• 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 3.1.1 bugs?

Loron

Member
Forum Donor
Joined
May 17, 2021
Messages
41
Likes
32
Two hours ago I did the first measurement with the UMIK-1 I received yesterday.

Setup:

NAD M33 on newest BluOS 3.14.26
MacBook Pro with latest MacOS Big Sur 11.6
UMIK-1 ver.2 attached to the MacBook Pro via USB-C
Dirac Live 3.1.2

Result:

Speakers are out of phase :facepalm:

Any advice?
Hi Stefan,

Try to disable delay and gain compensation before Dirac export the filter to your AVR.
Dirac will redo the filter design process then you can export these to your AVR.

It sounds much better in my case.

Let us know if you see any improvement.
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
I’ve tested the included microphone and it wasn’t meaningfully different than my UMIK1 so go with the NAD mic if it’s better for you.
To be honest, I ordered the UMIK-1 with the expectation that at least from the measurement capability it is going to be a step upwards from the NAD supplied mic. If one can here a clear difference after applying the DL filter is another story. I wanted to give it a try. But now DL has to sort out this issue first. Otherwise I would have to downgrade to 3.0.14 again to do a measuement with the UMIK-1 which I can compare to the NAD mic.
4. The microphone should point up
You can do also horizontally, this is why you get two correction files for the UMIK-1 from miniDSP, 0° and 90°.
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
Try to disable delay and gain compensation before Dirac export the filter to your AVR.
Dirac will redo the filter design process then you can export these to your AVR.

It sounds much better in my case.
I'm not really sure why this should actually lead to a better result. But thanks for the idea, I will give this a try out of interest.

Right now I think the best way is to raise a support ticket for DL 3.1.2 so they can work on the issue.
 
OP
F

fcracer

Member
Forum Donor
Joined
Nov 10, 2020
Messages
93
Likes
129
To be honest, I ordered the UMIK-1 with the expectation that at least from the measurement capability it is going to be a step upwards from the NAD supplied mic. If one can here a clear difference after applying the DL filter is another story. I wanted to give it a try. But now DL has to sort out this issue first. Otherwise I would have to downgrade to 3.0.14 again to do a measuement with the UMIK-1 which I can compare to the NAD mic.

You can do also horizontally, this is why you get two correction files for the UMIK-1 from miniDSP, 0° and 90°.
It’s a good thing that NAD supplied a microphone of sufficient quality that there’s no practical difference for DL purposes.

You can indeed use a 0d calibration and point the microphone at the speakers, however my point is that this is not what Dirac now recommends; they recommend pointing it up to capture more of the room reflections. More of an issue when doing a full correction, but might as well do it the recommended way so that you have the option to apply full correction in the future.
 
OP
F

fcracer

Member
Forum Donor
Joined
Nov 10, 2020
Messages
93
Likes
129
I'm not really sure why this should actually lead to a better result. But thanks for the idea, I will give this a try out of interest.

Right now I think the best way is to raise a support ticket for DL 3.1.2 so they can work on the issue.
Turning off delay and gain should be used as a debugging tool. By turning it off, if things improve dramatically, then it provides another data point to use with Dirac support.
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
You can indeed use a 0d calibration and point the microphone at the speakers, however my point is that this is not what Dirac now recommends
Then (at least for 2.0ch systems) dirac and miniDSP recommend opposite things:

"We provide two calibration files to be used depending on your application.

- For stereo system (e.g. 2ch dirac live, single speaker measurement), use the 0deg file and point the UMIK-1 at the speakers
- For multichannel system (E.g. 5.1/7.1) or a surround application where multiple speakers are spreadout around the room, use the 90deg file and point the UMIK-1 at the ceiling."

I guess is it good enough when the UMIK-1 is pointed between the speakers during measurement; or should someone tilt the mic towards the left channel during the left sweep and then towards the right channel during the right sweep?
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
Turning off delay and gain should be used as a debugging tool. By turning it off, if things improve dramatically, then it provides another data point to use with Dirac support.
Ah, now I see what you mean, will give it a try.
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
So, I had a closer look again at the measurement with DL 3.1.2 I did yesterday that yielded the non satisfactory result. Some things are really wrong looking at this.
1. The measurement at position "Back Bottom Right" shows a delay between the left and right speaker that isn't there on all the other measurement points:
Bildschirmfoto 2021-09-30 um 11.04.32.png

2. The impulse response of the left speaker is shown like this (as it always has):
Bildschirmfoto 2021-09-30 um 11.04.46.png

For the right speaker it looks like this:
Bildschirmfoto 2021-09-30 um 11.04.57.png
 
OP
F

fcracer

Member
Forum Donor
Joined
Nov 10, 2020
Messages
93
Likes
129
Then (at least for 2.0ch systems) dirac and miniDSP recommend opposite things:

"We provide two calibration files to be used depending on your application.

- For stereo system (e.g. 2ch dirac live, single speaker measurement), use the 0deg file and point the UMIK-1 at the speakers
- For multichannel system (E.g. 5.1/7.1) or a surround application where multiple speakers are spreadout around the room, use the 90deg file and point the UMIK-1 at the ceiling."

I guess is it good enough when the UMIK-1 is pointed between the speakers during measurement; or should someone tilt the mic towards the left channel during the left sweep and then towards the right channel during the right sweep?
I’ve tried my best to help you. There are contradictory comments on the internet, likely from Dirac having evolved so quickly from Dirac 1, to 2 and now 3. Even Dirac Live 3 is evolving quickly. I’ve shared what has worked for me and what Dirac/NAD told me to do.

Please feel free to use the advice given or try something else that works for you.
 
OP
F

fcracer

Member
Forum Donor
Joined
Nov 10, 2020
Messages
93
Likes
129
So, I had a closer look again at the measurement with DL 3.1.2 I did yesterday that yielded the non satisfactory result. Some things are really wrong looking at this.
1. The measurement at position "Back Bottom Right" shows a delay between the left and right speaker that isn't there on all the other measurement points:

2. The impulse response of the left speaker is shown like this (as it always has):

For the right speaker it looks like this:
A couple of questions:
1. Are you sure you’re using 3.1.2? The colour scheme bug from 3.1.1 appears in your screen shot. That should have been fixed in 3.1.2.
2. If that one measure is unusual, it’s ok to re-measure it.
3. The overall signal to noise for your graph looks a little low. it appears you can either turn the volume up another 5-10db. I use -53db on the M33 which is the highest I feel comfortable in an apartment environment. I’d prefer to go higher while reducing the mic sensitivity.
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
A couple of questions:
1. Are you sure you’re using 3.1.2? The colour scheme bug from 3.1.1 appears in your screen shot. That should have been fixed in 3.1.2.
2. If that one measure is unusual, it’s ok to re-measure it.
3. The overall signal to noise for your graph looks a little low. it appears you can either turn the volume up another 5-10db. I use -53db on the M33 which is the highest I feel comfortable in an apartment environment. I’d prefer to go higher while reducing the mic sensitivity.
1. Yes, and the color scheme is still off:
Bildschirmfoto 2021-09-30 um 11.52.25.png

2. I did, please see below.
3. Yes, for the re-measurement I set main volume higher, so mic was set at 0dB. Thanks, I didn't notice first time around that mic was at +10dB. But shouldn't have had a big impact on the measurement.
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
So I re-measured the ""Back Bottom Right" position and guess what? It all went fine! Now the impulse response of the right speaker is calculated correctly also and exporting the filter yields a satisfactory result again. Too early to say if it is an improvement against the filter with DL 3.0.14 though. Need to give it more listening time.

But, and that is a big BUT: Why would this happen in the first place? The mic didn't move and everything was the same. So I did some re-measures of the other positions, too. And guess what? On another position the measurement of the right speaker was capped at 8kHz! Measured again it is fine. As I did more measurements more of those irregularities occurred. This basically says that DL 3.1.2 (at least on MacOS) has still serious issues that need to get sorted out with the measurements.
 

Dumdum

Active Member
Joined
Dec 13, 2019
Messages
128
Likes
85
Location
Nottinghamshire, UK
It’s a good thing that NAD supplied a microphone of sufficient quality that there’s no practical difference for DL purposes.

You can indeed use a 0d calibration and point the microphone at the speakers, however my point is that this is not what Dirac now recommends; they recommend pointing it up to capture more of the room reflections. More of an issue when doing a full correction, but might as well do it the recommended way so that you have the option to apply full correction in the future.
This is likely not why they recommend this aim, from my work in cars, putting the mic at 90 degrees to all the drivers with high freq gives a polar response that’s the same to all drivers, so although the high freqs will roll off a little for all, at least it’s consistent for all speakers and the low end doesn’t care

In a car I worked out a fair while ago that with left and right drivers being asymmetrical if I pointed the mic forwards that then put one driver 15 degrees off axis (drivers side) and one 60 degrees off axis and so the top end on the drivers side looked way better, so I aim it upwards and compensate for the inaccuracy with the high freq roll off when I tune to a house curve

It works very well, I have a jig that just mounts to the headrest metal uprights and puts the mic in the right place for timing and phase measurements, plus eq for freqs below 300-400hz, above that I use a moving mic, again this points upwards also
 
OP
F

fcracer

Member
Forum Donor
Joined
Nov 10, 2020
Messages
93
Likes
129
So I re-measured the ""Back Bottom Right" position and guess what? It all went fine! Now the impulse response of the right speaker is calculated correctly also and exporting the filter yields a satisfactory result again. Too early to say if it is an improvement against the filter with DL 3.0.14 though. Need to give it more listening time.

But, and that is a big BUT: Why would this happen in the first place? The mic didn't move and everything was the same. So I did some re-measures of the other positions, too. And guess what? On another position the measurement of the right speaker was capped at 8kHz! Measured again it is fine. As I did more measurements more of those irregularities occurred. This basically says that DL 3.1.2 (at least on MacOS) has still serious issues that need to get sorted out with the measurements.
Maybe your new UMIK is defective? I don’t know if they fixed this on UMIK1 v2 but the cable needs to be pushed in pretty hard on the mic end of v1 to firmly seat it.

Another possibility is that Dirac is using the laptop/computer microphone through user error or a bug. I don’t know why they even leave that option in the software.

Anyway, I’m glad to hear that you’ve now got a good set of measurements and outcome. The end result is worth the journey with Dirac :)
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
The end result is worth the journey with Dirac :)
This I totally agree. I have a dedicated listening room with good treatment of basotect absorbers and diffusers and a RT60 of 0.4s, but still Dirac optimizes (for the better) almost all acoustic properties to some degree regarding the higher frequencies, heavily the lower frequencies. To be honest, once used to it, I can not think of ever returning to a system, where frequencies below 200Hz are not running through a DSP.
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
1. Are you sure you’re using 3.1.2? The colour scheme bug from 3.1.1 appears in your screen shot. That should have been fixed in 3.1.2.
Maybe this has something to do with me using "dark mode" in MacOS, in this case they have forgotten about the modes.
 

Stefan

Member
Joined
May 29, 2021
Messages
50
Likes
5
I try to narrow down the bug I am experiencing; did another measurement today. Here you can see that both channels appear to have inverse impulse responses (measured vs. corrected):
Bildschirmfoto 2021-10-01 um 08.51.03.png

Bildschirmfoto 2021-10-01 um 08.51.16.png
 

Dolor

Member
Joined
Mar 13, 2021
Messages
7
Likes
2
I would like to ask if the Dirac Live 3 or 3,.. for Mac works properly now on OSX 12 or 12.1 Monterey?
I am afraid to update, but my wife needs this Mac Book too. It’s hers…. ;0)
 
Top Bottom