• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

does anyone have the good old Dirac Live linear target curve?

Florin Andrei

Member
Joined
Feb 21, 2019
Messages
38
Likes
36
For many years, I've used Dirac Live on a miniDSP nanoAVR HD for a 5.1 system where all connections were HDMI. I set it up once and it worked flawlessly for a long time.

Recently, I moved things around, so a re-calibration was necessary. Also, the nanoAVR HD is not supported anymore, the old apps are old and creaky, etc. So I got an Onkyo AVR that supports Dirac Live. I did the measurements with the latest Dirac Live app.

To my shock and dismay, the old linear target curve is gone. It looked like this:

target-default-control-points.png


Apparently, in 2022 they came up with some "new and improved" commercial yadda-yadda. There does not seem to be a way in the new app to get the old curve back. This boggles the mind, since the old curve was spot-on in terms of results, it's literally the main reason I use Dirac. The new stuff manages to sound both bloated and thin at once, and I can't stand it.

In theory, I could try to draw my own curve, but the old default was perfect in every way, and I do not remember the slope, etc.

Please, does anyone have the old, perfectly linear curve saved in a file somewhere?

Thank you.
 
I created my own for my miniDSP. I used -1dB per octave.

I sent you a private message with the file attached. Perhaps it works for you. Change the extension of the file from ".txt" to ".targetcurve". I had to change it to attach the file to the private message.
 
Last edited:
Thank you @terryforsythe I will give it a try.

Does anyone know what is the slope of the original, linear target curve? Just eyeballing the image, it seems closer to -0.6 dB / octave.

Also, seems like just two points should be enough to define the slope, no?
 
Plot twist:

I have opened a support case with Dirac. They said if you choose the point-by-point mode and select “Set default target curve” in the drop-down menu, the expectation is that you should get the linear curve as before.

The fact that I am not getting the linear curve was surprising to the support engineer, they were able to reproduce the issue I'm seeing, and they've escalated the case.

I'll keep an eye on the support case. Now this is interesting.
 
Last edited:
Does anyone know what is the slope of the original, linear target curve? Just eyeballing the image, it seems closer to -0.6 dB / octave.
I experimented with various slopes from -0.75dB/octave to -1.2dB/octave. Originally I started using -0.9dB/octave. My son had a listen and thought it was too bright. The more I listened to it the more I tended to agree. So, I changed it to -1.0dB/octave. Both he and I like it better.

But, that doesn't mean it is best for everybody. My in-room response probably is not typical - I have a large room with some highly reflective elements, such as a front wall with a lot of tile. The floor also is tile, but I have a throw rug on it.

Also, seems like just two points should be enough to define the slope, no?
Perhaps. It seems like it would work, but I have not tried it.
 
I wouldn’t be obsessed to force a linear target curve if your actual measurements differ a lot from the target. Excess EQ treatment creates un-natural sound. I suggest to “mimetic” slope and tendency, just improving some frequencies where you find peaks or valleys, avoiding any +5dB or any -15dB. And difference between actual slope and target slope should be +\- 3dB max. Other than this will sound artificial.
 
"Un-natural" or "artificial" in the vast, vast majority of cases simply means "my ears have become accustomed to a certain sound".

What is truly artificial in an ear-independent way is sound that deviates from the original sound recorded in the studio. Deviations are removed via EQ. That's all.
 
When you buy a speaker I assume you like the sound emits. That’s what I would call its natural sound. EQing should be the trick to perfect the sound. When you have to change a lot that base sound you should change the speakers.
 
There is no advantage for using a current version of Dirac for your use case. Uninstall what you have and install version 3.4.4. It will have the target curve you seek which is 1/2 dB per octave.
You download from Dirac here:
 
@Miniyouuuu Buying audio gear based on feelings is exactly the wrong assumption here.

Yes, I am aware that most "audiophiles" out there have a tendency to worship their own equipment. I am not an "audiophile". Equipment is just a bunch of tools - cattle, not pets, nor idols. It's there to serve your goals, not the other way around. Hopefully that clarifies things a bit.
 
Here's the target in your screenshot digitized, then analyzed in REW:
View attachment 419381

I've also attached the target below.
This will work but will make the target curve appear really ‘thick’ as Dirac places a “handle” at each point in the file. iirc there’s only about 10 points in a Dirac target curve file along with some other info.
 
@JStewart Yeah, that's what I assumed. They had a blog post around April 2022 where they announced a big change in the way filter design works. It's full of commercial yadda-yadda, but the gist appears to be that now shelf-based target curves reign supreme. It seems quite clear they've moved away from measurements-based criteria, and towards satisfying the taste of a large chunk of their customers, which is dubious to say the least, especially in the audio world. But a company needs to do what it must to turn a profit.

I am going to test the last version that came out before the announcement. I just didn't have time to do it yet. I assume the old target can be exported from it somehow.

And there's also the option of crafting your own target by directly placing points on the graph, so the original linear target could probably be rebuilt regardless of the app version, as long as you know the slope. You would think, in this day and age, a computer could handle that labor, lol.
 
@Miniyouuuu Buying audio gear based on feelings is exactly the wrong assumption here.

Yes, I am aware that most "audiophiles" out there have a tendency to worship their own equipment. I am not an "audiophile". Equipment is just a bunch of tools - cattle, not pets, nor idols. It's there to serve your goals, not the other way around. Hopefully that clarifies things a bit.
I don’t say feelings. You buy a speaker because you have your own criteria. Some will buy because of its aesthetic others by the response graph or specs they see on a review or the bass the feel in the store. But finally you have them at home and you want to adjust their response to your room by EQing. If you need to adjust them too much I say that maybe you buyed the wrong speaker.
 
It seems quite clear they've moved away from measurements-based criteria, and towards satisfying the taste of a large chunk of their customers,
I have different thoughts on their reasoning.
1. Floyd Toole and I’m sure others, have found EQ above the room’s transition frequency is not a good idea when based on listening position measurements and when anhecoic response and directivity are either unknown or not conducive to EQ. Dirac’s new target curve will not alter this response as much as a fixed slope target could. So one could make the case it’s a better method. From a marketing perspective I’m sure it’s just easier for Dirac to say the target retains your speaker’s character, which they do.
2. Over the years many customers complained about having their bass “neutered”. The new target curve will leave more bass intact , if present in the measurements. This saves time building target curves on one’s own. The levels at both ends of the frequency band can be adjusted to taste without having to recalculate slopes by other means.
Summary: an attempt at more science based approach and more convenience for customers.

I also heard Dirac removed the straight line slope target from the most recent version. If true I think this is a mistake because some customers, including me, prefer it with the individual setup we have. Individual preference is never wrong. Choice is good, in this case.

I am going to test the last version that came out before the announcement. I just didn't have time to do it yet. I assume the old target can be exported from it somehow.

The version I suggested 3.4,4 was the last version before Dirac started adding code to the tool to support Dirac ART. It had fixed all of the prior bugs and it totally stable. It supports both target curve methods. There are zero improvements in construction of the actual filters since, i.e. no improvements to sound quality.
Target curves can be exported from and imported between every version of the tool to date.
 
@JStewart Yeah, that's what I assumed. They had a blog post around April 2022 where they announced a big change in the way filter design works. It's full of commercial yadda-yadda, but the gist appears to be that now shelf-based target curves reign supreme. It seems quite clear they've moved away from measurements-based criteria, and towards satisfying the taste of a large chunk of their customers, which is dubious to say the least, especially in the audio world. But a company needs to do what it must to turn a profit.

I am going to test the last version that came out before the announcement. I just didn't have time to do it yet. I assume the old target can be exported from it somehow.

And there's also the option of crafting your own target by directly placing points on the graph, so the original linear target could probably be rebuilt regardless of the app version, as long as you know the slope. You would think, in this day and age, a computer could handle that labor, lol.
Assuming the file I sent you works in your version of Dirac, you can open the file with a text editor and change the amplitude values to whatever you want. It takes me a minute or so to do so.
 
From a marketing perspective I’m sure it’s just easier for Dirac to say the target retains your speaker’s character, which they do.

Sure, but the problem with that is that it's a backslide into subjectivity and hardware worship.

I don't care about hardware. I do care, first and foremost, about sound. Can I have the actual sound that was recorded, the way it was envisioned by the sound engineer, please? Dirac used to be the best tool to get closer to that ideal in the speaker world, similar to the oratory1990 presets for headphones.

If "audiophiles" were clear-minded about it, they would think the same. But often the human element takes over, and now we're sipping something old on the rocks while pretending to admire the futuristic sculpture that costs tens of thousands of dollars, just happens to have some speakers embedded in it, and is objectively garbage, far less accurate than the cheap bookshelf speakers + a subwoofer all with Dirac applied to linearize the response.

I still strongly believe the push for shelf-based target curves is by and large the classic tension between doing the right thing, and doing the thing that caters to the crowd's imagination and "creates value for the shareholders", with the latter winning.

JStewart said:
2. Over the years many customers complained about having their bass “neutered”.

This is nothing but Pavlovian conditioning from poorly designed audio systems, which train a large part of the consumers to "like" bloated bass and other deficiencies.

"But they don't sound like my Beats headphones!" Yeah, thank God.

Don't get me wrong, I understand this whole industry is full of perverse incentives. I don't know how honest companies stay alive in this context.

JStewart said:
I also heard Dirac removed the straight line slope target from the most recent version.

I asked them point-blank if this is a bug in the latest 3.11 version. They said it's not a bug, but it's the intended behavior of the app now. :(

JStewart said:
The version I suggested 3.4,4 was the last version before Dirac started adding code to the tool to support Dirac ART. It had fixed all of the prior bugs and it totally stable. It supports both target curve methods.

I will try that, thank you.

All I want is figure out the characteristics (slope, mostly) of the old linear target curve. That would be a good starting point. I can then create my own target, but I'm missing this piece of information as a starting point. Seems like I could simply extract that info from 3.4.4. Then, in any future version, I could create targets that resemble it, with variations where necessary.

Assuming, of course, that the points-based curve is not removed completely by Dirac in the future, lol, because "customers demand shelves" or something.

terryforsythe said:
Assuming the file I sent you works in your version of Dirac, you can open the file with a text editor and change the amplitude values to whatever you want. It takes me a minute or so to do so.

Yeah, I know. I just don't have time right now to run the tests. But I will set time aside soon to get to the bottom of this. Thank you.
 
Back
Top Bottom