Sharing your Dirac target curves

[RiseFall123]

Hi all,

 

it is possibile for you that are using Dirac Live, to share an image (and possibly paste the TXT) of your target curves?

 

That will be useful to me to understand what kind of curve is the most used and most appreciate.

 

Actually I am using the stereo version of Dirac and a slightly modified curve taken from this document:

 

https://www.minidsp.com/support/community-powered-tutorials/263-custom-dirac-live-target-curves

 

The result is good but sometimes I think the sound is too bright.

 

Thank you!

[sebastien696]

Hi all,

 

it is possibile for you that are using Dirac Live, to share an image (and possibly paste the TXT) of your target curves?

 

That will be useful to me to understand what kind of curve is the most used and most appreciate.

 

Actually I am using the stereo version of Dirac and a slightly modified curve taken from this document:

 

https://www.minidsp.com/support/community-powered-tutorials/263-custom-dirac-live-target-curves

 

The result is good but sometimes I think the sound is too bright.

 

Thank you!

Hello,

 

I try to stay flat and follow my speakers for high frequencies. I have -8db at 20KHz at my listening point, which is 3m away from the speakers (natural roll-off from my speakers). I believe it might be more or less depending on your listening position (and I suppose also depending on the room and your speakers). I found information at http://www.ohl.to/calculators/targetcurve.php

and also in French at Courbes ISO 2969 X et Bruel & Kjaer

 

[Fitzcaraldo215]

Hi all,

 

it is possibile for you that are using Dirac Live, to share an image (and possibly paste the TXT) of your target curves?

 

That will be useful to me to understand what kind of curve is the most used and most appreciate.

 

Actually I am using the stereo version of Dirac and a slightly modified curve taken from this document:

 

https://www.minidsp.com/support/community-powered-tutorials/263-custom-dirac-live-target-curves

 

The result is good but sometimes I think the sound is too bright.

 

Thank you!

 

Just because you can modify the target curve does not mean you should. I myself have been perfectly happy using the downward sloping default Dirac target curve, unmodified, for years. Before that, I was very happy for years using the similar Audyssey downward sloping default target curve for years. (I did modify it in Audyssey Pro, but only to eliminate the poorly chosen Midrange Compensation dip, aka the BBC dip. Fortunately, Dirac does not have that particular wrinkle.)

 

I think there is also excellent theoretical support for a downward sloping target curve in most any real world room. Instrumentation mike maker B&K, for example recommends a target curve almost identical to Dirac's. Many other acoustics experts also agree.

 

Austinjerry's curve emphasizes the bass but it leaves the high frequencies flat. I am not surprised if it sounds bright, because I am convinced that in-room curves should slope downward in the highs.

 

Dirac chose their default curve for a reason and based on considerable research. Austinjerry is a nice guy who I have interacted with in forums, but I do not think his DIY research is as definitive or credible.

 

But, if you insist on experimenting, I would start with the downward sloping Dirac default and go from there.

[wgscott]

I'm also a big fan of the default curve, but I also sometimes use one with a little bit of a bass hump. I also made a flat one, and a few others, because I am a scientist and I think you should play with your equipment until it breaks. (This does not apply to your personal "equipment", BTW.)

[Fitzcaraldo215]

I'm also a big fan of the default curve, but I also sometimes use one with a little bit of a bass hump. I also made a flat one, and a few others, because I am a scientist and I think you should play with your equipment until it breaks. (This does not apply to your personal "equipment", BTW.)

 

The Dirac default does boost the bass slightly in linear fashion with decreasing frequency, as does the B&K curve.

 

But, some are more "bass freakish". Others listen at lower volumes, so deep bass needs more boosting per Fletcher-Munson.

 

I agree, if more bass boost sounds better to you, crank that into the target. But, at the other end of the spectrum, slight high end rolloff, as per the Dirac default, sounds best, I think.

[Le Concombre Masqué]

this works for me

The Dirac default does boost the bass slightly in linear fashion with decreasing frequency, as does the B&K curve.

 

But, some are more "bass freakish". Others listen at lower volumes, so deep bass needs more boosting per Fletcher-Munson.

 

I agree, if more bass boost sounds better to you, crank that into the target. But, at the other end of the spectrum, slight high end rolloff, as per the Dirac default, sounds best, I think.

[Le Concombre Masqué]

Then (successfully) felt like trying a new target curve, the one with the 1 slope from 200 towards LF/0.4 towards HF. Note that it's the best fit with the uncorrected response : maybe there's no universal best target curve ; fit matters too. A 70/+3 Low Shelf does the trick when it comes to match the target curve in the LF ; I leave the rest to REW

this works for me[ATTACH=CONFIG]33137[/ATTACH]

[sebastien696]

Hello,

 

Your measurement is at a single location? as opposed to multiple locations with Dirac? (which I suppose is averaged)

My speakers are supposed to be more or less flat on-axis but are more decreasing with frequency off-axis (more especially over 10Khz). Measuring with Dirac at multiple locations is supposed to give me a more decreasing curve than a single measurement on-axis. I also took my measurements at different heights. the mic was not at tweeter level.

 

 

[ATTACH=CONFIG]33181[/ATTACH][ATTACH=CONFIG]33182[/ATTACH][ATTACH=CONFIG]33183[/ATTACH]Then (successfully) felt like trying a new target curve, the one with the 1 slope from 200 towards LF/0.4 towards HF. Note that it's the best fit with the uncorrected response : maybe there's no universal best target curve ; fit matters too. A 70/+3 Low Shelf does the trick when it comes to match the target curve in the LF ; I leave the rest to REW

[Le Concombre Masqué]

One can use REW the Dirac way, averaging different measurements... Actually, I picked one and you're right there's no such thing as a single point sweet spot. Hence I chose the measurements that allowed the best sounding corrections. Plus I don't over correct : +-3 (settings ; actual cutting filters go further) that's it. One should not mess too much with high frequencies : shorter wave lengths, greater risks of miscorrections , beyond, I think, the averaging mitigation

Hello,

 

Your measurement is at a single location? as opposed to multiple locations with Dirac? (which I suppose is averaged)

My speakers are supposed to be more or less flat on-axis but are more decreasing with frequency off-axis (more especially over 10Khz). Measuring with Dirac at multiple locations is supposed to give me a more decreasing curve than a single measurement on-axis. I also took my measurements at different heights. the mic was not at tweeter level.

[Fitzcaraldo215]

One can use REW the Dirac way, averaging different measurements... Actually, I picked one and you're right there's no such thing as a single point sweet spot. Hence I chose the measurements that allowed the best sounding corrections. Plus I don't over correct : +-3 (settings ; actual cutting filters go further) that's it. One should not mess too much with high frequencies : shorter wave lengths, greater risks of miscorrections , beyond, I think, the averaging mitigation

 

Ok, glad you are happy with your approach and that it sounds good to you.

 

However, I am not sure I have ever seen any confirmation of your views anywhere. I do not think anyone can find confirmation of "one should not mess too much with high frequencies", "greater risks of miscorrections", etc. So, I think you will find no theoretical basis to support your view among credible acousticians. But, on the other hand, no one can argue with your subjective preferences. Happy listening!

[Le Concombre Masqué]

sheer logic : the shortest the wave length, the more pinpoint microphone placement specific the measure is ; I think you could compare multiple measurements done at slightly different locations and see marked differences. Averaging is a solution, not going too far correcting responses that would be different if measurements were done inches away is my suggestion. Here is my corrected response. I can show it smoothed or... with no smoothing : have a look at the highs...

Ok, glad you are happy with your approach and that it sounds good to you.

 

However, I am not sure I have ever seen any confirmation of your views anywhere. I do not think anyone can find confirmation of "one should not mess too much with high frequencies", "greater risks of miscorrections", etc. So, I think you will find no theoretical basis to support your view among credible acousticians. But, on the other hand, no one can argue with your subjective preferences. Happy listening!

[Fitzcaraldo215]

[ATTACH=CONFIG]33281[/ATTACH][ATTACH=CONFIG]33282[/ATTACH][ATTACH=CONFIG]33283[/ATTACH]sheer logic : the shortest the wave length, the more pinpoint microphone placement specific the measure is ; I think you could compare multiple measurements done at slightly different locations and see marked differences. Averaging is a solution, not going too far correcting responses that would be different if measurements were done inches away is my suggestion. Here is my corrected response. I can show it smoothed or... with no smoothing : have a look at the highs...

I am well aware of the comb filtering due to reflections and other seemingly erratic behavior in high frequency measurements due to the short wavelengths. That is why tools such as Dirac employ techniques like multipoint averaging and frequency smoothing. They avoid foolishly correcting every little squiggle in the measured frequency response. These issues are well known in acoustics.

 

If you look at measured HF response up close, unfiltered and unsmoothed, yes it will be extremely jagged. Step back, smooth the response, and you will see the average tendencies of the frequency response. It is known that psychoacoustically the human hearing mechanism provides perceptual response smoothing as a function of increasing frequency. The higher the frequency, the greater the degree of natural smoothing.

 

Dirac's filtering affects the "envelope" or the "average response" in the high frequencies, which is more like our perception, and it generally avoids the problems that seem to be of concern to you.

 

By the way, I assume your measurements are single point, meaning not spatially averaged. Personally, I prefer spatially averaged measurements. Many professional acoustics experts agree. There will likely be a discrepancy between your single point measurements and the corrective filters Dirac has calculated using spatial averaging.

[Le Concombre Masqué]

I tried Dirac first ; REW allows you to average measurements and create corrective filters based on averaging.

 

BTW seems that my target is not far from Dirac's default. They also certainly know that filters should apply in minimum phase zones : our opinions and drc philosophy might not differ as much as your post suggested... While our French friend's target is far from Dirac's default...

I am well aware of the comb filtering due to reflections and other seemingly erratic behavior in high frequency measurements due to the short wavelengths. That is why tools such as Dirac employ techniques like multipoint averaging and frequency smoothing. They avoid foolishly correcting every little squiggle in the measured frequency response. These issues are well known in acoustics.

 

If you look at measured HF response up close, unfiltered and unsmoothed, yes it will be extremely jagged. Step back, smooth the response, and you will see the average tendencies of the frequency response. It is known that psychoacoustically the human hearing mechanism provides perceptual response smoothing as a function of increasing frequency. The higher the frequency, the greater the degree of natural smoothing.

 

Dirac's filtering affects the "envelope" or the "average response" in the high frequencies, which is more like our perception, and it generally avoids the problems that seem to be of concern to you.

 

By the way, I assume your measurements are single point, meaning not spatially averaged. Personally, I prefer spatially averaged measurements. Many professional acoustics experts agree. There will likely be a discrepancy between your single point measurements and the corrective filters Dirac has calculated using spatial averaging.