AVS Forum banner

3321 - 3340 of 4511 Posts

·
Premium Member
Joined
·
2,084 Posts
That is a terrible video. If a first-time user was looking for advice on how to set up Dirac, this is not the video for it. How he goes through the level-setting screen is peculiar, and not anything like I would do on the HTP-1. And his advice on adjusting the target curve to avoid anything more than a 6dB correction results in abrupt changes in the curve--not good. And he never even clicks on the "Corrected" button to see what the result of his tweaks actually look like. This video is on the same level of some of the Youthman videos, IOW pure garbage. I hope you didn't actually follow his advice.
Oh absolutely not. Funny you mentioned Youthman. He has reviewed many subs in his room. Recently he went to see JTR show room and from there got in touch with an AVS member who has JTR stuff. That member offered him the assistance with his room. When they took reading of his own subs, they needed major EQ. I think they used REW to EQ his subs in minidsp. Then he ended up posting how amazing the subs sound. But he said similar things he reviewed about other subs. If he didn't have his own subs properly dialed in, how did he review other subs then? I mean what if he was listening to major peaks.... What if there were dips because of his room and he ended up saying its not a good sub (which I don't know if he did). Goes to show the level of all these youtube reviewers.
 
  • Like
Reactions: AVTimme

·
Registered
Joined
·
248 Posts
I'd like to see if other DLBC users like @AustinJerry or anyone else can reproduce the same issue I've encountered with multichannel/upmixed content and DLBC. DLBC optimizes stereo pairs, but it does not optimize any other combination of main speakers, such as L+C. That means DLBC can result in reduced bass at some frequencies when playing content from some pairs of speakers, depending on the all-pass filters used in the calibration.

To test, set the REW driver to Java and sweeps set to L+R, set the processor to all mono mode, and turn off all speaker amps, leaving only sub amps on. This allows us to measure the bass-managed subwoofer response when the same signal is sent to all speakers simultaneously. In the processor, disable all main channels except L+R. Keep the same number of subs that was used for the DLBC calibration. For my system this was a 2.2.0 speaker layout.
  1. With 2.x.0 speaker layout, take one measurement with regular Dirac, and another with DLBC. These are the reference measurements.
  2. Enable one other speaker or pair, and repeat measurements with regular Dirac and with DLBC. With regular Dirac, adding more speakers simply increases the SPL of the measurement across the whole frequency range, as expected. But with DLBC this was not always the case, depending on the particular speaker layout.
  3. Take as many combinations of speaker layouts as needed until the issue is found.
In my case, enabling the center speaker resulted in less bass centered at around 60 Hz with DLBC. And the speaker layout that demonstrated the issue most clearly on my particular setup and calibration was 4.2.4, shown below.

3086758
 
  • Like
Reactions: philipbtz

·
** Man of Leisure **
Joined
·
24,609 Posts
So how does this "issue" affect the audio when playing normal content, say stereo music sources, or 5.1 movie sources? Are you saying there is audible loss of bass?

This is a rather involved test and I am not sure how soon I would be able work on it. Why are you using the Java drivers? You can output a L+R signal using HDMI drivers, which is what I typically would use.
 

·
Registered
Joined
·
248 Posts
So how does this "issue" affect the audio when playing normal content, say stereo music sources, or 5.1 movie sources? Are you saying there is audible loss of bass?

This is a rather involved test and I am not sure how soon I would be able work on it. Why are you using the Java drivers? You can output a L+R signal using HDMI drivers, which is what I typically would use.
Yes, the most audible example is upmixing stereo content with Auro 3D, which loses a considerable amount of bass with my DLBC calibration. I believe I've heard the same issue with movies but not to the same extent. I didn't do too many listening comparisons with movies so I wouldn't put as much faith in that.

Java drivers are probably not necessary as long as you make sure the receiver sees a stereo signal so that it does the upmix correctly.
 

·
** Man of Leisure **
Joined
·
24,609 Posts
Yes, the most audible example is upmixing stereo content with Auro 3D, which loses a considerable amount of bass with my DLBC calibration. I believe I've heard the same issue with movies but not to the same extent. I didn't do too many listening comparisons with movies so I wouldn't put as much faith in that.

Java drivers are probably not necessary as long as you make sure the receiver sees a stereo signal so that it does the upmix correctly.
When you play a mono signal through multiple speakers, you are going to get interference patterns that are not necessarily associated with issues with how DLBC is phase matching speaker pairs. I am not convinced that your test is revealing a real-world issue. While I typically use DSU and not Auro3D for up mixing for 2.0 music content, I have never perceived any issues with the bass response when using DLBC.

While I mean no disrespect, especially since you have made a such a significant contribution with your custom GUI, I must decline running the measurements you requested, since I don't believe they will reveal any meaningful results.
 

·
Registered
Joined
·
248 Posts
When you play a mono signal through multiple speakers, you are going to get interference patterns that are not necessarily associated with issues with how DLBC is phase matching speaker pairs. I am not convinced that your test is revealing a real-world issue. While I typically use DSU and not Auro3D for up mixing for 2.0 music content, I have never perceived any issues with the bass response when using DLBC.

While I mean no disrespect, especially since you have made a such a significant contribution with your custom GUI, I must decline running the measurements you requested, since I don't believe they will reveal any meaningful results.
Edit: I realized that my original instructions were inaccurate, I fixed them. All main speaker amps should be off during the measurements. This lets us measure the bass managed response from the subwoofers, when the same signal is sent to all speakers. So these measurements should not have any interference patterns.

Are you willing to test Auro 3D upmix of stereo music and compare how the bass response sounds between a standard Dirac calibration and DLBC? That is a five second test. If you do this test, please also include your current speaker layout.
 

·
Registered
Joined
·
1,297 Posts
When you play a mono signal through multiple speakers, you are going to get interference patterns that are not necessarily associated with issues with how DLBC is phase matching speaker pairs. I am not convinced that your test is revealing a real-world issue. While I typically use DSU and not Auro3D for up mixing for 2.0 music content, I have never perceived any issues with the bass response when using DLBC.
The issue is real and I can hear it with 2ch music Auro enabled with 4.2.4 speaker setup. Auromatic adjustment (because it adjusts upmix channels phase/level) effects what bass frequency is attenuated and you need to have music that happens to have bass impact/kick in that frequency range. But once you notice it, you sadly can't unnotice it... And because it just nulls bass, even when you add bass with tone or loudness, it just doesn't sound right.
And please notice that this is not just Auro problem, it's DLBC problem, Auro just happens to be an easy way to show it.
 

·
Registered
Joined
·
14,328 Posts
Discussion Starter · #3,328 · (Edited)
I'd like to see if other DLBC users like @AustinJerry or anyone else can reproduce the same issue I've encountered with multichannel/upmixed content and DLBC. DLBC optimizes stereo pairs, but it does not optimize any other combination of main speakers, such as L+C. That means DLBC can result in reduced bass at some frequencies when playing content from some pairs of speakers, depending on the all-pass filters used in the calibration.

To test, set the REW driver to Java and sweeps set to L+R, set the processor to all mono mode, and turn off all speaker amps, leaving only sub amps on. This allows us to measure the bass-managed subwoofer response when the same signal is sent to all speakers simultaneously. In the processor, disable all main channels except L+R. Keep the same number of subs that was used for the DLBC calibration. For my system this was a 2.2.0 speaker layout.
  1. With 2.x.0 speaker layout, take one measurement with regular Dirac, and another with DLBC. These are the reference measurements.
  2. Enable one other speaker or pair, and repeat measurements with regular Dirac and with DLBC. With regular Dirac, adding more speakers simply increases the SPL of the measurement across the whole frequency range, as expected. But with DLBC this was not always the case, depending on the particular speaker layout.
  3. Take as many combinations of speaker layouts as needed until the issue is found.
In my case, enabling the center speaker resulted in less bass centered at around 60 Hz with DLBC. And the speaker layout that demonstrated the issue most clearly on my particular setup and calibration was 4.2.4, shown below.

View attachment 3086758
I'd says that's expected behavior whenever you apply delay and/or all-pass filters and weight the optimization towards a certain combination of speakers, e.g. L/R/SW, C/SW, LS/RS/SW, etc. It's a tradeoff. Of course you could weight the optimization differently but DL doesn't have any controls for this.

If I were to do a crossover splice optimization manually I'd limit the filters on individually redirected bass channels to a common all-pass and (if the situation permits) a common low pass.
 

·
Registered
Joined
·
1,297 Posts
I'd says that's expected behavior whenever you apply delay and/or all-pass filters and weight the optimization towards a certain combination of speakers, e.g. L/R/SW, C/SW, LS/RS/SW, etc. It's a tradeoff. Of course you could weight the optimization differently but DL doesn't have any controls for this.
But the end result of DLBC's "expected behavior" can be pretty sad if you listen 2ch music with Auromatic upmixer with heights or tops. I never expected it to remove all bass impact of some music scores.
 

·
Registered
Joined
·
248 Posts
I'd says that's expected behavior whenever you apply delay and/or all-pass filters and weight the optimization towards a certain combination of speakers, e.g. L/R/SW, C/SW, LS/RS/SW, etc. It's a tradeoff. Of course you could weight the optimization differently but DL doesn't have any controls for this.

If I were to do this manually I'd limit the filters on individually redirected bass channels to a low pass.
Thanks. I agree DLBC is probably working as intended, but for multichannel processors I hope that they do end up making those types of controls available, because otherwise DLBC can end up with worse bass than vanilla Dirac in these particular edge cases.

I was wondering if that user reporting bass issues with the Power Rangers 2017 Atmos mix was actually encountering this issue, but unfortunately the user didn't provide enough raw data to be able to draw any conclusions. I'm planning to do some tests with that movie soon.

It seems many people don't use Auro, which is currently the easiest way to demonstrate this issue so far, so it doesn't seem like a big deal for those users. But if it can audibly affect movies as well, the issue probably deserves more attention.
 

·
Registered
Joined
·
14,328 Posts
Discussion Starter · #3,331 ·
Thanks. I agree DLBC is probably working as intended, but for multichannel processors I hope that they do end up making those types of controls available, because otherwise DLBC can up with worse bass than vanilla Dirac in these particular edge cases.

I was wondering if that user reporting bass issues with the Power Rangers 2017 Atmos mix was actually encountering this issue, but unfortunately the user didn't provide enough raw data to be able to draw any conclusions. I'm planning to do some tests with that movie soon.

It seems many people don't use Auro, which is currently the easiest way to demonstrate this issue so far, so it doesn't seem like a big deal for those users. But if it can audibly affect movies as well, the issue probably deserves more attention.
I agree. Will also try to run some measurements.
 

·
Registered
Joined
·
781 Posts
Ho ALL!

I would like to enquire regarding the use of a NAS, with a VM, running Windows 10, for REW and DIRAC. My particular NAS had a built in 3.5mm jack for mic and speaker, as well as a small speaker in the back of the NAS itself. ANyone with experience or advice regarding these two programs and the propriety of using a USB to 3.5mm adapter for the UMIK mic? The NAS also has an HDMI out, which I currently run to my HTP-1 and the video from the latter to my projector.

Thank you very much

FURY
 

·
Registered
Joined
·
188 Posts
USB to 3.5mm adapter for the UMIK mic?
Are you looking for a way to connect the male USB-A plug on a cable connected to the UMIK to the analog microphone in of your NAS? That's just plain not possible. You'll need to share a hardware USB port of the NAS into the VM, or use a USB Port network server and appropriate drivers (although the latter may be a source of many "interesting" problems).
 

·
Registered
Joined
·
14,328 Posts
Discussion Starter · #3,334 · (Edited)
I agree. Will also try to run some measurements.
@flax Houston, we have a bass control problem.

Looks like DLBC is applying individual all-pass filters to each redirected bass channel. This creates huge magnitude response distortions in the summed bass channel (orange trace below). Below some graphs for illustration.

The solution to this issue is to remove the individual all-pass filters from each redirected bass channel and use (and optimize) just a single all-pass filter downstream of the bass managed subwoofer summing stage.

My recommendation for now is NOT to use Dirac Live Bass Control and go back to "regular" DL or "upmix only".
Note: Use "upmix only" when you have a single sub or single sub cluster (multiple subs fed by a single subwoofer output) connected. Do NOT use "upmix only" with individual subs on individual outputs (unfortunately some processors do support this) as this will result in distortion similar to what is shown below.

3087256
 

Attachments

·
Registered
Joined
·
1,297 Posts
Note: Use "upmix only" when you have a single sub or single sub cluster (multiple subs fed by a single subwoofer output) connected. Do NOT use "upmix only" with individual subs on individual outputs (some processors support this) as this will result invalidate optimization.
Please explain what in "upmix only" is different vs. regular Dirac (I know it uses Dirac's bass management engine instead of device's own). I notice very little difference between those, using two subs (front/rear) on individual outputs.
 

·
Registered
Joined
·
14,328 Posts
Discussion Starter · #3,336 ·
Please explain what in "upmix only" is different vs. regular Dirac (I know it uses Dirac's bass management engine instead of device's own). I notice very little difference between those, using two subs (front/rear) on individual outputs.
The only difference to my knowledge is that "upmix only" uses Dirac's own bass management code instead of the processors's own bass management.
 

·
Registered
Joined
·
1,275 Posts
@pink soda @markus767

Many thanks to you both for putting the time in to find & verify this bug with DLBC, has a support ticket been opened with Dirac about this issue?

I would imagine it already has & I am very interested in what Dirac has to say about this & to know if they are aware of it.
 

·
Registered
Joined
·
14,328 Posts
Discussion Starter · #3,338 ·
@pink soda @markus767

Many thanks to you both for putting the time in to find & verify this bug with DLBC, has a support ticket been opened with Dirac about this issue?

I would imagine it already has & I am very interested in what Dirac has to say about this & to know if they are aware of it.
I didn't open a ticket. Do you want to do the honors? To be honest I'm a bit exhausted spending my time on doing QA for free. Especially when such fundamental design flaws are discovered in a software that has been years in the making from a company that specializes in room correction. They even might have been fully aware of the issue but did it anyway because it produces better looking graphs. You decide what is worse.
 
3321 - 3340 of 4511 Posts
Top