Should we correct to Schroder, or full range?

[goat76]

Genelec's GLM and Neumann's MA-1 EQ correct above the room's Schroeder freq. to a limited extent. Does that mean they "tell you lies"?

I don't know what GLM and MA-1 use to calculate their corrections for different parts of the frequency range, but what we can assume is that those two programs have the benefits of knowing the "subject" in and out. They know exactly how the direct response looks like for the particular speaker model, and can therefore use both the measured room response and the known direct response when calculating what corrections are needed for the different parts of the frequency range.

Dirac on the other hand has no such knowledge of what the particular speaker's direct response looks like, for them, it's just an unknown "subject" and the only thing they can know when it comes to the frequency response of the unknown subject, is what the microphone has picked up at a bunch of non-reflection-free spots around the main listening area.

 

[CapMan]

Dirac on the other hand has no such knowledge of what the particular speaker's direct response looks like, for them, it's just an unknown "subject" and the only thing they can know when it comes to the frequency response of the unknown subject, is what the microphone has picked up at a bunch of non-reflection-free spots around the main listening area

I’m going to make the heroic assumption here that the development team at Dirac aren’t idiots, that they know something of acoustics, maths and coding…

I don’t think any of us know how their algorithm is built.

Maybe I’m wrong.

 

[Deleted member 48726]

I’m going to make the heroic assumption here that the development team at Dirac aren’t idiots, that they know something of acoustics, maths and coding…

I don’t think any of us know how their algorithm is built.

Maybe I’m wrong.

We all know the real experts reside in hibernation here on ASR. -For some odd reason they are not out there, establishing the truth in room correction software and making big bucks in the process.

/Some slight sarcasm may have been used in the above. If someone feels violated I will of course be willing to call it satire.

 

[Keith_W]

I used Acourate a while back and recall it uses windowed measurements and a psycho-acoustic representation of the response to avoid over correction. @Keith_W is way more experienced and could probably confirm.

Yup, I can confirm that this is what Acourate does. By default, Acourate lets you set the resolution of low and high frequencies independently.

 

[sigbergaudio]

I know you want something to pick apart.

You also want to make it into good or bad. How about that I think it sounds amazing. The approach must be good. In my case at least. I don't know how Dirac works.
I don't boost, only cutting.

I am happy that it sounds amazing.

But that you got Dirac to work for you doesn't necessarily mean full range correction is a great idea in any situation, especially when done manually. On the contrary we know for a fact that it can be detrimental. That one person has a good result isn't exactly conclusive. Some automatic correction systems are careful with what they do above Schroeder, and then it works fine. Some take the brute force approach, and then it will be hit and miss depending on the room and the speakers.

 

[vicenzo_del_paris]

Do you have measurements of what Dirac does in your system above say 500hz? So before after graphs from MLP?

An example of dirac full range correction ( measured at MLP - 1/12 smoothing)
mdat file also included.

 

[Deleted member 48726]

I am happy that it sounds amazing.

But that you got Dirac to work for you doesn't necessarily mean full range correction is a great idea in any situation, especially when done manually. On the contrary we know for a fact that it can be detrimental. That one person has a good result isn't exactly conclusive. Some automatic correction systems are careful with what they do above Schroeder, and then it works fine. Some take the brute force approach, and then it will be hit and miss depending on the room and the speakers.

I have never said that it's magic. Only that it works wonders for me. I'm against the concept of putting things in boxes in black or white theme. Especially just because one guy has written a book once. It reminds me of religion in a bad way and it seems to me like audiophile gibberish but from the other camp.. Like opposing something new you don't quite understand.
I don't know how Dirac Lives' calculations are done either. I can only relate to the result.

 

[Deleted member 48726]

An example of dirac full range correction ( measured at MLP - 1/12 smoothing)
mdat file also included.

View attachment 347006

Thanks. You've done what I'm too lazy to do. That looks like a very nice correction. My curve (as seen from Dirac Live) looks like that but with a bit steeper decline. It does however looks like it's boosting a bit too much for my taste around 150 and 18 Hz.
-How does it sound?

 

[vicenzo_del_paris]

Thanks. You've done what I'm too lazy to do. That looks like a very nice correction. My curve (as seen from Dirac Live) looks like that but with a bit steeper decline. It does however looks like it's boosting a bit too much for my taste around 150 and 18 Hz.
-How does it sound?

This is not the best correction I managed to get both in terms of measurements and sound. I recently had to move speakers and changed some furnitures.
Still not very satisfied with non corrected baseline as the dips around 130hz and 2800hz were not there before the changes in the room.
But Dirac always did a good job for me.
I tried limiting to few hundreds Hz but I always preferred full range (in different rooms and speakers over years).
Current setup sounds good, while there still is some room for enhancement.
Most noticeable difference (except of course better smooth bass and room modes controlled) is about vocals that are still very centered but less detached from instruments ans also less forward (due to FR being smoothed.)

 

[CapMan]

Some take the brute force approach, and then it will be hit and miss depending on the room and the speakers.

Which are brute force and which are not , it would useful info for people making a choice on the tool they use. Tnx

 

[vicenzo_del_paris]

Thanks. You've done what I'm too lazy to do. That looks like a very nice correction. My curve (as seen from Dirac Live) looks like that but with a bit steeper decline. It does however looks like it's boosting a bit too much for my taste around 150 and 18 Hz.
-How does it sound?

This is not default correction. Had to reduce the low end boost Dirac proposed. And will reduce it again I think.
But first I want to find the optimal speakers placement with the changes made to my living room

 

[goat76]

I’m going to make the heroic assumption here that the development team at Dirac aren’t idiots, that they know something of acoustics, maths and coding…

I don’t think any of us know how their algorithm is built.

Maybe I’m wrong.

I don't think they are idiots, and I hope you didn't get the impression that I think they are idiots.

I think they are a team of highly knowledgeable people, but they are also fully aware and smart enough to know they are selling a product that is highly dependent on the 'ease of use' for the end consumer. Most of the users are not "measuring nerds" like people on this forum, the measuring part must therefore be as straightforward and as uncomplicated as possible. If they start asking the consumers to drag their speakers out in the open, to get a reflection-free measurement as possible that could be used to generate a direct response of their loudspeakers, they would flush their selling point of 'ease of use' down the toilet and probably lose many of their customers who would now likely think Dirac is way too complicated.

Most consumers want the measurement part done quickly and with easy instructions, and they want the program to handle the corrections automatically with as little user involvement as possible. Most users have full trust in the program and they can easily see that the resulting response curve looks way better than before. Confirmation bias now kicks in, and it now also sounds way better than before which is probably true. But, it's still possible it could have been even better if Dirac had given the additional instructions to make measurements that could be gated, which then could generate a direct response from the speaker that they could use as a complement to the other measurements. The last part is unfortunately hard to sell to the large consumer market.

So if we look at the big picture, I don't think Dirac is fully committed to just sound quality, even if the result of course must sound better with the corrections than without. But it's also about selling a product which goes hand-in-hand with the ease of use.

 

[Flaesh]

An example of dirac full range correction ( measured at MLP - 1/12 smoothing)
mdat file also included.

View attachment 347006

There is only one uncorrected measurement

 

[CapMan]

So if we look at the big picture, I don't think Dirac is fully committed to just sound quality, even if the result of course must sound better with the corrections than without. But it's also about selling a product which goes hand-in-hand with the ease of use.

I understand your points, but are ease of use and achieving a high quality result mutually exclusive?

My iPhone camera is a breeze to use, but the work that has gone into the software that drives it yields incredibly good and consistent results.

Honestly I think it’s a leap to suggest Dirac aren’t committed to sound quality, but just my option.

Have a great weekend

 

[vicenzo_del_paris]

There is only one uncorrected measurement

In the posted mdat file, I kept 1 measurement uncorrected and 1 corrected for simplicity and clarity.
I could have included few of each.
Discarded ones were very close to kept ones for both uncorrected and corrected.

 

[GXAlan]

On the contrary we know for a fact that it can be detrimental. That one person has a good result isn't exactly conclusive. Some automatic correction systems are careful with what they do above Schroeder, and then it works fine. Some take the brute force approach, and then it will be hit and miss depending on the room and the speakers.

There is a difference between possibility and probability. Correcting a Bose 901 using full range Dirac works great since the MLP measurement matches the Bose factory EQ for the most part. Phase is not an issue since you have so many reflections anyway, although I wouldn’t be surprised if it ends up producing some interaural crosstalk cancellation.

Here is my question. We KNOW that correcting above Schroeder is UNPREDICTABLE for all the reasons that can be cited.

The real question is: how often do you win the lottery, correct above transition frequency, and get a better result through pure luck? 50%? 25% 75%?

 

[vicenzo_del_paris]

Sorry, I just realized what you meant.
The file I posted indeed contained only the uncorrected measurement.
Here is the mdat with both

 

[Deleted member 48726]

I don't think they are idiots, and I hope you didn't get the impression that I think they are idiots.

I think they are a team of highly knowledgeable people, but they are also fully aware and smart enough to know they are selling a product that is highly dependent on the 'ease of use' for the end consumer. Most of the users are not "measuring nerds" like people on this forum, the measuring part must therefore be as straightforward and as uncomplicated as possible. If they start asking the consumers to drag their speakers out in the open, to get a reflection-free measurement as possible that could be used to generate a direct response of their loudspeakers, they would flush their selling point of 'ease of use' down the toilet and probably lose many of their customers who would now likely think Dirac is way too complicated.

Most consumers want the measurement part done quickly and with easy instructions, and they want the program to handle the corrections automatically with as little user involvement as possible. Most users have full trust in the program and they can easily see that the resulting response curve looks way better than before. Confirmation bias now kicks in, and it now also sounds way better than before which is probably true. But, it's still possible it could have been even better if Dirac had given the additional instructions to make measurements that could be gated, which then could generate a direct response from the speaker that they could use as a complement to the other measurements. The last part is unfortunately hard to sell to the large consumer market.

So if we look at the big picture, I don't think Dirac is fully committed to just sound quality, even if the result of course must sound better with the corrections than without. But it's also about selling a product which goes hand-in-hand with the ease of use.

That's speculative at best. At worst it's FUD.

 

[Tangband]

I don't think they are idiots, and I hope you didn't get the impression that I think they are idiots.

I think they are a team of highly knowledgeable people, but they are also fully aware and smart enough to know they are selling a product that is highly dependent on the 'ease of use' for the end consumer. Most of the users are not "measuring nerds" like people on this forum, the measuring part must therefore be as straightforward and as uncomplicated as possible. If they start asking the consumers to drag their speakers out in the open, to get a reflection-free measurement as possible that could be used to generate a direct response of their loudspeakers, they would flush their selling point of 'ease of use' down the toilet and probably lose many of their customers who would now likely think Dirac is way too complicated.

Most consumers want the measurement part done quickly and with easy instructions, and they want the program to handle the corrections automatically with as little user involvement as possible. Most users have full trust in the program and they can easily see that the resulting response curve looks way better than before. Confirmation bias now kicks in, and it now also sounds way better than before which is probably true. But, it's still possible it could have been even better if Dirac had given the additional instructions to make measurements that could be gated, which then could generate a direct response from the speaker that they could use as a complement to the other measurements. The last part is unfortunately hard to sell to the large consumer market.

So if we look at the big picture, I don't think Dirac is fully committed to just sound quality, even if the result of course must sound better with the corrections than without. But it's also about selling a product which goes hand-in-hand with the ease of use.

Ofcourse its all about the money and there is a market for this which is shown in this thread . They are not idiots, but they cant make it to complicated for the customers.
Its no different than selling loudspeaker cables for 1000 dollars, but to be fair, Dirac really works as they say below the transition area.

And everyone can do corrections that makes the sound better with a good dsp , a soundcard and a measurement microphone, but you have to do it right:

1. Measure and correct frequencies below transition area with the mic at listening position .
2. Measure and correct frequencies above transition ( 200-400 Hz ) with the microphone 1 meter from one of the loudspeakers.

 

[CapMan]

Its no different than selling loudspeaker cables for 1000 dollars,

That’s just nonsense.