Targets for Dirac Live

[popej]

I have replaced my Yamaha AVR with Pioneer. While Yamaha does its room correction automatically, Dirac Live expect users to provide target curve. Now I'm trying to get the sound I like. My target consist of Harman curve with added loudness correction. Actually I have prepared a whole set of targets, which I have attached to this post.

Target names are "target-rXX-sYY-hZZ.txt", where XX is reference level of recording, YY is SPL level of playback and ZZ is Harman curve slope from 20Hz to 20kHz. I have used reference level 85dB SPL. This is a value I see when dealing with Dolby or THX. Maybe other values could be interesting too, like for example 89dB used for reply gain. Loudness is calculated according to ISO 226. It is defined up to 12.5kHz. I haven't tried to extrapolate for higher frequencies, I have repeated the last provided value for 16k and 20k.

I hope you can try these curves and provide opinions

 

[Mikel_Aguirre]

Thanks for sharing. It looks like Harman curves taking into account the Fletcher-Munson equal loudness contours. The h5 looks to me the one that have more similar slope to the traditional Harman curve. If I usually listen to an average of 75 db and want something close to Harman slope but adapted to that loudness, will it be correct to assume that the best target to use in my case would be target-85-s75-h5.txt?
This is very useful for people like me. I have my main system but also a secondary one in the bedroom that usually listen to at low volumes. I use dirac on both.

 

[popej]

If I usually listen to an average of 75 db and want something close to Harman slope but adapted to that loudness, will it be correct to assume that the best target to use in my case would be target-85-s75-h5.txt?

Yes, that's the idea. I don't know if Harman slope should change with volume, I think we can experiment a bit.

 

[popej]

Pioneer has 9 bands EQ. This can be used for loudness, but I don't know how precise are EQ settings. Assuming that we set Dirac for flat frequency response and reference level is 85dB SPL, then EQ for lower SPL would be following:

freq:    63    125    250    500    1000    2000    4000    8000    16000
80dB:    1.9    1.3    0.8    0.3    0.0    -0.2    -0.2    0.0    0.8
75dB:    3.8    2.7    1.6    0.6    0.0    -0.4    -0.4    0.1    1.5
70dB:    5.6    4.0    2.4    0.9    0.0    -0.5    -0.6    0.1    2.3
65dB:    7.5    5.4    3.1    1.2    0.0    -0.7    -0.8    0.2    3.0
60dB:    9.3    6.7    3.9    1.5    0.0    -0.9    -1.0    0.2    3.7
55dB:    11.2   8.0    4.7    1.8    0.0    -1.1    -1.2    0.3    4.4
50dB:    13.0   9.3    5.4    2.1    0.0    -1.3    -1.4    0.3    5.1

It can be usable to set EQ for 10dB volume reduction without changing perceived tone balance:

freq:   63      125     250     500     1000    2000    4000    8000    16000
-10dB: -6.5    -7.5    -8.5    -9.5    -10.0   -10.5   -10.5   -10.0   -8.5

This should work in quite wide volume range.

 

[Mikel_Aguirre]

I haven’t had the chance to experiment with these. But it looks like it could work quite nicely. In the past I had set similar V shaped eq-s in Roon with different degrees of slope to use when listening at low volumes. It was all done subjectively. Your approach is much more scientific. I really want to give these a try when I have some free time

 

[Davide]

I recently fine-tuned the Dirac BC target and was amazed by how it determines realism or not.
Clearly the process consisted of measurements with REW to verify the result. And in the listening point the curve is actually a straight line if represented logarithmically.
I only added a small boost around 4.5 kHz, which picks up a bit on the original response of the speakers, because it actually gives it more presence.

That's all:

This sounds very good at the standard 80dB level, but lowering there is a noticeable lack of bass.
I also had the idea of implementing ISO226 curves with Dirac but switching presets is really boring ...
If anyone knows a plugin it could be a great alternative.

 

[Ciobi69]

I recently fine-tuned the Dirac BC target and was amazed by how it determines realism or not.
Clearly the process consisted of measurements with REW to verify the result. And in the listening point the curve is actually a straight line if represented logarithmically.
I only added a small boost around 4.5 kHz, which picks up a bit on the original response of the speakers, because it actually gives it more presence.

That's all:
View attachment 361795

This sounds very good at the standard 80dB level, but lowering there is a noticeable lack of bass.
I also had the idea of implementing ISO226 curves with Dirac but switching presets is really boring ...
If anyone knows a plugin it could be a great alternative.

whice one is the sub? it may be over boosted by a lot

 

[Davide]

whice one is the sub? it may be over boosted by a lot

The subs response is blue and yellow. If you are referring to the boost to the lower limit of 20Hz, it is not overboosted. Neither as Response (measured with REW) nor as XMAX of drivers (I have accurately calculated the maximum power to not exceed it).
Also, consider that the subs are two and that graph shows the estimated response of their sum, individually they have a lower boost than represented.

I add to the discussion, that according to ISO226 curves, from a mathematical point of view, the extremes of frequency should be attenuated beyond the reference level (to maintain the same perceived loudness). But in fact this greatly alters the perception of the timbre, so these curves are to be taken a bit with a grain of salt.
Surely they work good when attenuating with respect to the reference.

 

[Ciobi69]

The subs response is blue and yellow. If you are referring to the boost to the lower limit of 20Hz, it is not overboosted. Neither as Response (measured with REW) nor as XMAX of drivers (I have accurately calculated the maximum power to not exceed it).
Also, consider that the subs are two and that graph shows the estimated response of their sum, individually they have a lower boost than represented.

I add to the discussion, that according to ISO226 curves, from a mathematical point of view, the extremes of frequency should be attenuated beyond the reference level (to maintain the same perceived loudness). But in fact this greatly alters the perception of the timbre, so these curves are to be taken a bit with a grain of salt.
Surely they work good when attenuating with respect to the reference.

Good job then you know what you are doing. What subs are those?

 

[Davide]

Good job then you know what you are doing. What subs are those?

DIY sealed with Scan Speak Discovery 12".
Here and here I posted some details. In the second post you can see SPL vs Freq graph, both with Xmax and Power limits. I should reach 96dB SPL at 20hz at 3 meters remaining in the linear excursion range.

Anyway, returning to the topic.
To provide Dirac targets with ISO226 curves integrated it is necessary that the user know exactly the SPL level at the listening point.
Also, this spl level is referred to 1kHz in case of ISO226.
Maybe it is appropriate to specify it for those who intend to try to use them correctly.