[Pictures Inside] Dirac Correction 0-300Hz Vs Full Range Correction

[Andrej]

Any target curve is only valid for one volume level, thanks to our hearing acuity at different frequencies. I also feel (no concrete evidence)
that the flat curve is the desired target, which is then modified based on Fletcher-Munson curves and the target listening volume. (Who are we to tell recording engineers that they screwed it up by changing the frequency response of their creations?)
Since I often listen at wide range of volumes, dynamic equalization (Audyssey, Dolby,...) is my solution, with the flat target curve.
And since everybody's hearing is a little different, finding the "perfect" target curve for any one person is non trivial. Room treatments, and how directional the speakers are (monopole, dipole, etc...) can also influence what the preferred target curve should be.

 

[abdo123]

If you have the inclination, we are very interested in your thoughts and comments about a target curve custom designed according to the measurements of each specific room (it can be eventually modified with sliders)
Such a target curve can be tested using a smartphone or tablet and the new mobile apps that we are currently testing:

Downloads - Dirac

www.dirac.com

Thanks for your time
Flavio

I thought this was the multi-speaker whole room optimisation that you guys have been teasing for a while but alas.

 

[Johnp]

I've been experimenting with full range correction that has a flat target at the listening position up to 400Hz and correction for on-axis flat speaker output above 400Hz. I've been running it for a few weeks, and it works well in my system. I end up with different targets above 400Hz for each speaker group, but the results are very good and balanced to my ears. This is the LR target that results in flat speaker output when measured at about 30 cm:

 

[Dj7675]

I've been experimenting with full range correction that has a flat target at the listening position up to 400Hz and correction for on-axis flat speaker output above 400Hz. I've been running it for a few weeks, and it works well in my system. I end up with different targets above 400Hz for each speaker group, but the results are very good and balanced to my ears. This is the LR target that results in flat speaker output when measured at about 30 cm:

View attachment 198961

So are you trying to do a type of speaker eq with Dirac then, instead of correcting based on the listening position?

 

[Johnp]

So are you trying to do a type of speaker eq with Dirac then, instead of correcting based on the listening position?

Yes. I attempted this by using a listening position target curve that creates a flat speaker response above 400 at the speaker. Here's how I generated the target curve:

1. Do a normal Dirac calibration, 17-point in my case.
2. Load a flat target curve, 0 at 20Hz to 0 at 20kHz.
3. With the flat target loaded, measure the response at the speaker (about 30 cm) at 75dB in the midrange using horizontal mic position and 0 degree calibration file. Since it's targeting flat at the listening position, it is boosting the high end to offset the in-room drop and natural speaker rolloff.
4. Export the near-field speaker measurement from REW with one-third smoothing.
5. Invert the measurement around the center point by subtracting all measurements from 75. (Boosts become cuts and cuts become boosts.)
6. Create the target in Dirac using a text editor, flat from 20 to 400 and the inverted measurement above 400.
7. Load into Dirac. When measured in the near-field, the response is perfectly flat at the speaker. I really like how it sounds at the listening position.

I've gone a couple of weeks without feeling the need to tweak the curve, which is amazing for me. I did create versions with Harman +2dB and +4dB in the low end in case I want a little more bass.

I was concerned at the start that this might be problematic for the speakers or the correction, but when I see the target, it's a very reasonable target. I did this for all speaker groups, so the target is different above 400 for all groups. It's a bit of a hassle, and I'm not sure it would be of benefit with other rooms/speakers, but it's been a success for me.

 

[Dj7675]

Yes. I attempted this by using a listening position target curve that creates a flat speaker response above 400 at the speaker. Here's how I generated the target curve:

1. Do a normal Dirac calibration, 17-point in my case.
2. Load a flat target curve, 0 at 20Hz to 0 at 20kHz.
3. With the flat target loaded, measure the response at the speaker (about 30 cm) at 75dB in the midrange using horizontal mic position and 0 degree calibration file. Since it's targeting flat at the listening position, it is boosting the high end to offset the in-room drop and natural speaker rolloff.
4. Export the near-field speaker measurement from REW with one-third smoothing.
5. Invert the measurement around the center point by subtracting all measurements from 75. (Boosts become cuts and cuts become boosts.)
6. Create the target in Dirac using a text editor, flat from 20 to 400 and the inverted measurement above 400.
7. Load into Dirac. When measured in the near-field, the response is perfectly flat at the speaker. I really like how it sounds at the listening position.

I've gone a couple of weeks without feeling the need to tweak the curve, which is amazing for me. I did create versions with Harman +2dB and +4dB in the low end in case I want a little more bass.

I was concerned at the start that this might be problematic for the speakers or the correction, but when I see the target, it's a very reasonable target. I did this for all speaker groups, so the target is different above 400 for all groups. It's a bit of a hassle, and I'm not sure it would be of benefit with other rooms/speakers, but it's been a success for me.

I have seen a similar process used using Audyssey doing speaker correction. I’m guessing you don’t have anechoic measurements for your speakers, otherwise you could just use those.
I do something sort of similiar.. I correct under 500hz with Dirac/DLBC. Then my process is..
-Look at anechoic measurements (on axis and PIR) for areas that may need to be addressed-
-Take MMM measurements to see what in room measurements show
-Create PEQ filters and apply post Dirac above 500 hz. Test filters while listening turning them on/off to confirm if they improve anything. If I can’t tell the difference or it sounds worse, I discard them. Filters are made primarily by eyeballing on axis, PIR, and in room measurements. Corrections are the broader variety.
I am still experimenting with full range DIRAC correction. It is starting to get close to my manual correction but I still prefer the manual correction above 500hz so far.
I voted 0-300hz even though I am primarily doing under 500hz.

 

[tjcinnamon]

The right answer is the room transition frequency. However, the sound I prefer is the full range but following the speaker roll off. Depending on the music I like full range flat… which sounds very artificial but it’s pretty chaotic music and artificial already.

So perhaps that’s a long way of saying: “tweak to taste”

 

[Flak]

The new desktop version has just been released...

 

[tjcinnamon]

The new desktop version has just been released...
View attachment 199204

Fantastic! Do you know if it's roadmapped to allow us to see the resultant trims and delays before sending it to the AVR? Additionally, possibly modify the trims before sending it to the AVR? My AVR doesn't allow modding trims after the fact and 95% of my calibrations kills the center image. They always skew left (despite my tape measure) and playing with Dirac off, it being centered.