• 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). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Dirac Live, FIR-filter - Yes, No?

Joined
Dec 27, 2018
Messages
18
Likes
11
#1
Hello again,
atm Im struggeling with the thought to buy a MiniDSP DDRC-24 or something that can handle FIR filters like this for example https://www.hifiberry.com/blog/announcing-the-dac-dsp/
ATM Im using Sure/Wondom DSP boards, UMIK1 (one for each channel) with AutoEQ(IIR), feeded by REW curves. The sound/difference is really great.

My questions are now:
1. What can I expect from Dirac Live in practice?
2. What is the benefit from Dirac against DIY EQ or DIY FIR filter?
3. Is there such a big difference between IIR filters and FIR filters?
4. Do I need a minimum equipment for one of the two?

Im interested in your experiences(ear), not some type of datasheet war of some devices or methods.

Thank you!
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
20,619
Likes
25,439
Location
Seattle Area
#2
In my opinion, what the equalization does is far more important than how it does it. So I would not worry about such things as IIR vs FIR, etc. From what I recall, DIRAC uses a hybrid approach anyway.

The main advantage of Dirac is that it is fully automatic. You position the mic in different locations as the wizard tells you and it generates the filters for you. You can then modify them by not using them above certain frequency for example and such but otherwise, there is no work on your part. You don't need to understand the science or how to implement it. All you need is a USB measurement mic which is pretty cheap these days.

Going the manual method requires a lot of understanding and you have the potential to completely blow up your speakers or amps! All it takes is one wrong metric and you could be blasting one frequency out at maximum level.

As a first step, buy a measurement mic, download REW and follow my tutorials here on how to measure your room. You will then know the magnitude of problem.

FYI, I know how to do EQ manually but I still use Dirac Live on my PC audio server.
 
Joined
Dec 27, 2018
Messages
18
Likes
11
#3
DIRAC uses a hybrid approach anyway.
Yep, they call it mixed-phase filters.

I own a UMIK1 and with REW and the Analog Devices Sigma Studio and the ADAU1701 Im quite familiar meanwhile. My room it a nightmare, but with hours of measurement it sound really good now. Can you please describe why you prefer to use Dirac Live if you can measure and correct yourself? Exactly that reason Im interested in.
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,067
Likes
3,391
Location
Riverview, FL
#4
Can you please describe why you prefer to use Dirac Live if you can measure and correct yourself?
I use an automated solution, because, well, its automated. MiniDSP OpenDRC-DI and AcourateDRC.

Its correction can be a mix of IIR and FIR.

Manual correction can become a bit tedious, depending on what you are trying to do, as a change here changes something else over there.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
20,619
Likes
25,439
Location
Seattle Area
#5
Yep, they call it mixed-phase filters.

I own a UMIK1 and with REW and the Analog Devices Sigma Studio and the ADAU1701 Im quite familiar meanwhile. My room it a nightmare, but with hours of measurement it sound really good now. Can you please describe why you prefer to use Dirac Live if you can measure and correct yourself? Exactly that reason Im interested in.
At the time Roon did not have a DSP function and I was not looking forward to getting yet another piece of software to make the correction in there. I tried Dirac, it worked very nicely and that was that. :) FYI I have it only correcting < 200 Hz.
 
Joined
Dec 31, 2018
Messages
16
Likes
17
#6
I use the APL1 in my car and bedroom, I have great results with it.

The measurement process isn't all that different from Dirac, but the software has a much steeper learning curve. Now that I know the software really well I can tune a car that's already been roughed in within ten minutes. I always take measurements after tuning and the results are usually spots on, at least within the limits of the equipment.

Here is the before and after graphs from my bedroom:


Green is the target curve, yellow was before the tune.

I plan to get three APL1012's for my theater setup.
 
Joined
Dec 27, 2018
Messages
18
Likes
11
#7
Thank you for the information and the graph. So I buy a DIRAC capable device and give it a try. Looks like its worth it.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
3,863
Likes
1,988
Location
Zg, Cro
#8
In my opinion, what the equalization does is far more important than how it does it. So I would not worry about such things as IIR vs FIR, etc. From what I recall, DIRAC uses a hybrid approach anyway.

The main advantage of Dirac is that it is fully automatic. You position the mic in different locations as the wizard tells you and it generates the filters for you. You can then modify them by not using them above certain frequency for example and such but otherwise, there is no work on your part. You don't need to understand the science or how to implement it. All you need is a USB measurement mic which is pretty cheap these days.

Going the manual method requires a lot of understanding and you have the potential to completely blow up your speakers or amps! All it takes is one wrong metric and you could be blasting one frequency out at maximum level.

As a first step, buy a measurement mic, download REW and follow my tutorials here on how to measure your room. You will then know the magnitude of problem.

FYI, I know how to do EQ manually but I still use Dirac Live on my PC audio server.
Does Dirac Live offer you a possibilty to measure the actual response after correction or you are left with the predicted response?
 

DonH56

Technical Expert
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
3,009
Likes
3,582
Location
Monument, CO
#9
It shows what it thinks the response should be like all the other programs I have used (Audyssey, MCACC, YPAO, etc.) To measure the actual response these days I use REW and a calibrated UMIK-1. In the past I used "professional" programs and an Earthworks M30. The results are essentially identical, so a $100 mic and REW replaced the several thousands of dollars of gear and SW I was using. Bittersweet progress is.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
3,863
Likes
1,988
Location
Zg, Cro
#10
It shows what it thinks the response should be like all the other programs I have used (Audyssey, MCACC, YPAO, etc.) To measure the actual response these days I use REW and a calibrated UMIK-1. In the past I used "professional" programs and an Earthworks M30. The results are essentially identical, so a $100 mic and REW replaced the several thousands of dollars of gear and SW I was using. Bittersweet progress is.
IMHO in that case measuring actual corrected response is a must as you cannot possibly be sure that predicted response matches the actual.
 

DonH56

Technical Expert
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
3,009
Likes
3,582
Location
Monument, CO
#11
I often often wondered why programs do not provide a post-analysis-and-correction capability. I suspect it is partly marketing and partly the difficulty in explaining to users why their real-world response does not look like the pretty flat curve shown and why it does not matter. The vast majority (IMO) of audiophiles do not know how to measure their response nor really want to. Nor do I, for that matter; I do it to dial in my subs and as a sanity check plus a bit of tweaking to preference. With my workload I'd rather spend my free time watching and listening than tweaking. Or maybe I've just outgrown the urge; moved from focusing on the means to an end to just wanting to get to the end.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
3,863
Likes
1,988
Location
Zg, Cro
#12
I choosed to do room correction manually. I guess I'm the "old school" guy. I did measurement with REW with cheap calibrated mic and make FIR filters using rePhase applied via BruteFIR convolver running on top of Volumio on a cheap fanless PC. Loudspeakers are Castle Harlech S2, here is how it looks:

Left channel:


right channel:


I don't have any doubts how the future will look - in a near future we'll have AI that will via DSP control some sophisticated active speakers which will do much better job than me correcting the speakers response. To make a car analogy, that same AI will probably drive a circle at Nurburgring at least 10 seconds faster than me with a car that will have 4 electric motors running very fast and silent. And that is the way it should be, that is the point of making things better in the future, the future our kids will hopefully enjoy living in.

But today.. today I'm enjoying Diana Krall singing "Temptation" from my Castle's as real as I can imagine her singing in my room. To continue with the car analogy, today I enjoy eery scream of Borg& Warner turbo in my Volvo when I downshift to 3rd gear at 140km/h approaching the curve at a Nurburgring circuit coupled with deep rumble of Ferrita stainless steel exhaust fighting to deliver 420HP to the driveshafts. And all of that sends shivers down my spine as car struggles to stay on the curve while tires are painfully screaming..

Sure, tomorrow some AI will make a better correction than I do. It will also drive better than I do, and that is the way it should be.

But today, today I'm driving.
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
20,619
Likes
25,439
Location
Seattle Area
#13
Does Dirac Live offer you a possibilty to measure the actual response after correction or you are left with the predicted response?
It can't because it has you measure multiple spots so naturally you can't repeat the same measurements afterward. Instead it shows you what it predicts will happen as do some other systems. I did measure mine with REW and the graph I got is very different than what it said was happening.

The only multi-measurement system I know that tells the truth here is Harman ARCOS. It uses multiple microphones in a fixed configuration so it can remeasure with consistency. Unfortunately the measurement system costs thousands of dollars so it is offered as a service by the dealer so not DIY.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
20,619
Likes
25,439
Location
Seattle Area
#14
FYI manual systems fall apart when we are talking multiple channels and multiple subs. Then the iterations can be measured in millions and no way can you hand optimize. The ARCOS system for example will attempt to optimize the crossover between multiple subs and multiple speakers to make sure there is smooth hand off from one to the other. This is just something that is not possible to do manually. See: https://www.audiosciencereview.com/...s/computer-optimization-of-room-acoustics.12/
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
3,863
Likes
1,988
Location
Zg, Cro
#15
It can't because it has you measure multiple spots so naturally you can't repeat the same measurements afterward.
I see no reason why they couldn't do it. They can simplly allow you to run the same measurement with the correction being active.

I did measure mine with REW and the graph I got is very different than what it said was happening.
IMHO that is not a good sign. I tried to take several measurements with the correction being active and I always get pretty much the same graph I posted above, and that is how I percieve the things should be.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
3,863
Likes
1,988
Location
Zg, Cro
#16
FYI manual systems fall apart when we are talking multiple channels and multiple subs.
I can't comment on that as I have tried it with stereo only without subs, but I can tell you that whenever I take measurements with correction being active I do get the same consistent results. Personally, I wouldn't be happy at all if it wouldn't be like that.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
20,619
Likes
25,439
Location
Seattle Area
#17
That is another difference: automatic EQ systems aim to improve fidelity for larger are than one spot. That necessitates multi-location measurements as otherwise you would be making things worse for another seat while improving another. That is why your case is more consistent: you are not trying to do this.
 

pos

Senior Member
Patreon Donor
Joined
Feb 13, 2018
Messages
332
Likes
301
#18
That is another difference: automatic EQ systems aim to improve fidelity for larger are than one spot.
Manual EQ does not imply single point measurement, and I know for a fact @Krunok did use both MMM and spacial averages methods when designing his correction.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
3,863
Likes
1,988
Location
Zg, Cro
#19
That is another difference: automatic EQ systems aim to improve fidelity for larger are than one spot. That necessitates multi-location measurements as otherwise you would be making things worse for another seat while improving another. That is why your case is more consistent: you are not trying to do this.
Not true. When I was taking MMM RTA measurements with REW I used the same area as covered by Dirac Live measuring spots.

@amirm , you're an engineer, I know you are aware that if Dirac Live correction cannot be verified by REW measurement it simply isn't good.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
3,863
Likes
1,988
Location
Zg, Cro
#20
Manual EQ does not imply single point measurement, and I know for a fact @Krunok did use both MMM and spacial averages methods when designing his correction.
Exactly. Thank you for chiming in on that.

Btw, I also verified the RTA MMM measurements with averaged sweeps taken from 9 points on the same points as Dirac Live.

This is MMM RTA vs average of sweeps taken at 9 points as suggested by Dirac Live (left channel):

 
Last edited:
Top Bottom