Jump to content
IGNORED

Article: Calibrating Desktop Speakers using Focus Fidelity Filter Designer


Recommended Posts

Another great article, @mitchco! My question is:

 

Does applying speaker EQ from anechoic data before running the room correction software improve the end result? Or does a SOTA room correction  software obviate the need to apply anechoic speaker EQ?

 

The iLoud MTM has been measured by Amir so we have the full anechoic speaker data. So we can EQ the CEA-2034 listening window (LW) curve for nearfield listening, or the predicted in room response curve (PIR) for far field.

 

For example, user Maiky76 provided these EQs, which can be used at a starting point. For iLoud MTM desktop use, you could apply the listening window EQ settings, sticking to above 700hz or so, and taking care to avoid overly narrow EQ adjustments. I would personally skip the large 4.2db boost at 18.2kHz, leaving only a few small PEQ adjustments that are between 1.4 to 2.3 dB of cut or boost.

 

Then, apply the SOTA room correction as normal on top of this speaker “pre-eq,” starting with a more accurate speaker baseline.

 

Having anechoic speaker data is becoming more common thanks to Amir and Erin.

 

But does a SOTA render this unnecessary?

 

Finally, I will note that high end processors such as JBL Synthesis SDP-75 and Trinnov Altitude support loading anechoic speaker EQ before applying Trinnov room eq. And Storm audio supports this, but with Dirac as the room eq. Vendors such as Wisdom Audio and JBL Synthesis provide their anechoic speaker eq for all these platforms.

 

So having axes to publicly posted anechoic speaker data as with the iLoud MTM could be a great gift—that is if adds value in combination with SOTA room eq.

 

Link to comment
6 hours ago, Yourmando said:

Another great article, @mitchco! My question is:

 

Does applying speaker EQ from anechoic data before running the room correction software improve the end result? Or does a SOTA room correction  software obviate the need to apply anechoic speaker EQ?

 

The iLoud MTM has been measured by Amir so we have the full anechoic speaker data. So we can EQ the CEA-2034 listening window (LW) curve for nearfield listening, or the predicted in room response curve (PIR) for far field.

 

For example, user Maiky76 provided these EQs, which can be used at a starting point. For iLoud MTM desktop use, you could apply the listening window EQ settings, sticking to above 700hz or so, and taking care to avoid overly narrow EQ adjustments. I would personally skip the large 4.2db boost at 18.2kHz, leaving only a few small PEQ adjustments that are between 1.4 to 2.3 dB of cut or boost.

 

Then, apply the SOTA room correction as normal on top of this speaker “pre-eq,” starting with a more accurate speaker baseline.

 

Having anechoic speaker data is becoming more common thanks to Amir and Erin.

 

But does a SOTA render this unnecessary?

 

Finally, I will note that high end processors such as JBL Synthesis SDP-75 and Trinnov Altitude support loading anechoic speaker EQ before applying Trinnov room eq. And Storm audio supports this, but with Dirac as the room eq. Vendors such as Wisdom Audio and JBL Synthesis provide their anechoic speaker eq for all these platforms.

 

So having axes to publicly posted anechoic speaker data as with the iLoud MTM could be a great gift—that is if adds value in combination with SOTA room eq.

 

 

Since my link to the iLoud MTM EQ was not allowed in my post above, I will paste in a couple of experts from the EQ. All credit to ASR user Maiky76:

Quote

 

Score EQ LW: 5.76

with sub: 7.96

 

Score EQ Score: 6.27

with sub: 8.04

 

Code:

iLoud MTM APO EQ LW 96000Hz

December132020-110110

 

Preamp: -4.7 dB

 

Filter 1: ON PK Fc 47.3 Hz Gain 0 dB Q 1.09

Filter 2: ON PK Fc 817 Hz Gain 1.5 dB Q 3.35

Filter 3: ON PK Fc 1608 Hz Gain 2.5 dB Q 3.48

Filter 4: ON PK Fc 3248 Hz Gain 1.35 dB Q 3.02

Filter 5: ON PK Fc 8410 Hz Gain 2 dB Q 1.19

Filter 6: ON PK Fc 18250 Hz Gain 4.39 dB Q 2.08

 

 

I put the 47hz highpass & 18.2kHz eq that I wouldn't apply in strikeout. The high pass would be solved by dragging the left curtain up to 45hz or so like Mitch did, or crossing over to a subwoofer. The big boost at 18kHz is unnecessary for most ears, I would dare to say :) So we are left with 4 small anechoic speaker eq adjustments.

 

iLoud MTM Zoom EQ Design.png

iLoud MTM Zoom PIR-LW-ON.png

iLoud MTM Radar.png

Link to comment

 

23 hours ago, Yourmando said:

Does applying speaker EQ from anechoic data before running the room correction software improve the end result? Or does a SOTA room correction  software obviate the need to apply anechoic speaker EQ?

 

@Yourmando thanks for your comments. 

 

I have performed this myself, using the same anechoic data, and my tests show that it does not improve the end result. Of course, if one is not using room correction eq, it helps.

 

SOTA DRC/DSP, that is correctly applied, not only alleviates the need for anechoic speaker eq, but can be applied at a much finer resolution than a few PEQ filters can do. I.e. the response tracks closer to the ideal minimum phase response. In addition, the left and right channels have virtually identical frequency response which means the phantom stereo image is dead centre along with a number of other benefits as described in this article.

 

As you know, the much bigger problem to solve is the room itself - room modes, standing waves, non-minimum phase response in the low frequencies with +20 to 30 dB SPL response variances are large issues to deal with. Using SOTA DSP/DRC makes way for even sounding bass that is crystal clear. When I say SOTA room eq, I mean DSP that works both in the frequency and time domain, like David's Focus Fidelity as reviewed here.

 

The issue I have with h/w DRC (i.e. Trinnov, JBL Synthesis, Storm et al) is that the onboard DRC has very little resolution below 100 Hz where there are only 2 bands of eq. So right where you need the power of DRC those devices have none. You can find a comparison along with the math behind it if you search my threads a bit.

 

Don't get me wrong, I appreciate the anechoic data and if you are not using room eq, it is beneficial. But if you are using SOTA room eq, there is no real benefit to applying eq using anechoic data before DRC.

 

Link to comment
On 4/19/2021 at 9:16 AM, mitchco said:

 

 

@Yourmando thanks for your comments. 

 

I have performed this myself, using the same anechoic data, and my tests show that it does not improve the end result. Of course, if one is not using room correction eq, it helps.

 

SOTA DRC/DSP, that is correctly applied, not only alleviates the need for anechoic speaker eq, but can be applied at a much finer resolution than a few PEQ filters can do. I.e. the response tracks closer to the ideal minimum phase response. In addition, the left and right channels have virtually identical frequency response which means the phantom stereo image is dead centre along with a number of other benefits as described in this article.

 

As you know, the much bigger problem to solve is the room itself - room modes, standing waves, non-minimum phase response in the low frequencies with +20 to 30 dB SPL response variances are large issues to deal with. Using SOTA DSP/DRC makes way for even sounding bass that is crystal clear. When I say SOTA room eq, I mean DSP that works both in the frequency and time domain, like David's Focus Fidelity as reviewed here.

 

The issue I have with h/w DRC (i.e. Trinnov, JBL Synthesis, Storm et al) is that the onboard DRC has very little resolution below 100 Hz where there are only 2 bands of eq. So right where you need the power of DRC those devices have none. You can find a comparison along with the math behind it if you search my threads a bit.

 

Don't get me wrong, I appreciate the anechoic data and if you are not using room eq, it is beneficial. But if you are using SOTA room eq, there is no real benefit to applying eq using anechoic data before DRC.

 


Thanks, @mitchco I’ve long wondered about whether applying anechoic speaker correction prior to room correction is beneficial.

 

Makes sense that a SOTA room correction doesn’t really benefit from the speaker correction data.

 

I do wonder why Trinnov, JBL Synthesis SFP-75, and Storm audio tout this as such a good feature to have.

 

Presumably, both Trinnov and Dirac, even on lower powered hardware like the high end processors, have plenty of taps/filters available to correct above Schroeder (vs a few PEQs that an anechoic “speaker tuning” would provide). And below Schroeder the speaker tunings don’t help because to room takes over.

 

Maybe there is a threshold where adding the above transition PEQs would help. (Such as hardware starved of taps, but not sure what number of taps that would be. Or maybe the anechoic tunings are just redundant and used as marketing hype.

 

Thanks for your thoughts and excellent article!

Link to comment
  • 3 weeks later...

I am just trying to improve my own understanding of how this software works, so apologies if these are basic questions.

 

Looking at the Focus Fidelity website, it shows this for the target response:

 

image.thumb.png.6fb2f11140f7c523223e972a0bc7b7db.png

 

In the above example, for the frequencies between 100Hz and 1000Hz, the majority of the in room measured response is below the target line.  So what does the software do here in terms of frequecy SPL correction?  Does it ignore this region or does it provide a "boost" to these frequencies?  From my limited experiments with REW creating relatively crude PEQ type correction, you essentially cut the peaks above the line but any "boosting" below the line is either ignored, or at least severely limited. 

 

In the above example, I can see that the peak at 50Hz or could be easily corrected, as could the peak at about 10kHz, it is just the 100Hz to 10000Hz zone that puzzles me.  Maybe I am misunderstanding how the target works?

 

Windows 11 PC, Roon, HQPlayer, Focus Fidelity convolutions, iFi Zen Stream, Paul Hynes SR4, Mutec REF10, Mutec MC3+USB, Devialet 1000Pro, KEF Blade.  Plus Pro-Ject Signature 12 TT for playing my 'legacy' vinyl collection. Desktop system; RME ADI-2 DAC fs, Meze Empyrean headphones.

Link to comment
On 5/10/2021 at 8:44 PM, Confused said:

I am just trying to improve my own understanding of how this software works, so apologies if these are basic questions.

 

Looking at the Focus Fidelity website, it shows this for the target response:

 

image.thumb.png.6fb2f11140f7c523223e972a0bc7b7db.png

 

In the above example, for the frequencies between 100Hz and 1000Hz, the majority of the in room measured response is below the target line.  So what does the software do here in terms of frequecy SPL correction?  Does it ignore this region or does it provide a "boost" to these frequencies?  From my limited experiments with REW creating relatively crude PEQ type correction, you essentially cut the peaks above the line but any "boosting" below the line is either ignored, or at least severely limited. 

 

In the above example, I can see that the peak at 50Hz or could be easily corrected, as could the peak at about 10kHz, it is just the 100Hz to 10000Hz zone that puzzles me.  Maybe I am misunderstanding how the target works?

 

Hi,

 

It will apply boost when the measured frequency response is below the target, the maximum amount of boost is adjustable (on the filter magnitude page). I should also point out that the software determines which dips/notches are not minimum phase and limits to the boost to those areas even further. This gives a more optimal result in the time domain.

 

Link to comment
9 hours ago, Focus Fidelity said:

Hi,

 

It will apply boost when the measured frequency response is below the target, the maximum amount of boost is adjustable (on the filter magnitude page). I should also point out that the software determines which dips/notches are not minimum phase and limits to the boost to those areas even further. This gives a more optimal result in the time domain.

 

Thanks for the response.  

 

This makes me wonder, I note the software includes phase corrections.  So where you have minimum phase induced dips / nulls (particularly in the bass region) do these phase corrections serve to mitigate the these dips / nulls (in terms of frequency vs SPL) , at least in part?

Windows 11 PC, Roon, HQPlayer, Focus Fidelity convolutions, iFi Zen Stream, Paul Hynes SR4, Mutec REF10, Mutec MC3+USB, Devialet 1000Pro, KEF Blade.  Plus Pro-Ject Signature 12 TT for playing my 'legacy' vinyl collection. Desktop system; RME ADI-2 DAC fs, Meze Empyrean headphones.

Link to comment
15 hours ago, Confused said:

Thanks for the response.  

 

This makes me wonder, I note the software includes phase corrections.  So where you have minimum phase induced dips / nulls (particularly in the bass region) do these phase corrections serve to mitigate the these dips / nulls (in terms of frequency vs SPL) , at least in part?

The software focuses on attenuating resonant modes and improving time domain criteria (ringing, step response, excess phase). Boosting a sharp dip/notch will result in a filter with significant ringing (pre or post depending on if the dip is minimum or maximum phase) so the software will seek to largely ignore the dip.  The relationship between phase, time domain and magnitude response gets rather complex and mathematical quickly.

Link to comment
3 hours ago, Focus Fidelity said:

The software focuses on attenuating resonant modes and improving time domain criteria (ringing, step response, excess phase). Boosting a sharp dip/notch will result in a filter with significant ringing (pre or post depending on if the dip is minimum or maximum phase) so the software will seek to largely ignore the dip.  The relationship between phase, time domain and magnitude response gets rather complex and mathematical quickly.

Yes, the one thing I do understand is that boosting nulls / dips is a bad idea and does not really work, in particular for lower frequencies, hence my question on Monday.

 

My later question was rather different, what I was trying to understand is what influence the phase adjustments (or possibly the other criteria you mention) made by the software might have on a low frequency null / dip, if any?

Windows 11 PC, Roon, HQPlayer, Focus Fidelity convolutions, iFi Zen Stream, Paul Hynes SR4, Mutec REF10, Mutec MC3+USB, Devialet 1000Pro, KEF Blade.  Plus Pro-Ject Signature 12 TT for playing my 'legacy' vinyl collection. Desktop system; RME ADI-2 DAC fs, Meze Empyrean headphones.

Link to comment
4 hours ago, Focus Fidelity said:

If the dip is minimum phase it will apply boost (gain) with matching phase shift to improve both the frequency response and impulse response.

OK, so the phase shift should have some positive impact on frequency response.  (I guess this will be room / case dependent to a degree)

 

Anyway, I feel that I have been laboring this point a little, so apologies for that, but I am just trying to get a better understanding of mechanisms involved here.  Ever curious.

 

For what it's worth, I have pretty much decided to buy the software and give it a try.  The only thing stopping me is that I am busy with other projects in the short term, no point buying software and not having time to use it, but I should get there in a couple of weeks or so.

Windows 11 PC, Roon, HQPlayer, Focus Fidelity convolutions, iFi Zen Stream, Paul Hynes SR4, Mutec REF10, Mutec MC3+USB, Devialet 1000Pro, KEF Blade.  Plus Pro-Ject Signature 12 TT for playing my 'legacy' vinyl collection. Desktop system; RME ADI-2 DAC fs, Meze Empyrean headphones.

Link to comment
15 hours ago, Confused said:

OK, so the phase shift should have some positive impact on frequency response.  (I guess this will be room / case dependent to a degree)

 

Anyway, I feel that I have been laboring this point a little, so apologies for that, but I am just trying to get a better understanding of mechanisms involved here.  Ever curious.

 

For what it's worth, I have pretty much decided to buy the software and give it a try.  The only thing stopping me is that I am busy with other projects in the short term, no point buying software and not having time to use it, but I should get there in a couple of weeks or so.

It's fairly complex, some of the relevant math https://en.wikipedia.org/wiki/Hilbert_transform

 

Time domain performance is the end goal really, an example showing a waterfall plot and step response can be found here https://www.focusfidelity.com/solutions

Link to comment
14 hours ago, Confused said:

I had a few plans for last weekend that were all basically abandoned due to very wet weather in the UK.  So finding myself with a little unexpected free time, I decided I would re-measure my room using REW, and with that done buy myself a copy of the Focus Fidelity Filter Designer software.  

 

I suspect that anyone who follows the topic of room correction on this forum will acknowledge that Mitchco is an expert in this field, by contrast, I am not. 

 

So when I read Mitcho's article regarding Focus Fidelity Filter Designer and he states that the software is "easy to use", I did kind of think in my head "yes, for you maybe, but I bet I'll get stuck somewhere".  The point being, that I can of many time in my life when I have tried some new software package and have found myself wasting hours on end wondering why it does not work as it should.  Maybe a bug somewhere, some odd setting that I have not got right, unexplained weird results, this kind of thing.  I have a fear of new software, which I see as route to frustration and hours of free time wasted.

 

So having sorted out some new REW measurements, I took the plunge and purchased the Filter Designer software.

 

Completely new to the software, I ran trough the instructions in the manual, step by step.  Where default settings were available, I simply kept with the defaults, I just wanted to see if I could get this thing working.  Anyway, within about one hour of starting the download for the software package itself, I had created a zip file for my convolutions, loaded them into Roon and had music playing.  I was actually expecting to hit a few issues and problems, but no, the software did it's stuff and everything seemed to work as it should. 

 

With a quick listen, I was not that happy with the result, I had just used the automatically generated target curve, and the tonal balance did not sound right at all to my ears.  To be honest, I was just pleased I had got everything working.  So, a second attempt, this time making some minor adjustments in the target frequency curve editor.  The results were better, this was moving in the right direction, but it still did not sound as I had hoped.  Then a third attempt, a bit more fine tuning of the target curve, this was much better.

 

After that, I just sat down and listened to selection of music.  Everything sounded good, from acoustic stuff, to rock, to electronica.  The main difference is a simple step forward in clarity, the effect is remarkable.  Sound that is a little muddled and irritating is replaced by music.  Maybe when people talk about "digititis" what they are actually suffering is "roonititis", but this is the effect, music sounds less reproduced by digits and more real.  Improvements in sound staging were obvious, the speakers become a less obvious source of the sound, moving towards invisibility, and a definite step up in clarity.  As a bit of an aside, I use a Devialet amp, which perhaps unusually these days has tone controls.  In the past, I have pretty much found the bass control to be useless in practical terms.  OK, it works, but if you happen to listen to an album that is inherently bass light, think of the classic "80's CD sound", then adding bass gain often just makes things sound worse, it just adds mud to the sound and sounds wrong.  Running the Focus Fidelity convolutions, suddenly the bass control does what it should, you add bass gain, you get more bass, clear bass.  I think this is clearly indicative of how Focus Fidelity is "cleaning" the sound of the system in the room

 

I was worried that would would struggle to get anywhere with this software, but it basically took my one Sunday morning from scratch to get a result that I was pleased with, and I am a room correction novice.  

 

I am sure that I still have a lot to learn, and I am also sure that I can improve on this early result.  The good thing is that the software is quick and easy to use, so I am happy that once I have had some time to listen more to the convolutions that I have, I will be able to investigate some of the settings in more detail, it should be easy to "fine tune", create new convolutions, and experience the results for myself.  This is where I think this software will be invaluable to myself.  Room correction is very complex topic, but there is nothing to beat being able to fine tune some parameter or another, and within a few minutes being able to listen to the results for yourself.

 

To be clear, the hardest part of this entire process was getting the actual measurements made in REW.  So I would say that for anyone who is at a level where they can perform the required REW measurements, using Focus Fidelity Filter Designer should be relatively simple, it is far more intuitive than REW.

 

It is very early days for myself, I have a lot to learn, a lot to investigate and try, but this is a great tool to start the convolution journey.  The fact that someone like myself can get good results in one rainy Sunday morning perhaps tells you all you need to know. 

 

Being in the UK the cost of the software worked out at £185.  Based on my experience of the software so far, this is £185 well spent and an absolute bargain in audiophile terms.

 

I also note that at the time of writing only 7 people are following this thread, whereas you get hundreds of followers following other "tweaker" threads.  Seems a shame, as there are some very real benefits to sound quality available here, for those will to make the effort.

Hi,

 

Thanks for taking the time to share your experience in detail.

Tweaking the target curve is to be expected in general, while the initial target is a best fit through the measurement data it can still also be a large change from what you might be used to or simply not be what you prefer.

If you do find yourself with any questions or have anything you'd just like doubled checked simply e-mail your project file to the Focus Fidelity support e-mail address.

 

Kind regards,

David

 

Link to comment
  • 1 month later...
  • 4 months later...

As a REW user, and considering Dirac is useless nowadays, I warmly welcome new DRC with time/phase corrections. I don't know how this software compare to the few real DRCs out there, like Acourate and Audiolense, but if nothing else it sure looks much more modern.

 

Good news anyway, I might give this a try!

Link to comment
On 3/15/2021 at 6:49 PM, Focus Fidelity said:

Hi @dm68 The filters are exported as stereo *.wav files so both channels in one file. There is no feature right now to export as mono *.wav files. I can however add this as a feature if needed.

I know of one very good reason for export to two mono convolution files: HQPlayer. I use that for its filters (I recommend it strongly), but also for its excellent DSP/convolution engine. However, it only accepts 2 mono files for convolution (as far as I know).

Link to comment
3 hours ago, MagnusH said:

As a REW user, and considering Dirac is useless nowadays, I warmly welcome new DRC with time/phase corrections. I don't know how this software compare to the few real DRCs out there, like Acourate and Audiolense, but if nothing else it sure looks much more modern.

 

Good news anyway, I might give this a try!

Hi @MagnusH, I can say that David's software is in the same class as Acourate and Audiolense wrt to time/phase corrections.

Link to comment
  • 1 month later...
On 11/5/2021 at 7:11 PM, MagnusH said:

As a REW user, and considering Dirac is useless nowadays, I warmly welcome new DRC with time/phase corrections. I don't know how this software compare to the few real DRCs out there, like Acourate and Audiolense, but if nothing else it sure looks much more modern.

 

Good news anyway, I might give this a try!

Hi @MagnusH, would you care to elaborate why you claim that "Dirac is useless nowadays"? I am considering Dirac Live alternatives and Focus Fidelity sounds like a great candidate for that. 

 

@mitchco, your Dirac review back in 2020 was very positive: https://audiophilestyle.com/ca/reviews/dirac-live-2-digital-room-correction-software-walkthrough-r884/. The obvious question is: any benefit for the final outcome of one system vs the other (DL vs FF)?

 

Thank you!

 

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...