• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Dirac and similar

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,406
Location
Sweden
Yes, multiple subs are good, I have only two. :( If space would admit it I would use the Ino Audio Infra-Y-12 (12x15 inch subs). I don't have the space though. Or absolute need... :)
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,406
Location
Sweden
A ooaurse room simulation would place some dips both above and below 90 Hz when I have the microphone at 90 cm height.

Skärmavbild 2019-09-08 kl. 14.29.44.png


The right speaker is having some other stuff around it (LP storage, pushing the rear wall to the right of the speaker forward, which changes the SBIR significantly). So there i100-160 Hz effect is mostly likely SBIR effects from corner/sidewall. The peak and dip below 100 Hz is not affected though.

left+right.png
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
A ooaurse room simulation would place some dips both above and below 90 Hz when I have the microphone at 90 cm height.

View attachment 32898

The right speaker is having some other stuff around it (LP storage, pushing the rear wall to the right of the speaker forward, which changes the SBIR significantly). So there i100-160 Hz effect is mostly likely SBIR effects from corner/sidewall. The peak and dip below 100 Hz is not affected though.

View attachment 32899

Hard to tell. But you can easilly tell once you try to correct such dips: whatever boost you throw at dip caused by SBIR it practically doesn't react to it. If it reacts nicely than it's not SBIR dip.

Anyway, from what I can tell by looking at those graphs your room is pretty much "standard". From 300Hz upwards speakers have nice linear response, below that is room dominated, especially bellow 100Hz, which is again case with most roos.

Any chance for with/without rear panels measurements? :)
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,406
Location
Sweden
I'll make a try later this evening.
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,406
Location
Sweden
Here is a test with no panels vs two panels behind on the wall behind the speakers. Blue=panels, red=no panels. (You can also see my intentional voicing with a little more boost 1-2 kHz compared to 2-5 kHz, which I have applied according to Shirley et al. But that is my preference to get a more neutral response for stereo listening.)

First the frequency response 100-20,000 Hz:
Panel test.png

1/6 octave 20-20,000 Hz
Panels test 1_6 octave.png

And the reverberation time:
Panel test 2.png
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Here is a test with no panels vs two panels behind on the wall behind the speakers. Blue=panels, red=no panels. (You can also see my intentional voicing with a little more boost 1-2 kHz compared to 2-5 kHz, which I have applied according to Shirley et al. But that is my preference to get a more neutral response for stereo listening.)

First the frequency response 100-20,000 Hz:
View attachment 32904
1/6 octave 20-20,000 Hz
View attachment 32906
And the reverberation time:
View attachment 32905

Very interesting! The panels seem to help with reverb time noticeably but with FR it's a kind of a mixed bag, don't you think? :)
 
Last edited:

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,406
Location
Sweden
Very interesting! The panels seem to help with reverb time noticeably but with FR it's a kind of a mixed bag, don't you think? :)

Yes, you can only say that it changes and that it is audible; measurements are made at one point at listening seat so you see all kinds of room effects. I did not make a close up measurement (1 meter), I was short of time. Perhaps next time. Ideally, to isolate the effects of the panel I should do the measurements against a wall outside.

And as you say, the benefit I hear with the panels are more details and better dynamics. More of silence when it should be silent.
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,406
Location
Sweden
And FYI, a measurement of direct response at 1,5 meter, in room, average between 0 and 15 deg horizontally. Heavily gated so very low resolution below 1 kHz.

0 and 15 deg average.png
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,406
Location
Sweden
Another analysis when I have shortened the time window, not start 7 ms, length 70 ms. As above, panel blue, no panel red.

Shorter time window.png
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,406
Location
Sweden
Yeah that's what I figure... but as the post was preceded by:

I found it a little confusing... but it is late and I am quite tired and maybe a little stupid (well a little more than normal I mean).

It may have been confusing, but in design meaning that the speaker (driver to wall) distance is around 35-40 cm, pushing the first SBIR cancelation from back wall higher up in frequency, enough to be dealt with using a thinner damping panel. Since there is a boost below that 1st cancellation frequency, you need to calculate with that in crossover and driver output.

giphy.gif

(This is not related to room resonant modes, or later comb filtering effects by sound waves bouncing around.)
 

digicidal

Major Contributor
Joined
Jul 6, 2019
Messages
1,981
Likes
4,838
Location
Sin City, NV
It may have been confusing, but in design meaning that the speaker (driver to wall) distance is around 35-40 cm, pushing the first SBIR cancelation from back wall higher up in frequency, enough to be dealt with using a thinner damping panel. Since there is a boost below that 1st cancellation frequency, you need to calculate with that in crossover and driver output.

giphy.gif

(This is not related to room resonant modes, or later comb filtering effects by sound waves bouncing around.)
Yes that's what I figured the case was. I just thought your "already fixed" and "design and setup" - meant that you were saying you had already handled cancellations of that nature and therefore everything shown was only what you would fix via DSP... which seemed in conflict with the rest of the post. Especially confusing considering trims are much more likely to be effective than boosts via EQ in most situations.
 
Last edited:

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Yes that's what I figured the case was. I just thought your "already fixed" and "design and setup" - meant that you were saying you had already handled cancellations of that nature and therefore everything shown was only what you would fix via DSP... which seemed in conflict with the rest of the post. Especially confusing considering trims not boosts are usually all that's effective via EQ.

Boosts are also effective, but not if you are trying to correct SBIR related dip.
 

digicidal

Major Contributor
Joined
Jul 6, 2019
Messages
1,981
Likes
4,838
Location
Sin City, NV
Boosts are also effective, but not if you are trying to correct SBIR related dip.
Correct, I edited my original statement to be more clear.

I suppose I could have also said "trims are less likely to exacerbate problems in other areas and listening positions than boosts are"...
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,406
Location
Sweden
Yes that's what I figured the case was. I just thought your "already fixed" and "design and setup" - meant that you were saying you had already handled cancellations of that nature and therefore everything shown was only what you would fix via DSP... which seemed in conflict with the rest of the post. Especially confusing considering trims are much more likely to be effective than boosts via EQ in most situations.

With respect to room modes, I would proabably just reduce the 47 Hz mode and then adjust level of the bass. As mentioned I did this previously using a Behringer parametric EQ and it was quite sufficient for my needs
 

Flak

Senior Member
Industry Insider
Joined
Nov 18, 2018
Messages
385
Likes
593

Snarfie

Major Contributor
Forum Donor
Joined
Apr 30, 2018
Messages
1,168
Likes
925
Location
Netherlands
I'm using Mathaudio room eq incombination with foobar2000 as add In under Win 10. It is free (incombination with foobar2000) work simpel with excellent results. https://mathaudio.com/room-eq.htm
 

Flak

Senior Member
Industry Insider
Joined
Nov 18, 2018
Messages
385
Likes
593
I'm using Mathaudio room eq incombination with foobar2000 as add In under Win 10. It is free (incombination with foobar2000) work simpel with excellent results. https://mathaudio.com/room-eq.htm

Being this a thread dedicated to "Dirac and others" it gives the opportunity to detail about different approaches.

MathAudio specifies that it works with zero latency, as a result (and correctly stated) it doesn't delay the audio track... but "no latency at all" also points to a major difference.

A speaker in a room can be measured and showed to have some frequency (magnitude) response.
What we try to do with Dirac live is to improve this response in frequency (to have the desired target) and in time (we want the impulse response to be as close to a perfect impulse as possible).

When considering only the frequency response we can (in theory) apply any filter that has the inverse frequency response and the resulting response will be flat.
This filter is often a minimum phase filter, and if the system (the speaker together with the room) was already minimum phase, the result will be a flat frequency response and the impulse response will be perfect.

If the system is not minimum phase (and this is the case in our normal listening rooms) the frequency response will still be flat, but the impulse response will not be ideal (exactly how it looks will depend on the phase).
In order to get the impulse response correct for a non minimum phase system you need to use a filter that is not minimum phase.

What this filter will do is shift certain frequencies in time in such a way that they all arrive at the same time, thereby achieving the desired result.
Obviously we cannot move them all to time zero, as this requires us to know about the future.... we have advanced technology but not that advanced :)

Instead we shift all of them to the latest frequency, that is we delay frequencies to match the latest one (within reasonable limits of course).
As a result the Dirac correction requires some latency which is irrelevant for audio only applications and normally low enough for audio&video.
 

hugodlc

Member
Joined
Jun 26, 2019
Messages
18
Likes
42
In my setup I first treated the room, (basement home theater) by opening the ceiling and adding rock wool to the hollow cavity between the joists, then added 2" and 4" panels around the room to tame reflections and absorbing some of the nasties.
Then measured using REW and applied the suggested correction to my BSS 160 unit. Then re-measured and re-adjusted, this was repeated 3 or 4 times per speaker never boosting or cutting more than a couple dB's per filter. (And then adjusted volume to the EQ bypass setting, and A-B re-checked by ear to make sure it was actually helping and not the other way around) I did EQ above the room transition frequency for a particular spot in the room (my mixing position) but I tested uniformity and in the end softened the filters so minimal EQ was being done, while achieving the most transparent results.
It is very time consuming, but in the end I achieved the curve I wanted, and the work I do in my room translates exceptionally well to other mixing stages.

But the approach is first treat the room, then get exceptionally flat speakers wit good amplification, then just as the cherry on the pie, apply subtle EQ.
You can't fix a bad room/speaker with eq.
 

Attachments

  • Center.jpg
    Center.jpg
    89.8 KB · Views: 122
Last edited:

Snarfie

Major Contributor
Forum Donor
Joined
Apr 30, 2018
Messages
1,168
Likes
925
Location
Netherlands
Being this a thread dedicated to "Dirac and others" it gives the opportunity to detail about different approaches.

MathAudio specifies that it works with zero latency, as a result (and correctly stated) it doesn't delay the audio track... but "no latency at all" also points to a major difference.

A speaker in a room can be measured and showed to have some frequency (magnitude) response.
What we try to do with Dirac live is to improve this response in frequency (to have the desired target) and in time (we want the impulse response to be as close to a perfect impulse as possible).

When considering only the frequency response we can (in theory) apply any filter that has the inverse frequency response and the resulting response will be flat.
This filter is often a minimum phase filter, and if the system (the speaker together with the room) was already minimum phase, the result will be a flat frequency response and the impulse response will be perfect.

If the system is not minimum phase (and this is the case in our normal listening rooms) the frequency response will still be flat, but the impulse response will not be ideal (exactly how it looks will depend on the phase).
In order to get the impulse response correct for a non minimum phase system you need to use a filter that is not minimum phase.

What this filter will do is shift certain frequencies in time in such a way that they all arrive at the same time, thereby achieving the desired result.
Obviously we cannot move them all to time zero, as this requires us to know about the future.... we have advanced technology but not that advanced :)

Instead we shift all of them to the latest frequency, that is we delay frequencies to match the latest one (within reasonable limits of course).
As a result the Dirac correction requires some latency which is irrelevant for audio only applications and normally low enough for audio&video.
Thanx for your explanation this is what I'm after (as long as i can comprehend it:facepalm:). All these solutions like REW, Dirac live. Mathaudio, Lyngdorf, Sonarworks, Minidsp an probably more it would be handy to have a sort of overview/test what are the benefits or drawbacks of all these solutions for instance Mathaudio avoids pre-echo which will have an affect on the neutrality of the sound whatever that means?. Does that mean that other solutions like Dirac live also avoids pre-echo and if not why, are ther certain disadvantages to do so?. Mine impression is that most of these solutions don't want to reveal how there solution (algo's) really works what is understandable considering competitive advantage. What for me as a consumer important is is that i can choose the best solution for my needs. This could be Mathaudio but another solution fit mabey better. For that i need tests/comparisons, overviews an reviews. IMO room correction is the biggest change atleast for me in the last 50 years in reproducing music (much much) better.
 
Last edited:

Igor Kirkwood

Active Member
Joined
Jan 28, 2020
Messages
109
Likes
238
Get good results for loudspeaker measurements in anechoic room or with a Klippel system is satisfying
for the manufacturer and for the customer.
But to optimize his system, the listener needs real measurements in his room.
Some sofwares such as Dirac or hardware have been discussed here.
I’d like to present another solution that is used in my system.

pano.jpg


As a sound engineer for classical music, I’m working generally with a minimal recording setup : 2
Neumann TLM50 mics, no compression, no DSP,...
So my listening setup is used to check my recordings :
-a room of 130 cubic meters, walls are treated with nearly 5 cubic meters of glass-wool
listening distance is 3.1m, distance between front speakers is 2.9m
-stereo front channels are Yamaha NS1000X in a 3 ways activ configuration, original tweeter was replaced
by a 27mm Focal Be, foam on frontwall to minimize diffraction
-2 surrounds with Yamaha NS1000X passives
-4 subwoofers SVS PC-2000 at mid walls
-QSC Q-Sys 110F is doing FIR filtering for subwoofers (FIR filter at 90Hz, 70dB/oct), crossovers for
NS1000X (375 and 1800Hz, 70dB/oct), FIR and IIR EQ correction, delays,,...

This processor also deals with source choice, stereo/multichannel, phantom center, LFE and bass-
management, ...

The subwoofer management is also special : you can choose between mono, stereo, M/S, Griesinger
(enhanced L-R),... and possibility to modify L+R (on phase) and L-R (out of phase) bass levels
For listening tests, all configurations can be switched and compared instantaneously on a computer with a
friendly user interface
The system setup has been done by Jean-Luc Ohl (who introduced MMM in www.ohl.to)
Correction is based on MMM for the amplitude (min phase only) and sweeps for the excess phase, and
loaded in Q-Sys with 8192 taps FIR plus IIR (programming and measurements analysis are done
remotely).
Let’s now see some measurements :

1/ no EQ on front NS1000X, just crossovers and delays

91.1.2-p1 (3).png



2/ front NS1000X with EQ

91..2.30-p1 (2).png



3/front NS1000X with subwoofers : this configuration is optimized to enhance L-R (stereo signals on low frequencies, see Griesinger)


92.15CL.9.20-p1 (1).png
 
Last edited:
Top Bottom