SW couldn't be smarter and somehow understand from a single point which peaks and dips to leave alone, which is what I believe you're arguing - though in that case the FR wouldn't be ruler-flat anymore (which is fine anyway).

As I can't inspect the Dirac correction filter directly, or know what is going on in the DiracLiveProcessor (i.e. is it "just" a convolution engine or is there additional processing?) it is difficult to draw any real conclusions.

Just wanted to add that although we can't know exactly how the internal processing is done, based on what I've seen and what the SW is meant to do, I'd say that it should be safe to assume it is in practice a linear, time-invariant (LTI) process. Signal processing theory tells us that LTI systems are fully described by their impulse response / transfer function.

If we turn our attention to the time domain, I see you have some step response comparisons, but your horizontal scale only goes out to about 7ms or so. While that is great to show the time alignment capabilities it is missing the raison d'etre of DRC, which is being able to smooth out the bass response over time. This is what what the low frequency excess phase correction is designed for and one of the "key" differences between these DRC products (i.e. most don't have it).

Mine definitely doesn't look as nice

Vs uncorrected:

I guess part of the reason is my system not playing as low, but still the reflections in my system seem more severe - with or without correction.
Here's also a plot of the RT20 from MLP (without correction) for reference:

Not sure if the peak at 62,5Hz is to be trusted, though - the result changes between measurements by quite a lot.

PS. and it is just not at one mic location as I have other posts on ASR that show the same results using 14 measurements over a 6ft x 2ft grid area

Would you by any chance be able to repost your FR measurements of this by any chance? Very interested in this but I only managed to find the step responses, not the FR. Thanks!

Again, good comparison, but I think you are just scratching the surface wrt tools like Audiolense (and Acourate and like @dasdoing mentions DRC) can achieve. You should be getting results similar to what I show above and in this article on what is accurate sound.

Keep up the good work!

Thanks again! Absolutely I don't see myself as any kind of power user (though I wouldn't say I'm completely clueless either) - just hoping my experiences would help casual users as I don't think many will consider doing a deep dive into room EQ and DSP theory.

 

[mitchco]

Hi @dominikz

Thanks! I just used the 'Default (True Time Domain)' profile:

Yes, this is a point of departure from other DSP/DRC software and entering the realm of custom designing FIR filters to match speakers to rooms with a high degree of accuracy and precision. Adjusting these controls makes the difference between a good correction and near perfect text book response. Part of the equation is understanding the dimensions of ones listening space, where the transition frequency is and using Frequency Dependant Windowing (FDW) math to figure out how much excess phase correction is applied at low frequencies and how little after the transition frequency. Aside from partial correction, both in the frequency and time domain and using midband filtering in the transition zone when the wavelengths transition from waves to rays can also be useful. Lots of parameters to play with...

Absolutely, didn't mean to say or imply single-point measurements don't have their place, just that:
1) I seem to be getting slightly worse results with single-point in one of my systems (though I fully admit it might just be lack of competence on my side ), and
2) I'm having a hard time understanding how could a very flat measured FR at a single point ever translate to similarly flat measured FR across multiple points - it seems to me one would have to see at least some peaks at other points.

There is some, as you will see below, but it is important to understand the type of analysis that is being performed.

ust wanted to add that although we can't know exactly how the internal processing is done, based on what I've seen and what the SW is meant to do, I'd say that it should be safe to assume it is in practice a linear, time-invariant (LTI) process. Signal processing theory tells us that LTI systems are fully described by their impulse response / transfer function.

Yes to LTI, but it is still a black box and additional filtering could be applied for all we know. It would be nice to inspect the filter directly.

Mine definitely doesn't look as nice

Yes, another point of departure to move away from the default settings. This will get you much closer, but does take time. I have been working with DSP/DRC software for 10 years and have used on a couple hundred systems. There is definitely a design pattern to match virtually any speaker to any room to get the results I have shown.

Not sure if the peak at 62,5Hz is to be trusted, though - the result changes between measurements by quite a lot.

T20 is good to look at broadband decay times down to 100 Hz or around the rooms transition frequency. REW's waterfall display provides a better display for looking at low frequency decay times.

Would you by any chance be able to repost your FR measurements of this by any chance? Very interested in this but I only managed to find the step responses, not the FR. Thanks!

This set of frequency responses belongs to these step responses:

The chapter Filter Design Verification in my DSP book goes into gory detail about these measurements that were based on a single analysis measurement in Acourate. I was advised at the time to use psychoacoustic smoothing "to produce a plot that more closely corresponds to the perceived frequency response." I think 1/6 octave smoothing is generally close enough to follow the envelope of what we are hearing.

Another system, currently in my room, that matches these sets of step responses:

Ignore the overall slope as I was experimenting with sloped targets and this one is (way) too steep. Again, this is based on one analysis measurement, this time in Audiolense. The chart shows how smooth the response is, even across a 6ft area. The speakers are in a 9ft equilateral triangle. So with the mic positioned 3 ft to the left of center and measuring the right speaker, it is actually some 12ft away and represents the extreme ends on my 6ft couch. When I say I don't hear any tonal changes across my couch area, the chart verifies. It does help that I am using a constant directivity waveguide above 630 Hz.

As a side note, it is unfortunate that there is so much disinformation about DRC on the interwebs. Stuff like moving the mic 6" produces a completely different response. Or single point analysis measurements can't be as good as multipoint analysis measurements - as seen here, simply not so. Or for the love of Pete don't ever eq a speaker above the transition frequency as you will wreck the off axis response. Maybe using crappy speakers and/or crappy DRC, perhaps so, but certainly not in my case as shown above. In fact, I have a number of clients that use the Revel Salon2 in a variety of rooms. Each one received up to 6 correction filters to listen to. All had partial corrections up to the rooms transition frequency in addition to full range corrections. Not one chose the partial correction. They all preferred the full range correction.

Thanks again! Absolutely I don't see myself as any kind of power user (though I wouldn't say I'm completely clueless either) - just hoping my experiences would help casual users as I don't think many will consider doing a deep dive into room EQ and DSP theory.

Like I say you did a great great job! I just wanted to point out that where most DRC's end in features and performance, is just the beginning for Audiolense and Acourate. These are SOTA custom FIR filter designers and the others are not. It is the difference between a"general purpose" solution versus obtaining near perfect textbook responses both in measurements and listening result. There is a big difference.

 

[Rottmannash]

Hi @dominikz



Yes, this is a point of departure from other DSP/DRC software and entering the realm of custom designing FIR filters to match speakers to rooms with a high degree of accuracy and precision. Adjusting these controls makes the difference between a good correction and near perfect text book response. Part of the equation is understanding the dimensions of ones listening space, where the transition frequency is and using Frequency Dependant Windowing (FDW) math to figure out how much excess phase correction is applied at low frequencies and how little after the transition frequency. Aside from partial correction, both in the frequency and time domain and using midband filtering in the transition zone when the wavelengths transition from waves to rays can also be useful. Lots of parameters to play with...



There is some, as you will see below, but it is important to understand the type of analysis that is being performed.



Yes to LTI, but it is still a black box and additional filtering could be applied for all we know. It would be nice to inspect the filter directly.



Yes, another point of departure to move away from the default settings. This will get you much closer, but does take time. I have been working with DSP/DRC software for 10 years and have used on a couple hundred systems. There is definitely a design pattern to match virtually any speaker to any room to get the results I have shown.



T20 is good to look at broadband decay times down to 100 Hz or around the rooms transition frequency. REW's waterfall display provides a better display for looking at low frequency decay times.



This set of frequency responses belongs to these step responses:

View attachment 106523

The chapter Filter Design Verification in my DSP book goes into gory detail about these measurements that were based on a single analysis measurement in Acourate. I was advised at the time to use psychoacoustic smoothing "to produce a plot that more closely corresponds to the perceived frequency response." I think 1/6 octave smoothing is generally close enough to follow the envelope of what we are hearing.

Another system, currently in my room, that matches these sets of step responses:

View attachment 106524

Ignore the overall slope as I was experimenting with sloped targets and this one is (way) too steep. Again, this is based on one analysis measurement, this time in Audiolense. The chart shows how smooth the response is, even across a 6ft area. The speakers are in a 9ft equilateral triangle. So with the mic positioned 3 ft to the left of center and measuring the right speaker, it is actually some 12ft away and represents the extreme ends on my 6ft couch. When I say I don't hear any tonal changes across my couch area, the chart verifies. It does help that I am using a constant directivity waveguide above 630 Hz.

As a side note, it is unfortunate that there is so much disinformation about DRC on the interwebs. Stuff like moving the mic 6" produces a completely different response. Or single point analysis measurements can't be as good as multipoint analysis measurements - as seen here, simply not so. Or for the love of Pete don't ever eq a speaker above the transition frequency as you will wreck the off axis response. Maybe using crappy speakers and/or crappy DRC, perhaps so, but certainly not in my case as shown above. In fact, I have a number of clients that use the Revel Salon2 in a variety of rooms. Each one received up to 6 correction filters to listen to. All had partial corrections up to the rooms transition frequency in addition to full range corrections. Not one chose the partial correction. They all preferred the full range correction.



Like I say you did a great great job! I just wanted to point out that where most DRC's end in features and performance, is just the beginning for Audiolense and Acourate. These are SOTA custom FIR filter designers and the others are not. It is the difference between a"general purpose" solution versus obtaining near perfect textbook responses both in measurements and listening result. There is a big difference.

Were you speaking English? May I ask what the transitional frequency is?

 

[mitchco]

Were you speaking English? May I ask what the transitional frequency is?

It is the rooms transition frequency, also known as the Schroeder frequency. Figure 1 in the 2nd link is a great view of what is going on in a room. You can calculate your own rooms transition frequency by using one of many online room mode calculators like this one: https://amcoustics.com/tools/amroc

Another picture to illustrate:

 

[dominikz]

Part of the equation is understanding the dimensions of ones listening space, where the transition frequency is and using Frequency Dependant Windowing (FDW) math to figure out how much excess phase correction is applied at low frequencies and how little after the transition frequency. Aside from partial correction, both in the frequency and time domain and using midband filtering in the transition zone when the wavelengths transition from waves to rays can also be useful. Lots of parameters to play with...

Thanks for the link, I'll be sure to give it a read! I do believe I understand what FDW does in principle, and I did play a bit with the settings to see how it changes the result, but I fully admit I don't know how to tune it for best results (which is why I used the default profile for my comparison).

Yes to LTI, but it is still a black box and additional filtering could be applied for all we know. It would be nice to inspect the filter directly.

Though if we agree that it is LTI then we know that the impulse response / transfer function completely describes what the system does to any input signal. However I assume your comment is meant to say that we don't know how the filters themselves are implemented (e.g. IIR, FIR or some combination of the two, how they do their averaging and smoothing of responses etc.) which I of course agree with completely. It's just that I'm not sure that is really important to know for the purpose of a functional comparison such as this one, since we have a full functional description of the calculated correction with the impulse response / transfer function measurement.

This set of frequency responses belongs to these step responses:

First of all, thanks a lot for reposting all these measurements! I've adapted my diagram scale and smoothing to match yours to compare, seems there's some similarity as you move away further from the MLP, e.g.:

vs:

Though your definitely looks tidier, seems like we both get some peaks as we move further from the MLP (in your case 3ft positions, in my around 2.5ft to the side MLP) - this is what I was describing - though obviously I didn't do the best version of the correction and I can see how therefore the peaks in my case could have been more audibly offending.

Anyway, thanks for all your suggestions and insights - very useful!

 

[richard12511]

Or for the love of Pete don't ever eq a speaker above the transition frequency as you will wreck the off axis response. Maybe using crappy speakers and/or crappy DRC, perhaps so, but certainly not in my case as shown above. In fact, I have a number of clients that use the Revel Salon2 in a variety of rooms. Each one received up to 6 correction filters to listen to. All had partial corrections up to the rooms transition frequency in addition to full range corrections. Not one chose the partial correction. They all preferred the full range correction.
.

I think this thought is so widespread (especially here) because it was said by Floyd Toole. The vast majority here advise against EQing above the transition frequency, so it definitely is a commonly held opinion.

With the help of a friend, I was able to compare Dirac full range correction against Dirac to 650Hz correction this weekend. To my surprise, I actually preferred full range correction, and really consistently. Wasn't a huge difference, but I somehow still (fairly) consistently preferred it. I've always preferred the limited range corrections in sighted comparisons, so this came as a bit of a surprise. I guess that means there is an expectation bias in play for me, likely since I know Toole's recommendation. It's not due to bad speakers, either (still waiting for my center channel ), as the speakers are textbook Harman style speakers. I do wonder how much of Floyd's advice was based on the DRC available back then, and would he still have the same advice with software like Audiolense and Acourate in the mix?

I'd be really curious to see others try the same blind test, as I was truly surprised. I know @Steve Dallas is messing around with full range vs sub 1000(?)Hz Dirac correction atm.

 

[Χ Ξ Σ]

@mitchco
Hi Mitch, Are you familiar with these acoustic processors? http://en.sinemedia.com/products. From the manufacturer's write-up they use up to 16384 FIR filter taps, similar to Audiolense and Accurate. These peocesor units cost a few thousand US dollars whereas the two softwares costs a few hundreds. I think it make sense to have a physical processor unit in those use cases, and these processors might be automated so it's more convenient. But other than that, should I expect similar performance from these processors and Audiolense/Accurate given that they have similar amount of filter taps?

 

[Χ Ξ Σ]

Incredible work by @dominikz ! As a Dirac user, I did not bother to try other DRC products during my trial, but I have been curious about what the others can do. you experiment made me very happy about my purchase.

 

[Robbo99999]

I think this thought is so widespread (especially here) because it was said by Floyd Toole. The vast majority here advise against EQing above the transition frequency, so it definitely is a commonly held opinion.

With the help of a friend, I was able to compare Dirac full range correction against Dirac to 650Hz correction this weekend. To my surprise, I actually preferred full range correction, and really consistently. Wasn't a huge difference, but I somehow still (fairly) consistently preferred it. I've always preferred the limited range corrections in sighted comparisons, so this came as a bit of a surprise. I guess that means there is an expectation bias in play for me, likely since I know Toole's recommendation. It's not due to bad speakers, either (still waiting for my center channel ), as the speakers are textbook Harman style speakers. I do wonder how much of Floyd's advice was based on the DRC available back then, and would he still have the same advice with software like Audiolense and Acourate in the mix?

I'd be really curious to see others try the same blind test, as I was truly surprised. I know @Steve Dallas is messing around with full range vs sub 1000(?)Hz Dirac correction atm.

In your tests Richard, did you EQ above transition frequency using Anechoic EQ? I know you normally do that, so I wondered if you were comparing DIRAC below transition + Anechoic EQ above transition against DIRAC across the whole frequency range?

 

[dominikz]

I think this thought is so widespread (especially here) because it was said by Floyd Toole. The vast majority here advise against EQing above the transition frequency, so it definitely is a commonly held opinion.

With the help of a friend, I was able to compare Dirac full range correction against Dirac to 650Hz correction this weekend. To my surprise, I actually preferred full range correction, and really consistently. Wasn't a huge difference, but I somehow still (fairly) consistently preferred it. I've always preferred the limited range corrections in sighted comparisons, so this came as a bit of a surprise. I guess that means there is an expectation bias in play for me, likely since I know Toole's recommendation. It's not due to bad speakers, either (still waiting for my center channel ), as the speakers are textbook Harman style speakers. I do wonder how much of Floyd's advice was based on the DRC available back then, and would he still have the same advice with software like Audiolense and Acourate in the mix?

I'd be really curious to see others try the same blind test, as I was truly surprised. I know @Steve Dallas is messing around with full range vs sub 1000(?)Hz Dirac correction atm.

Just wanted to share some thoughts on this.
In my tests I thought the difference in tonality between partial and full range correction was pretty small as long as I used a reasonable target and I could live with it either way - it's just that my philosophy is in general to pick the variant that does less processing if both options sound fine to me.

However, from what I've seen, if the measurement position is in the diffuse field the DRCs I tested weren't able to extract the loudspeakers real anechoic on-axis response from the steady-state in-room measurement, regardless of FDW. This would imply a tonality change if EQing above the transition frequency (note that I'm not arguing whether this would be detrimental).

Let me show what I mean on an example - the next diagram shows:
1) Filters generated by Dirac Live and Audiolense for just the left channel, targeting the Olive/Toole slope of -10dB 20Hz-20kHz
2) Anechoic Listening Window (LW) measurements of M16 from two sources (ASR and Revel)
3) Combined filter + LW responses to show the simulated anechoic direct sound of the loudspeaker after the correction is applied.

Of course we can disregard the LF room correction part, but in the mid and HF we can see that the filtered LW responses deviate a little bit (especially the mid range bump at 2.5kHz) from the pretty flat anechoic LW measurements of these speakers.
The reason for the bump in the response is the fact that the steady-state measurement shows the dip there, but my point is that the dip is less pronounced in the actual anechoic LW response.
If I'm not misinterpreting something, this would speak in favor of Toole's argumentation.

 

[thewas]

With the help of a friend, I was able to compare Dirac full range correction against Dirac to 650Hz correction this weekend. To my surprise, I actually preferred full range correction, and really consistently. Wasn't a huge difference, but I somehow still (fairly) consistently preferred it.

Would be interesting to see the Dirac measured FR and target curve used. In m experience sometimes a higher target slope than the measured one can make a system sound more euphonic, especially with some rather bass lacking recordings.

 

[mitchco]

I think this thought is so widespread (especially here) because it was said by Floyd Toole. The vast majority here advise against EQing above the transition frequency, so it definitely is a commonly held opinion.

Hi Richard, for sure. My examples of the Revel Salon2's are part of a larger metric. The reality is that 99% of my clients prefer a full range correction over a partial correction. It is a fairly decent sample size given the number of clients. Each client has submitted clean IR's, sometimes taking multiple tries , room ratios, pictures of the room, descriptions of the gear and loudspeakers, room treatment if any, etc. It's a lot of data. I also have all of the target responses, corrections, etc. Some interesting trends and correlations present themselves. Mostly contradicting what is known as conventional wisdom. I intend to publish the anonymized data in the 2nd edition of my book with the analysis and overlays of the measurements and target responses. I think most will see the patterns that have emerged regardless of loudspeaker or room.

Hi Mitch, Are you familiar with these acoustic processors? http://en.sinemedia.com/products. From the manufacturer's write-up they use up to 16384 FIR filter taps, similar to Audiolense and Accurate. These peocesor units cost a few thousand US dollars whereas the two softwares costs a few hundreds. I think it make sense to have a physical processor unit in those use cases, and these processors might be automated so it's more convenient. But other than that, should I expect similar performance from these processors and Audiolense/Accurate given that they have similar amount of filter taps?

Hello @Χ Ξ Σ Thanks for the link. I had a look but I could not see which one is using 16,384 taps...? What I saw was 4096 or 8192 taps... I could not find any technical specifications at all on that site Any links? Audiolense and Acourate use 65,536 taps or 131,072 or more.

The reason for the bump in the response is the fact that the steady-state measurement shows the dip there, but my point is that the dip is less pronounced in the actual anechoic LW response.

Are you sure? If you were using the default correction window in Audiolense, I can say that the direct sound plus reflections at 2.5 kHz are being corrected. Tsk Tsk Same goes with Dirac, except you can't change any of the windowing parameters in Dirac.

 

[Χ Ξ Σ]

@mitchco I saw it on the individual product page after clicking on a product.
Whoa, I read about how many filter taps that Audiolense and Acourate have from your comments but the numbers just did not stick with me... They are 10 times what I remembered...

 

[Χ Ξ Σ]

I intend to publish the anonymized data in the 2nd edition of my book with the analysis and overlays of the measurements and target responses.

When can we expect the 2nd edition of your book?

 

[dominikz]

Are you sure? If you were using the default correction window in Audiolense, I can say that the direct sound plus reflections at 2.5 kHz are being corrected. Tsk Tsk Same goes with Dirac, except you can't change any of the windowing parameters in Dirac.

My reasoning was that since the speaker is in the diffuse field of this room, the 2,5kHz dip (and following boost) in the measured in-room response were the product of uneven vertical directivity and resulting vertical reflections summing with the on-axis response (and the nicer looking horizontal reflections) at MLP:

The basic shape/trend of Audiolense filters in this part of the response doesn't seem to change much with significant changes to the FDW (here shown from 0.1 cycles to 15 cycles, see graph legend for actual LF and HF values used)

We can see the small boost around 2,5kHz (visible in 1-15 cycle setting diagrams) followed by the wide dip (visible with all cycle settings) - correlating with the reflections mentioned above.

In addition, if we're only aiming to EQ the direct sound and not reflections at HF (by adapting FDW correspondingly), the basic shape of the in-room steady-state response should after correction be similar to the one before - i.e. the measured response still wouldn't be as flat as the target (due to the character of the off-axis reflection at MLP). But all of the above filters would change the measured response at HF to be closer to the target flat slope at MLP (i.e. they would compensate somewhat for vertical reflections).

Hope this explains a bit more what I meant! Of course, perhaps I'm overlooking something - any insight would be much appreciated!

 

[dominikz]

Hope this explains a bit more what I meant! Of course, perhaps I'm overlooking something - any insight would be much appreciated!

Hmmm, perhaps what I'm overlooking is the height of my speakers vs the MLP - it is at about 6-7° downward angle with regard to tweeter axis. Will look into this a bit more.

 

[dasdoing]

I think you are just scratching the surface wrt tools like Audiolense (and Acourate and like @dasdoing mentions DRC) can achieve.

people are not quiet believing in phase correction. it might be true that not everybody will hear the delays,
but the direct sound is also affected by phase:
https://www.audiosciencereview.com/...ase-distortion-abx-testing.18709/#post-657865

 

[landco]

Sonarworks Reference 4 Studio edition - I couldn't make myself agree with this one Sure, there are some nice features there (plus great systemwide version and some extensive headphone EQ options) so I'm sure it works great for many - but lack of filter sharpness, true correction range limiting and detailed target curve editing are for now deal-breakers for me.

In Sonarworks, there is a ready-made preset for stereo Hi-Fi listening in a typical room, this preset is designed by professionals and gives a good result in any typical room:

Dirac and ARC don't have this preset, which is bad, because it forces you to adjust the sound "according to the feeling", and not in the right way.

 

[sejarzo]

If you set a house curve that is flat to 200Hz, -3 dB at 2000Hz, and -6 dB at 20kHz, that's essentially the B&K 1974 target within a fraction of a dB.

Not very difficult.