How to use REW to measure Dirac ART sound improvement?

[dlaloum]

RC = Room Correction, BC = Bass Control.
If you directly compare Dirac off with ART, the differences in frequency response (FR) and sound pressure level (SPL)
are so large that it becomes unclear what is actually being evaluated.
If the goal were simply to correct the FR, that could be done with other EQs as well.
Waterfall graphs are difficult to read and therefore not very useful.

Hmm I disagree - I find that Waterfalls provide all the main measurements in a single easy to read image!

Frequency Response, SPL, Decay...

 

[Paul McNeil]

Hmm I disagree - I find that Waterfalls provide all the main measurements in a single easy to read image!

Frequency Response, SPL, Decay...

Agree, 100%!

 

[Paul McNeil]

When you ask yourself a question like this, "what has ART done", you need to have some idea of what you are going to look for. According to its marketing claims, it improves the bass by smoothing the frequency response and reducing decay. You also need to know how to look for unwanted side effects. With ANY DSP, the price for any improvement in the frequency domain is paid in the time domain.

So to answer your question, you need to look at:

1. Has it improved the frequency response? I am pretty sure you know how to look to see if the frequency response.
2. Has decay been reduced? Take a look at book 2 of the REW eBook Section 4.8 (p60 onwards) on the waterfall and spectrogram.
3. What has happened in the time domain? Again, book 2, Section 4.4 (p44 onwards) on the impulse and step response.

I gave you the link to the eBook earlier in the thread. It's clear you didn't look at it, because if you did, you wouldn't need to ask these questions.

Note that you need specific settings for waterfall/spectrogram to best demonstrate the difference. Remember that any null in the freq response will "improve" the decay time on a waterfall or spectro, so you need to tick the "normalize to peak at each frequency" option.

When you ask yourself a question like this, "what has ART done", you need to have some idea of what you are going to look for. According to its marketing claims, it improves the bass by smoothing the frequency response and reducing decay. You also need to know how to look for unwanted side effects. With ANY DSP, the price for any improvement in the frequency domain is paid in the time domain.

So to answer your question, you need to look at:

1. Has it improved the frequency response? I am pretty sure you know how to look to see if the frequency response.
2. Has decay been reduced? Take a look at book 2 of the REW eBook Section 4.8 (p60 onwards) on the waterfall and spectrogram.
3. What has happened in the time domain? Again, book 2, Section 4.4 (p44 onwards) on the impulse and step response.

I gave you the link to the eBook earlier in the thread. It's clear you didn't look at it, because if you did, you wouldn't need to ask these questions.

Note that you need specific settings for waterfall/spectrogram to best demonstrate the difference. Remember that any null in the freq response will "improve" the decay time on a waterfall or spectro, so you need to tick the "normalize to peak at each frequency" option.

Keith_W, I had a look at your REW eBook. An excellent reference, I wish I had looked at it earlier. Thanks for that resource. And I see that you have used (apparently, no source other than ASR is given) some of my older data files, not from this thread but an earlier one. However, it was not clear what you concluded in answer to my question, though the latest measurements here seem to indicate, to me, that it does improve all three: FR, Decay and Time Domain alignment. I mention this because those who have not read your eBook would not be aware of this, it seems to me.

 

[Keith_W]

I am not sure if it was your measurements I had used, but the measurements of most ART systems look similar. I have looked at a few on ASR. It improves FR and decay, but at the expense of time domain performance. In every measurement I have examined, the main subwoofer impulse appears before the main impulse. I have heard an ART system only ONCE at an audio show, and I did not notice any problem (apart from wayyy too much bass - but I blame that on the guy who tuned the system and not on ART). It is as I keep saying, don't worry about what the measurement looks like if it is inaudible. Yes it's not good that the sub impulse appears before the main impulse. But if you can't hear it, it's fine.

I really need to go listen to more ART systems to see if I can detect it.

 

[Oddball]

I am not sure if it was your measurements I had used, but the measurements of most ART systems look similar. I have looked at a few on ASR. It improves FR and decay, but at the expense of time domain performance. In every measurement I have examined, the main subwoofer impulse appears before the main impulse. I have heard an ART system only ONCE at an audio show, and I did not notice any problem (apart from wayyy too much bass - but I blame that on the guy who tuned the system and not on ART). It is as I keep saying, don't worry about what the measurement looks like if it is inaudible. Yes it's not good that the sub impulse appears before the main impulse. But if you can't hear it, it's fine.

I really need to go listen to more ART systems to see if I can detect it.

That is a great point and I guess we ART-ists know that. It is obviously there by "design" as ART was in development for a mighty long time. My best guess is that they just could not make it all work, so FQ response and decay were prioritized as "most audible".

Can I hear the delay? Well, my ART sounds great, so I guess I am more than happy with it. If there was an option to eliminate mains delay, it would be an interesting to blind test it. Delay can be 20ms or more, and is not necessarily uniform across the spectrum.

Not sure if I will ever have time to try to fix this by myself, but it would be relatively easy to delay subs a bit with their DSP. Except that it would likely wreck all kinds of chaos in other aspects of the performance .

 

[janbth]

It improves FR and decay, but at the expense of time domain performance. In every measurement I have examined, the main subwoofer impulse appears before the main impulse.

What do you mean by this, taking into account there is no conventional crossover in ART? Do you have an example of a «subwoofer impulse» appearing before the «main impulse»?

 

[Oddball]

What do you mean by this, taking into account there is no conventional crossover in ART? Do you have an example of a «subwoofer impulse» appearing before the «main impulse»?

Like this one.

 

[Keith_W]

What do you mean by this, taking into account there is no conventional crossover in ART? Do you have an example of a «subwoofer impulse» appearing before the «main impulse»?

Page 51 of the book.

 

[janbth]

Like this one.

View attachment 501599

Thanks! I assume you are referring to the dotted line? This is «peak energy time» according to the REW help pages. Looks like the bass region is about 10 ms ahead of the high frequencies in this measurement. An alternative view is excess group delay, as mentioned in the REW help pages on Minimum phase (scroll a bit down). Any chance of producing such a plot from your data? It would be interesting to see.

 

[janbth]

Page 51 of the book.

View attachment 501600

It’s not clear to me what the «main impulse» and «subwoofer impulse» is from this.

 

[Oddball]

Thanks! I assume you are referring to the dotted line? This is «peak energy time» according to the REW help pages. Looks like the bass region is about 10 ms ahead of the high frequencies in this measurement. An alternative view is excess group delay, as mentioned in the REW help pages on Minimum phase (scroll a bit down). Any chance of producing such a plot from your data? It would be interesting to see.

Sorry but deleted 30GB of Dirac and REW data as that was needed for other purposes. Just couple of screenshots left and a whole lot of great sound. Clear that subs fire a bit early though.

 

[Keith_W]

It’s not clear to me what the «main impulse» and «subwoofer impulse» is from this.

The main impulse is indicated in the diagram. The subwoofer impulse is that big long impulse appearing before the main impulse.

 

[OCA]

I have yet to see a single proper post ART measurement. Where is the channel name? Is this a sub measurement or is this a speaker measurement? Or are they all playing together? What are you doing about the LPF for LFE setting which caps sub response in the AVR? How about Atmos channels? You cannot measure them with REW internal sweeps. I will keep waiting...

Here's my mystery measurement with no ART (yes bass region is not supposed to be all red):

 

[KenMasters]

What are you doing about the LPF for LFE setting which caps sub response in the AVR?

The option is greyed out when ART is used. It seems to be bypassed entirely, replaced by the balance of the sub's measured upper EQ limit and ART's 150Hz cut-off.

 

[OCA]

The option is greyed out when ART is used. It seems to be bypassed entirely, replaced by the balance of the sub's measured upper EQ limit and ART's 150Hz cut-off.

Good to know thanks for that info. Does the LFE measurement in REW look like there's a lowpass filter applied in the AVR or is it the sub's natural roll off?

 

[KenMasters]

Good to know thanks for that info. Does the LFE measurement in REW look like there's a lowpass filter applied in the AVR or is it the sub's natural roll off?

Well, if you look at the pre-cal measurement in ART, it's clear to see there's no filter in place:

And the LFE measurement in REW aligns well with ART's predicted response (also in evidence is the transition from ART to vanilla Dirac).

 

[OCA]

Well, if you look at the pre-cal measurement in ART, it's clear to see there's no filter in place:

View attachment 502132

And the LFE measurement in REW aligns well with ART's predicted response (also in evidence is the transition from ART to vanilla Dirac).

View attachment 502133

Looks like ART is applying a Linkwitz-Riley 36dB/octave lowpass at 150Hz. Could be 48db/oct, hard to tell but interesting choice of slope.

 

[dlaloum]

I have yet to see a single proper post ART measurement. Where is the channel name? Is this a sub measurement or is this a speaker measurement? Or are they all playing together? What are you doing about the LPF for LFE setting which caps sub response in the AVR? How about Atmos channels? You cannot measure them with REW internal sweeps. I will keep waiting...

Here's my mystery measurement with no ART (yes bass region is not supposed to be all red):

View attachment 502120

My post ART measurements - with ART active, specify the channel (I usually use L) and that channel has support from SW1, SW2, R, C, SR, SL channels - as that is my configuration.

It is notable that I am seeing lower THD for a supported channel with ART than running with no EQ

 

[Oddball]

All I know is that long time ago I stopped arguing that better graphs make for a better sound. Graphs need to fall within the acceptable range, which is a difficult term to by itself. What is acceptable usually develops after excessive use of EQ systems and REW. Otherwise, it is easy to tell that a range that is 2dB tighter to another measures better.

Getting the confidence, but not arrogance to choose what you like is a delicate balance. On some other forums I am corresponding to some really good calibrators. Like really good. But can't say that tips I get are making me any happier.

I guess when you are a redneck dog with a penguin on your head, your can only go so far.

 

[OCA]

My post ART measurements - with ART active, specify the channel (I usually use L) and that channel has support from SW1, SW2, R, C, SR, SL channels - as that is my configuration.

It is notable that I am seeing lower THD for a supported channel with ART than running with no EQ

Can you hear SW1, SW2, R, C, SR, SL play anything while measuring L?