• 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 measurement beginner questions

JunkHippo

Member
Joined
Feb 8, 2021
Messages
17
Likes
20
Hi there,

I've been reading ASR for over a year now. Such a helpful page!
One of the things it inspired me to is trying out room correction. I have a pair of Neumann KH310 (great speakers!), in a relatively small room (15m²) and positioned not quite symmetrical - one is closer to a wall with floor-to-ceiling windows, and clearly there's potential in fixing at least bass in this room. I got a second hand minidsp SHD with a Umik 1 and after reading lots on the topic, did 3 measurement sessions so far.

The result: it sounds OK but it's clearly not the improvement I was hoping for. Specifically, even when applying some +6db or +8db lift in the bass region (not above the measured response), bass sounds kind of cleaner but still less defined, rather more diffuse than when disabling Dirac. Authority and presence are lacking, the pocket radio comparison comes to mind. It's not just the "less bass" phenomenon to get used to, I think. Something's clearly wrong there and I fear it's my measurements. On to my questions!

1. Volume Calibration
This part confuses me the most, because honestly Dirac's instructions and UI are not super clear on this, and other sources offer conflicting information.
So I'm supposed to get a reading on the speaker levels of 20-30db above noise floor I guess.
Dirac only says to set volume to a comfortable level and adjust mic gain until I get in the appropriate area. To get to a 25db reading, I have to set volume to close the maximum (where the lock is), a bit below 30db, which is definitely higher than my normal/ comfortable listening level. Then I still have to increase mic gain to the maximum red of +20db in order to get the desired response.
This can't be right? Some sources mention to use your usual listening volume, and to always leave mic gain at default 0db. But then I don't get any reading in the volume calibration UI.
I did make sure that the UMIK is selected as microphone, incl. correct calibration file. Also I got no errors during the measurement, though I noticed that frequencies with large dips fell off the chart during the measurement - but no error message.
2. Microphone Positioning
My listening position is 46cm from the back wall, and I used the 'focused' setting to cover the small sofa. I use a camera tripod and kept the sofa in place, but removed back cushions to get more clearance for the mic.
I also read that you shouldn't measure within 45cm of a boundary, so my "rear" positions are in line with listening position, and the rest moved forward accordingly. Positions were 30cm apart, which is the suggested minimum, but it's a small sofa and short listening distance of about 2.3m, so larger wouldn't make much sense.
The bottom rear positions are then fairly close the sofa's seating cushion.
I wonder if I should just leave out the 4 rear measurement points.
I also wonder if the listening position close to the wall (which can't be helped in this room) results in an overcorrection of bass frequencies perhaps? But then again the first measurement should be in my prime listening position.
3. Target Curves
I will post an example later. It's looking not too unusual I'd say. Some larger dips in there, which I tried to not correct too much (not over 5db boost), and then tried out standard curves, steeper linear curves, Harman curves... but no real winner so far.

Sorry for the long text, maybe some of the experienced Dirac users can spot some mistakes, or offer advice?
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Hi there,

I've been reading ASR for over a year now. Such a helpful page!
One of the things it inspired me to is trying out room correction. I have a pair of Neumann KH310 (great speakers!), in a relatively small room (15m²) and positioned not quite symmetrical - one is closer to a wall with floor-to-ceiling windows, and clearly there's potential in fixing at least bass in this room. I got a second hand minidsp SHD with a Umik 1 and after reading lots on the topic, did 3 measurement sessions so far.

The result: it sounds OK but it's clearly not the improvement I was hoping for. Specifically, even when applying some +6db or +8db lift in the bass region (not above the measured response), bass sounds kind of cleaner but still less defined, rather more diffuse than when disabling Dirac. Authority and presence are lacking, the pocket radio comparison comes to mind. It's not just the "less bass" phenomenon to get used to, I think. Something's clearly wrong there and I fear it's my measurements. On to my questions!

1. Volume Calibration
This part confuses me the most, because honestly Dirac's instructions and UI are not super clear on this, and other sources offer conflicting information.
So I'm supposed to get a reading on the speaker levels of 20-30db above noise floor I guess.
Dirac only says to set volume to a comfortable level and adjust mic gain until I get in the appropriate area. To get to a 25db reading, I have to set volume to close the maximum (where the lock is), a bit below 30db, which is definitely higher than my normal/ comfortable listening level. Then I still have to increase mic gain to the maximum red of +20db in order to get the desired response.
This can't be right? Some sources mention to use your usual listening volume, and to always leave mic gain at default 0db. But then I don't get any reading in the volume calibration UI.
I did make sure that the UMIK is selected as microphone, incl. correct calibration file. Also I got no errors during the measurement, though I noticed that frequencies with large dips fell off the chart during the measurement - but no error message.

you're supposed to put the microphone volume to where the room's noise floor is roughly -50 dB, for me that's around 90% with the UMIK-1, you can see the value right at this step where i highlighted.

As for the master volume, put it where the master volume on the SHD is typically when you listen.

1639748022158.png


2. Microphone Positioning
My listening position is 46cm from the back wall, and I used the 'focused' setting to cover the small sofa. I use a camera tripod and kept the sofa in place, but removed back cushions to get more clearance for the mic.
I also read that you shouldn't measure within 45cm of a boundary, so my "rear" positions are in line with listening position, and the rest moved forward accordingly. Positions were 30cm apart, which is the suggested minimum, but it's a small sofa and short listening distance of about 2.3m, so larger wouldn't make much sense.
The bottom rear positions are then fairly close the sofa's seating cushion.
I wonder if I should just leave out the 4 rear measurement points.
I also wonder if the listening position close to the wall (which can't be helped in this room) results in an overcorrection of bass frequencies perhaps? But then again the first measurement should be in my prime listening position.

Only take measurements where your head will actually be when you listen, the guidelines are just guidelines, I usually just take 4 to 5 measurement points.

the more measurements you take the less effective the EQ will be where you actually have your head, the average response will be skewed by the irrelevant measurements.

Don't worry about the boundary thing, unless you can actually do something about it, you always want to avoid boundaries in general, this is not specific to Dirac or room EQ, it's an acoustic thing.

3. Target Curves
I will post an example later. It's looking not too unusual I'd say. Some larger dips in there, which I tried to not correct too much (not over 5db boost), and then tried out standard curves, steeper linear curves, Harman curves... but no real winner so far.

Sorry for the long text, maybe some of the experienced Dirac users can spot some mistakes, or offer advice?

I made my own target curve based on the Olive / Harman preference score, with a 4dB boost in sub-bass. My speaker follows this response naturally till ~300Hz or so. As a result i limit the correction / curtains to 300Hz. Save it in a .targetcurve text file under C:\Users\*your username*\Dirac\targets

NAME
Unnamed
DEVICENAME

BREAKPOINTS
40 8.6
80 7.6
100.7 3.6
115.8 3.3
133.0 3
152.9 2.7
175.7 2.4
201.9 2.1
232.1 1.9
266.7 1.6
306.5 1.4
352.3 1.2
404.8 1
465.3 0.9
534.7 0.7
614.5 0.6
706.2 0.4
811.6 0.3
932.8 0.2
1072.0 0
1232.0 -0.1
1415.8 -0.2
1627.1 -0.3
1870.0 -0.4
2149.1 -0.4
2469.9 -0.5
2838.5 -0.6
3262.1 -0.7
3749.0 -0.8
4308.6 -0.8
4951.6 -0.9
5690.7 -1
6540.0 -1
7516.1 -1.1
8637.9 -1.1
9927.1 -1.2
11408.7 -1.2
13111.5 -1.3
15068.4 -1.4
17317.4 -1.4
18801.1 -1.4
LOWLIMITHZ
19.5
HIGHLIMITHZ
21797.2
 
Last edited:
OP
JunkHippo

JunkHippo

Member
Joined
Feb 8, 2021
Messages
17
Likes
20
Wow thanks! I'll try measuring with these hints next.
1. Noise floor was actually at -50db when I set the mic gain to the maximum +20db, but I was unsure if this maximum setting might have any disadvantages.
2. I will try with less points at more "realistic" positions (not at seating cushion hieght). Also makes the process quicker so win-win. I think Dirac state somewhere that measuring at more points is generally good, but probably it just depends on the actual room, how many points make sense.
3. I'll try that curve as well :) so far I also tend towards correcting only bass region, as I quite like the sound I already get, apart from room modes.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Wow thanks! I'll try measuring with these hints next.
1. Noise floor was actually at -50db when I set the mic gain to the maximum +20db, but I was unsure if this maximum setting might have any disadvantages.
That's a pretty quiet room! lucky for you!
 

phoenixdogfan

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
3,297
Likes
5,079
Location
Nashville
Wow thanks! I'll try measuring with these hints next.
1. Noise floor was actually at -50db when I set the mic gain to the maximum +20db, but I was unsure if this maximum setting might have any disadvantages.
2. I will try with less points at more "realistic" positions (not at seating cushion hieght). Also makes the process quicker so win-win. I think Dirac state somewhere that measuring at more points is generally good, but probably it just depends on the actual room, how many points make sense.
3. I'll try that curve as well :) so far I also tend towards correcting only bass region, as I quite like the sound I already get, apart from room modes.
I'm not sure I agree about the fewer points. I run mine with both the sofa setting and the chair setting, and even though I do all my listening in a chair midway between the two speakers in an equilateral triangle arrangement, I still prefer the greater number of points in the sofa correction. My theory is this corrects more for the entire room than the limited number of points in the chair setting..

It really doesn't take that much more time either. If you're using a mic boom, you just position for one side and rotate the boom to the other side while keeping it in the same place on the floor. Means you are taking 8 pairs of measurements plus the center. It can be done in 15 minutes once initial setup of mic levels, selection of measurement points, etc are done.

Of course, you could just run two sets of measurements (chair and sofa) and decide which you like best. I did that and it wasn't really all that much more work.
 
Last edited:

Pattern

Member
Joined
Dec 6, 2021
Messages
38
Likes
53
I'm actually finding in my own setup that for listening positions up against the back wall such as my sofa, a smaller gap between measurement positions and shifting the center position measurement a couple of feet forward into the room produces a more enjoyable result. Bass is still even and appropriate in level while sitting on the sofa (although slightly boosted; this is okay because I am running the default target curve to 500hz making the slight boost welcome) I like to use my presets to create the same target curve at different listening positions. Sometimes it's nice to wheel in an office chair, change the preset, and enjoy in near-mid field.
 

Beershaun

Major Contributor
Forum Donor
Joined
Oct 3, 2019
Messages
1,864
Likes
1,910
Thanks all for this thread. It's exactly what I'm working on right now as well and was trying to get my head around the best volume calibration level to go with.

I am using nthe full range version with my nad t778.

Here is what I'm debating and welcome thoughts and input.

1) current option: following the guidelines puts me at a volume level setting of -35db. This is much quieter than I generally listen at. So I bumped it up to my music listening volume setting of -30db. Note: I listen in Dolby surround mode so I don't need to worry about amp power and volume setting difference for 2ch.

2) option 2: do I increase the volume to target -86db in room measured volume level as Amir does with his tests in order to better highlight room modes and speaker flaws? Would this enable Dirac to do a better job of compensating?

3) option 3: do I set the volume level to -20 which is where I typically watch movies due to the lower input signal from my FireTV cube. This lower input requires me to turn up to hit the Same listening volume level.

Additional question re subwoofer levels. I would like to know what sub volume calibration people are targeting to make sure this line level output is high enough to make sure the sub amp gets enough voltage to hit full power. In my last run the sub ended up at -4db which significantly reduced the signal to the sub so it wasn't hitting nearly as hard as it should be.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Thanks all for this thread. It's exactly what I'm working on right now as well and was trying to get my head around the best volume calibration level to go with.

I am using nthe full range version with my nad t778.

Here is what I'm debating and welcome thoughts and input.

1) current option: following the guidelines puts me at a volume level setting of -35db. This is much quieter than I generally listen at. So I bumped it up to my music listening volume setting of -30db. Note: I listen in Dolby surround mode so I don't need to worry about amp power and volume setting difference for 2ch.

2) option 2: do I increase the volume to target -86db in room measured volume level as Amir does with his tests in order to better highlight room modes and speaker flaws? Would this enable Dirac to do a better job of compensating?

3) option 3: do I set the volume level to -20 which is where I typically watch movies due to the lower input signal from my FireTV cube. This lower input requires me to turn up to hit the Same listening volume level.

Additional question re subwoofer levels. I would like to know what sub volume calibration people are targeting to make sure this line level output is high enough to make sure the sub amp gets enough voltage to hit full power. In my last run the sub ended up at -4db which significantly reduced the signal to the sub so it wasn't hitting nearly as hard as it should be.

It doesn’t matter at which volume the speaker is set, as long as the output is above the microphone’s and the room’s noise floor the correction will be the same. Room modes are not a function of loudness, they’re a function of speaker directivity and room’s absorption coefficient.

Also Dirac will play a full amplitude signal, so use the level you would typically have it on when you’re playing music.

The subwoofer should be gain matched to the speakers, preferably, but that’s a tedious process. If your AVR’s setup wizard can SPL match them it’s fine. It shouldn’t be much of an issue if you have only one subwoofer.
 

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
2,116
Likes
2,782
Thanks all for this thread. It's exactly what I'm working on right now as well and was trying to get my head around the best volume calibration level to go with.

I am using nthe full range version with my nad t778.

Here is what I'm debating and welcome thoughts and input.

1) current option: following the guidelines puts me at a volume level setting of -35db. This is much quieter than I generally listen at. So I bumped it up to my music listening volume setting of -30db. Note: I listen in Dolby surround mode so I don't need to worry about amp power and volume setting difference for 2ch.

2) option 2: do I increase the volume to target -86db in room measured volume level as Amir does with his tests in order to better highlight room modes and speaker flaws? Would this enable Dirac to do a better job of compensating?

3) option 3: do I set the volume level to -20 which is where I typically watch movies due to the lower input signal from my FireTV cube. This lower input requires me to turn up to hit the Same listening volume level.

Additional question re subwoofer levels. I would like to know what sub volume calibration people are targeting to make sure this line level output is high enough to make sure the sub amp gets enough voltage to hit full power. In my last run the sub ended up at -4db which significantly reduced the signal to the sub so it wasn't hitting nearly as hard as it should be.
I find this to be quite good and straight forward to follow regarding speaker levels, sub levels etc
 

Beershaun

Major Contributor
Forum Donor
Joined
Oct 3, 2019
Messages
1,864
Likes
1,910
Thanks. Yes I found that site and used those instructions. I think it worked well for my speakers but didn't like the subwoofer results. Specifically it ended up setting my subwoofer line out level to -4db which made the level too low to really drive them. So curious why the instructions for the sub were different from the normal "set the sub volume at 12 o'clock position."

When I did a second calibration run with the sub at 12 o'clock and same level as the rest of the speakers it set the line level to -1db which worked much better nfor driving them.
 

Beershaun

Major Contributor
Forum Donor
Joined
Oct 3, 2019
Messages
1,864
Likes
1,910
Also Dirac will play a full amplitude signal, so use the level you would typically have it on when you’re playing music.
Thanks for your response. Can you clarify? The Dirac calibration signal is louder or softer depending on my volume calibration level. So I thought the room mode would likewise be louder or softer. Just not sure if there needs to be a minimum SPL to make sure the room gain is apparent.

The subwoofer should be gain matched to the speakers, preferably, but that’s a tedious process. If your AVR’s setup wizard can SPL match them it’s fine. It shouldn’t be much of an issue if you have only one subwoofer.
I have 2. Does this complicate things? The current calibration has the sub level at -1db when I level matched them during volume calibration. The previous one had it at -4db when I used the instructions linked above and this made the line level too low for the sub amps.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Thanks for your response. Can you clarify? The Dirac calibration signal is louder or softer depending on my volume calibration level. So I thought the room mode would likewise be louder or softer. Just not sure if there needs to be a minimum SPL to make sure the room gain is apparent.


I have 2. Does this complicate things? The current calibration has the sub level at -1db when I level matched them during volume calibration. The previous one had it at -4db when I used the instructions linked above and this made the line level too low for the sub amps.

 
OP
JunkHippo

JunkHippo

Member
Joined
Feb 8, 2021
Messages
17
Likes
20
Thanks again for the input here!
I made some new measurements. The most successful one was with first position where my head usually is, then covering all points with rather larger distances (45cm?) between them. I left out the 4 rear points that would be directly on the wall.
With this measurement and some tuning of the bass region I got to a point where I was satisfied with the bass response.

However I ended up selling my MiniDSP SHD.

Why?
Purely personal taste. Dirac is really powerful and the SHD is a fine unit. I liked the bass correction, but personally I could not get used to the time alignment, the way it affected stereo imaging. It nailed every instrument and voice to their own place, generally more "correct", but I never escaped the feeling that I'm sitting in an artificial sound bubble. I just couldn't get used to it, and relaxed when I went back to my usual setup that's less precise but feels more natural to me.
I do miss the bass correction somewhat, and would be intrigued to try other DSP options (Neumann, Trinnov).
Again, this is just my personal taste, not claiming that Dirac isn't doing a good job. I think it's pretty cool, would just wish for a bit more control for the user (e.g. limiting all processing to specific frequency range).
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Thanks again for the input here!
I made some new measurements. The most successful one was with first position where my head usually is, then covering all points with rather larger distances (45cm?) between them. I left out the 4 rear points that would be directly on the wall.
With this measurement and some tuning of the bass region I got to a point where I was satisfied with the bass response.

However I ended up selling my MiniDSP SHD.

Why?
Purely personal taste. Dirac is really powerful and the SHD is a fine unit. I liked the bass correction, but personally I could not get used to the time alignment, the way it affected stereo imaging. It nailed every instrument and voice to their own place, generally more "correct", but I never escaped the feeling that I'm sitting in an artificial sound bubble. I just couldn't get used to it, and relaxed when I went back to my usual setup that's less precise but feels more natural to me.
I do miss the bass correction somewhat, and would be intrigued to try other DSP options (Neumann, Trinnov).
Again, this is just my personal taste, not claiming that Dirac isn't doing a good job. I think it's pretty cool, would just wish for a bit more control for the user (e.g. limiting all processing to specific frequency range).
I understand completely what you're talking about. If you want only the bass adjustments then you can get the MiniDSP Flex (sans Dirac) for like 400$.
 

withoutsuit

Active Member
Joined
Dec 21, 2021
Messages
163
Likes
97
Again, this is just my personal taste, not claiming that Dirac isn't doing a good job. I think it's pretty cool, would just wish for a bit more control for the user (e.g. limiting all processing to specific frequency range).
Well, you already can do this, by placing the left and right "curtains" in the required frequency range. It´s made at the same step, as you choose your target curve.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Well, you already can do this, by placing the left and right "curtains" in the required frequency range. It´s made at the same step, as you choose your target curve.
This limits the frequency response magnitude correction, not the phase correction.
 
Top Bottom