• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Which is the best DSP option: DIRAC vs Acourate vs Audiolense vs RePhase vs ?

Snarfie

Addicted to Fun and Learning
Forum Donor
Joined
Apr 30, 2018
Messages
881
Likes
685
Location
Netherlands
Perhaps this post in my comparison thread on DRCs might be interesting; foobar2000 ABX comparator results are in the second half of the post.
Links to posts with various other DRC tests I did can be found at the end of the first post of the thread.
I would suggest to read dominikz extensive measurments it gave me for the first time a better insight in DRC. @dominikz after comparing all DRC solutions you ended up using REW as you prefereble DRC solution.?
 
Last edited:

Dathzo

Member
Joined
Jan 20, 2021
Messages
85
Likes
59
Perhaps this post in my comparison thread on DRCs might be interesting; foobar2000 ABX comparator results are in the second half of the post.
Links to posts with various other DRC tests I did can be found at the end of the first post of the thread.
Wow, this is exactly what I was looking for, thanks a lot for sharing!
That's excellent work you have done @dominikz . The community certainly benefits from such detailed testing.
Your conclusion of multi- vs single point measurements is different as the one from @mitchco, as he manages to get a consistent response across a couch area.

So you conclude 1) that all room corrections do the job well; 2) multipoint correction is better than single point, from the perspective of a better frequency response in average and 3) you managed to identify differences between Audiolense and Dirac and between Dirac and MathAudio.

What it is not clear from the post, is your preference (probably subjective). How would you rank the 3 systems from a sound quality achieved?
Thanks!
 

dlaloum

Addicted to Fun and Learning
Joined
Oct 4, 2021
Messages
874
Likes
572
Wow, this is exactly what I was looking for, thanks a lot for sharing!
That's excellent work you have done @dominikz . The community certainly benefits from such detailed testing.
Your conclusion of multi- vs single point measurements is different as the one from @mitchco, as he manages to get a consistent response across a couch area.

So you conclude 1) that all room corrections do the job well; 2) multipoint correction is better than single point, from the perspective of a better frequency response in average and 3) you managed to identify differences between Audiolense and Dirac and between Dirac and MathAudio.

What it is not clear from the post, is your preference (probably subjective). How would you rank the 3 systems from a sound quality achieved?
Thanks!
You missed this... further down the thread:

"Personally I thought Dirac was best overall, followed by ARC3."
 

Snarfie

Addicted to Fun and Learning
Forum Donor
Joined
Apr 30, 2018
Messages
881
Likes
685
Location
Netherlands
You missed this... further down the thread:

"Personally I thought Dirac was best overall, followed by ARC3."
Yes but the overal differences where minimal (regarding the sound results) i thougt. I'm not sure but he even ended up with REW (not Dirac) because of the flexibility.

Dominiukz conclusion.

Closing thoughts
Have to say I feel I've learned some more with each iteration of these comparisons, and this one was no exception :)
In general, my key takeaway is that with appropriate configuration and equivalent target curves, one can get what I feel are pretty similar audible results in most setups with either Audiolense, Dirac or MathAudio.

Most of the audible differences I heard in the end between Audiolense vs Dirac and MathAudio were IMHO related to differences intrinsic to multi-point vs single-point correction approach, rather than having to do with major differences in each SW implementation. Sadly I could not test Audiolense multi-point correction to verify :confused:, but based on all the other results I posted so far, I feel it likely my conclusion wouldn't change much.

In my experience through all these tests, multi-point corrections gave me a more consistent result when listening in the far-field (though it is a compromise at every listening point), whereas with single-point measurements I could sometimes hear some resonances - depending on where exactly I sat or which track I listened to. Maybe this could be further optimized somewhat in Audiolense even with a single-point measurement, but I doubt that in most acoustic environments one can get ruler-flat frequency response at a single measurement point without sacrificing the response somewhat at other points around it.


I'd also like to repeat that all of the audible differences I did hear were spectrum/tonality-related, and that I can't say that I heard any conclusive differences related to phase/time-domain correction done by Audiolense or Dirac in any of my tests - in any case none that I could then anchor myself to and recognize in blind listening. Perhaps others could do better with this - my ears may be far from golden :)
That being said, the bulk of the very real differences between Audiolense, Dirac, MathAudio and others for me is mainly in their ease of use, measurement/configuration/correction process robustness, configurability/customizability and price. Still, I'd say these are anyway very important considerations in themselves.
 
Last edited:

dominikz

Senior Member
Forum Donor
Joined
Oct 10, 2020
Messages
490
Likes
1,705
Dominikz after comparing all DRC solutions you ended up using REW as you prefereble DRC solution.?
Yes, I ended up using REW. I found that, when using MMM measurement as a baseline, it works just as well as the best automated DRCs - plus I like having full control of the filters. Also, I like using simple IIR filters as they don't introduce playback latency.
I recently posted a short explanation of the basic approach I use to measure and create filters here. At some point (and if I find the time) I may put together a more complete guide.

Wow, this is exactly what I was looking for, thanks a lot for sharing!
That's excellent work you have done @dominikz . The community certainly benefits from such detailed testing.
Thanks - much appreciated, and I'm happy you find it useful! :)

Your conclusion of multi- vs single point measurements is different as the one from @mitchco, as he manages to get a consistent response across a couch area.
To be honest multi- and single-point measurements can both sound great in practice.
In the end it is just a matter of what you're optimizing for - perfect response in a single position, or a better compromise across multiple positions (but not ideal in any single position). Of the two, I personally prefer the second approach. :)

So you conclude 1) that all room corrections do the job well; 2) multipoint correction is better than single point, from the perspective of a better frequency response in average and 3) you managed to identify differences between Audiolense and Dirac and between Dirac and MathAudio.
I find that once you solve the audibly offending resonances, the remaining small audible differences between most good DRCs are just different flavors of an already good sound. :) While I could identify differences in ABX, I wouldn't say any tested option sounded significantly better than others - just slightly different.
There are of course significant differences in DRCs with regard to ease of use, ergonomics, configurability, price, etc... which are all IMHO important as well.

What it is not clear from the post, is your preference (probably subjective). How would you rank the 3
You missed this... further down the thread:

"Personally I thought Dirac was best overall, followed by ARC3."
Please note that later in the thread I also tested other DRCs like Audiolense, MathAudio and REW (links to relevant posts are at the end of first post).
After all the tests I did I like the manual approach with MMM and REW the most. It perhaps takes a bit of time to learn and understand, but IMO works really well and gives you the choice how to implement the filters (DSP boxes like miniDSP, SW like EAPO, FIR...).
Of the automated solutions I still think Dirac was the best one - very good compromise of user-friendliness, configurability and great sound without having to spend too much time first to understand the ins-and-outs of room EQ.
 

Dathzo

Member
Joined
Jan 20, 2021
Messages
85
Likes
59
Yes, I ended up using REW. I found that, when using MMM measurement as a baseline, it works just as well as the best automated DRCs - plus I like having full control of the filters. Also, I like using simple IIR filters as they don't introduce playback latency.
I recently posted a short explanation of the basic approach I use to measure and create filters here. At some point (and if I find the time) I may put together a more complete guide.


Thanks - much appreciated, and I'm happy you find it useful! :)


To be honest multi- and single-point measurements can both sound great in practice.
In the end it is just a matter of what you're optimizing for - perfect response in a single position, or a better compromise across multiple positions (but not ideal in any single position). Of the two, I personally prefer the second approach. :)


I find that once you solve the audibly offending resonances, the remaining small audible differences between most good DRCs are just different flavors of an already good sound. :) While I could identify differences in ABX, I wouldn't say any tested option sounded significantly better than others - just slightly different.
There are of course significant differences in DRCs with regard to ease of use, ergonomics, configurability, price, etc... which are all IMHO important as well.



Please note that later in the thread I also tested other DRCs like Audiolense, MathAudio and REW (links to relevant posts are at the end of first post).
After all the tests I did I like the manual approach with MMM and REW the most. It perhaps takes a bit of time to learn and understand, but IMO works really well and gives you the choice how to implement the filters (DSP boxes like miniDSP, SW like EAPO, FIR...).
Of the automated solutions I still think Dirac was the best one - very good compromise of user-friendliness, configurability and great sound without having to spend too much time first to understand the ins-and-outs of room EQ.
Thanks a lot, all very clear! I had not looked further down the thread but now I have and read the great perspective that @mitchco brings as well.

This is probably a trivial question for many, but how can you see the filters applied from a certain correction with REW? I mean, from a measurement, I can read the magnitude, impulse, step responses, etc., but not the filters applied to your signal (like in the plot below). I am sure it is a simple way, but I have not cracked it yet (Acourate uses functions, i.e., subtract two curves, where this would be possible).
Any help is appreciated,
Thanks!
index.php
 

dominikz

Senior Member
Forum Donor
Joined
Oct 10, 2020
Messages
490
Likes
1,705
This is probably a trivial question for many, but how can you see the filters applied from a certain correction with REW? I mean, from a measurement, I can read the magnitude, impulse, step responses, etc., but not the filters applied to your signal (like in the plot below). I am sure it is a simple way, but I have not cracked it yet (Acourate uses functions, i.e., subtract two curves, where this would be possible).
Any help is appreciated,
Thanks!
index.php
There are a few ways to do it, but probably easiest is to do a direct loopback measurement with the DRC in the signal path.
E.g. apply the DRC filter to your DAC output, and then directly connect the DAC output to your ADC input and do a sweep measurement in REW.

With FIR-based DRCs that export filters as impulse responses you can also directly import the filter impulse response in REW by using the File > Import Impulse Response (from REW manual).
 

Dathzo

Member
Joined
Jan 20, 2021
Messages
85
Likes
59
There are a few ways to do it, but probably easiest is to do a direct loopback measurement with the DRC in the signal path.
E.g. apply the DRC filter to your DAC output, and then directly connect the DAC output to your ADC input and do a sweep measurement in REW.

With FIR-based DRCs that export filters as impulse responses you can also directly import the filter impulse response in REW by using the File > Import Impulse Response (from REW manual).
I would like to get the Dirac filters, which unfortunately I dont have a file for it. So I can have Dirac engaged in my streamer and do a sweep that will be measured by REW. I will get the corrected magnitude response. But how to get the filter from it? Do I need to process the magnitude response somehow? Again, sorry if it is a very trivial question, but have not gotten my head around that...
 

dominikz

Senior Member
Forum Donor
Joined
Oct 10, 2020
Messages
490
Likes
1,705
I would like to get the Dirac filters, which unfortunately I dont have a file for it. So I can have Dirac engaged in my streamer and do a sweep that will be measured by REW. I will get the corrected magnitude response. But how to get the filter from it? Do I need to process the magnitude response somehow? Again, sorry if it is a very trivial question, but have not gotten my head around that...
Dirac uses a proprietary format for the filters, yes, so you can't import them into REW directly.
Not sure if I fully understand your dilemma on loopback measurement, but if you're concerned that your ADC/DAC are introducing FR variations, you can first do a 'soundcard calibration' run in REW with Dirac disabled (see relevant REW manual section here). Though in general these days electronics have pretty flat FR so there shouldn't be much difference with and without calibration.

Please note that when I say 'loopback measurement' what I mean is that you connect the output of your DAC/streamer directly to your ADC. I.e. there are no speakers or microphones involved in this measurement - you're directly connecting the output to input with e.g. a cable.

Hope this helps!
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
2,085
Likes
1,479
Location
Canada
I would like to get the Dirac filters, which unfortunately I dont have a file for it. So I can have Dirac engaged in my streamer and do a sweep that will be measured by REW. I will get the corrected magnitude response. But how to get the filter from it? Do I need to process the magnitude response somehow? Again, sorry if it is a very trivial question, but have not gotten my head around that...

It’s not just the magnitude but the time information that needs to be deconvolved as well.

Haven’t tried something like that… but I think it would be easier to just do a simple A/B trace arithmetic function and recreate your own simplified magnitude-only version of Dirac’s filter correction via the REW EQ module by flattening the resulting SPL trace difference.
 

dominikz

Senior Member
Forum Donor
Joined
Oct 10, 2020
Messages
490
Likes
1,705
It’s not just the magnitude but the time information that needs to be deconvolved as well.

Haven’t tried something like that… but I think it would be easier to just do a simple A/B trace arithmetic function and recreate your own simplified magnitude-only version of Dirac’s filter correction via the REW EQ module by flattening the resulting SPL trace difference.
The loopback test will actually capture both magnitude and phase response of the filter. Here's an example for Dirac:
index.php

You can see the rising phase response - which is actually Dirac's crossover phase correction which results in time alignment between tweeter and woofer. Note however that this specific measurement is without soundcard calibration (this is why there is magnitude drop below 30Hz and above 10kHz). We also see some 'hairiness' in very high frequencies - but this was just a measurement artifact I had in this run, please disregard it.
 

fluid

Senior Member
Joined
Apr 19, 2021
Messages
436
Likes
764
When making these kind of measurements it is a good idea to have the output and input devices using the same clock source i.e. they are are part of the same device. If not there can be clock drift which can make the impulse response look odd. The frequency response of the filter should be well represented but evaluating the impulse response can get tricky.

Anyone reading the snippets from dominikz thread posted here should also read the whole thread to understand them in context.
 

fluid

Senior Member
Joined
Apr 19, 2021
Messages
436
Likes
764
Another way to get a version of a Dirac (or any VST Plugin) filter if it is a VST in a DAW or player host is to use a Dirac Pulse as the playback file/track and bounce or record the output to disk. Jriver has the Disk Writer function and most DAW software can do the same.
 

Dathzo

Member
Joined
Jan 20, 2021
Messages
85
Likes
59
Dirac uses a proprietary format for the filters, yes, so you can't import them into REW directly.
Not sure if I fully understand your dilemma on loopback measurement, but if you're concerned that your ADC/DAC are introducing FR variations, you can first do a 'soundcard calibration' run in REW with Dirac disabled (see relevant REW manual section here). Though in general these days electronics have pretty flat FR so there shouldn't be much difference with and without calibration.

Please note that when I say 'loopback measurement' what I mean is that you connect the output of your DAC/streamer directly to your ADC. I.e. there are no speakers or microphones involved in this measurement - you're directly connecting the output to input with e.g. a cable.

Hope this helps!


Ah, got it, that is it:
Please note that when I say 'loopback measurement' what I mean is that you connect the output of your DAC/streamer directly to your ADC. I.e. there are no speakers or microphones involved in this measurement - you're directly connecting the output to input with e.g. a cable.

I was not understanding the loopback principle. I have not applied it yet but will give it a go.
Thanks for your patience here
 
OP
J

JRS

Addicted to Fun and Learning
Joined
Sep 22, 2021
Messages
830
Likes
715
Location
Albuquerque, NM USA
Thoughtful and potentially $$ saving approach. And you are right: these particular drivers are fairly well matched and well behaved so they don't need a lot of "work." I do like getting yo a minimal phase situation, and I also like habing the option of very steep crossovers that allow extending the drivers bandwidth that typically requires good software, and not just a miniDSP or si,ilar bare bones approach. By a matrix I am imagining that to really dig out the IR of the drivers, phase, etc in a reverberant field requires a bit of math magic as in a set of simultaneous equations, but as I write this maybe its simply a system that needs correction in one blow, and all the valuable info can be obtained with a very truncated msmt of first arrival. On top of that it may not make a difference to get the native low frequencies of drivers as FR and room response can be lumped together and repaired in one swoop. I just assumed that Diracs insistence on mult msmts from different locations was in part to untangle the reflections with the other obvious part of enlarging the sweet spot. And you are probably right re Dirac and no xo's.
 
OP
J

JRS

Addicted to Fun and Learning
Joined
Sep 22, 2021
Messages
830
Likes
715
Location
Albuquerque, NM USA
I agree with your take on existing research and the majority of respected opinions....they usually diminish the importance of phase.
All that still doesn't smell right to me though...

Whenever i dig through the past studies quoted by those whose opinions we respect, i most often find the studies to really be lacking in terms of truly isolating phase audibility in real world terms.
It seems the studies have most often had to use headphones because speakers, powerful full range speakers that are necessarily multi-way to achieve realistic SPL, dynamics, and bass extension, have not had the ability to have flat phase until relatively recently.
IOW, we haven't had the ability to truly hear flat phase realistically, to even have a chance to know how rotated phase compares.
Most historical studies appear to me to have compared less-crooked to more-crooked.

A few xovers, not matter how low order, end up providing quite a bit of phase rotation. And i think there is fairly universal agreement that low order designs, in two-ways for instance, sound superior to high order designs. What is that but phase?
What is the perfect impulse and step response response everyone wants, other than both flat mag and phase?
What is impulse inversion, which is the heart of FIR, other than flattening both mag and phase?
And i guess finally, even simple minimum phase EQ's.....fix mag and you automatically fix phase.

Many, if not all, of the prosound speaker manufactures I've come to respect, stress phase. (perhaps just me looking for confirmation ??)
It's funny how the prosound world seems to be so far head of the home sound world, in terms of DSP and processing.

So anyway, there is some of my grounds for skepticism against phase doesn't matter...
and my gut says how can it not?
It's just time, how can time no matter how fine, not matter....until we can absolutely prove it doesn't?

Now, despite all the above, can i say for sure i know the flat phase i achieve in my speaker builds is the source of their great sound?
No, i cant. Like i replied earlier, it's not apples to apples exactly when i compare minimum phase processing to linear phase processing.
Minimum phase processing is much harder to dial in as smoothly as possible....simply because low order xovers are needed and drivers are called on to work wider overlapping bandwidths where rolloffs exist.
Linear phase processing is much easier to dial in smoothly.....because steep complementary xovers allow easy summation of drivers operating in their reduced bandwidth range, of better mag and phase response.

That may be the major reason my linear phase processing i use sounds better to...the simple likelihood of achieving superior frequency response.
I also have measured consistently better polars with the steep xovers.

I do wish i could figure out a better way to truly nail down phase audibility.

Gnarly wrote in his excellent post, "A few xovers, not matter how low order, end up providing quite a bit of phase rotation. And i think there is fairly universal agreement that low order designs, in two-ways for instance, sound superior to high order designs. What is that but phase?"

Not picking nits,but it is also a way to have a smoother and more gradual directivity change at crossover.

I may have mentioned this already but one experiment I did was to compare Jeff Bagby's Kairos with the first order filters vs an active version of same, using very steep filters with phase corrections via DEQX. I used the same xo points as the Satori drivers used are wide bandwidth and wanted to keep everything as constanf as possible. Couldn't AB them for obvious reasons as s lot of wiring changes needed, but had no preference much to my surprise. Turns out a well designed speaker using quality drivers doesn't need a lot of assistance. Room EQ helped both, but could have lived w/o the DRC and have been a happy guy indeed. Meanwhile Jeff seemed a bit surprised that the sound wasn't mangled.
 
OP
J

JRS

Addicted to Fun and Learning
Joined
Sep 22, 2021
Messages
830
Likes
715
Location
Albuquerque, NM USA
Perhaps this post in my comparison thread on DRCs might be interesting; foobar2000 ABX comparator results are in the second half of the post.
Links to posts with various other DRC tests I did can be found at the end of the first post of the thread.
Great read BTW.
 
OP
J

JRS

Addicted to Fun and Learning
Joined
Sep 22, 2021
Messages
830
Likes
715
Location
Albuquerque, NM USA
Truly the kind of citizen science that adds so much interest to the playback of music. And for having just gotten your feet wet,most impressive effort.
 

fluid

Senior Member
Joined
Apr 19, 2021
Messages
436
Likes
764
By a matrix I am imagining that to really dig out the IR of the drivers, phase, etc in a reverberant field requires a bit of math magic as in a set of simultaneous equations, but as I write this maybe its simply a system that needs correction in one blow, and all the valuable info can be obtained with a very truncated msmt of first arrival. On top of that it may not make a difference to get the native low frequencies of drivers as FR and room response can be lumped together and repaired in one swoop. I just assumed that Diracs insistence on mult msmts from different locations was in part to untangle the reflections with the other obvious part of enlarging the sweet spot. And you are probably right re Dirac and no xo's.
Nothing in Room Correction software is quite as complicated as that. Dirac is different in this regard as the multiple measurements are separate and information from each measurement can be processed separately, they do have some papers I have read before describing parts of it.

These are the steps performed by DRC_FIR
  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 pre-processed 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.
Frequency dependent windowing is the main function that assists with separating the speaker from the room. In REW the window is quite basic and is in cycles or octaves for the whole range. This still means that the window is short at the top and long at the bottom.
So you get progressively more room in the measurement as frequency goes down.

DRC_FIR goes further with a sliding window where the length of the window can be set different at both the low frequency and high frequency independently and there is also an exponent parameter to change how the two interact in the middle. Changing the window exponent is one factor that has a big impact on the way the correction sounds. Differences of 0.01 are audible as it strongly affects the midrange.

drc004.png

This really lets you tailor just how much speaker correction vs room correction is being done. I see very few speakers where the direct sound still couldn't be smoother but whether messing with this has a positive effect depends a lot on how even the directivity of the speaker being corrected is.
 
Top Bottom