Measurement/Correction & TTD Windows

vego99

Member
Thread Starter
Joined
Jul 8, 2018
Messages
31
Hello Everyone,

Ive been working with AL for over 1 year now and have made some great strides with learning the systems etc.

One section I’m still having trouble with under standing is the Measurement/Correction Window and the TTD Subwindows.

Looking at the impulse from my filtered measurement, it seems like nothing is happening after 500ms ( 400ms based on the impulse starting at 100ms.)

Does this mean I should be lowering the default time window from 8ms down to 4.5ms @10hz?
What about the upper region?

In Contrast, how should the TTD window get adjusted?

Thanks,

Erik
 

Attachments

  • Screenshot 2024-02-05 at 2.14.05 PM.png
    Screenshot 2024-02-05 at 2.14.05 PM.png
    273.8 KB · Views: 50
  • Screenshot 2024-02-05 at 2.14.29 PM.png
    Screenshot 2024-02-05 at 2.14.29 PM.png
    284.1 KB · Views: 52

juicehifi

Audiolense
Staff member
Joined
Feb 5, 2018
Messages
734
Not much will happen in the filtered response after 4-500 ms unless you extend the measurement and correction window in the correction procedure. But 4-500 ms is usually enoug for quite a few bounces back and forth in the room. And if the first reflections are covered and corrected, the later decay will also have been significantly silenced.

I recommend two ways to examine the effect of the TTD correction.

The first and perhaps the most important one is to look at the step response.

The second is to export an unfiltered simulation and examine it in REW. Then you can look at waterfall plots and other 3D plots and compare with the measurement ... which you can also export.

The size of the TTD window do not only limit the scope of the correction. It also has implications on how well Audiolense sees the time domain issues. So there are instances where e.gk. a 6-4 window works better than the default 5-3 window, especially if the selective preringing prevention is checked. When it works better it is not because it corrects better, but because it becomes better informed.

My approach to tuning the TTD window is to aim for the cleanest possible step response, with specific regards to any occurances of potentially audible pre-ringing. The trial and error involves playing with different TTD window sizes and with / without the selective preringing prevention engaged.
 

vego99

Member
Thread Starter
Joined
Jul 8, 2018
Messages
31
Bernt,

This is fantastic information, thank you very much!
 

vego99

Member
Thread Starter
Joined
Jul 8, 2018
Messages
31
With your guidance and after several hours of generating filters, I think we have a winner. Sounds pretty fantastic.
Screenshot 2024-02-07 at 8.54.01 PM.png
Screenshot 2024-02-07 at 8.52.32 PM.png
Screenshot 2024-02-07 at 8.52.45 PM.png
Screenshot 2024-02-07 at 8.53.29 PM.png
Screenshot 2024-02-07 at 8.54.16 PM.png
Screenshot 2024-02-07 at 8.57.38 PM.png
 

juicehifi

Audiolense
Staff member
Joined
Feb 5, 2018
Messages
734
Looking good.

It looks like you're using a linear phase target here. It should work perfectly well on most music, perhaps inluding everything you care to listen to. And it may be the best solution for acoustic music. But a linear phase target will have some low frequency pre-ringing that may be audible on music with e.g. synthetic bass and rhythm. If that ever becomes a problem you could switch to a minimum phase target.
 

moedra

Member
Joined
Mar 1, 2021
Messages
75
Not much will happen in the filtered response after 4-500 ms unless you extend the measurement and correction window in the correction procedure. But 4-500 ms is usually enoug for quite a few bounces back and forth in the room. And if the first reflections are covered and corrected, the later decay will also have been significantly silenced.

I recommend two ways to examine the effect of the TTD correction.

The first and perhaps the most important one is to look at the step response.

The second is to export an unfiltered simulation and examine it in REW. Then you can look at waterfall plots and other 3D plots and compare with the measurement ... which you can also export.

The size of the TTD window do not only limit the scope of the correction. It also has implications on how well Audiolense sees the time domain issues. So there are instances where e.gk. a 6-4 window works better than the default 5-3 window, especially if the selective preringing prevention is checked. When it works better it is not because it corrects better, but because it becomes better informed.

My approach to tuning the TTD window is to aim for the cleanest possible step response, with specific regards to any occurances of potentially audible pre-ringing. The trial and error involves playing with different TTD window sizes and with / without the selective preringing prevention engaged.
What is the proper way to export an unfiltered simulation? AL won't let me export anything unless it is "filtered" and I generate correction files. I can export an unsmoothed simulation, however. This by toggling through the smoothing options next to the Simulated Result in the legend/key to the right side of the interface. Is this what is meant by "unfiltered"?
 

vego99

Member
Thread Starter
Joined
Jul 8, 2018
Messages
31
What is the proper way to export an unfiltered simulation? AL won't let me export anything unless it is "filtered" and I generate correction files. I can export an unsmoothed simulation, however. This by toggling through the smoothing options next to the Simulated Result in the legend/key to the right side of the interface. Is this what is meant by "unfiltered"?
I had some similar confusion when trying to do this.

First thing you need to change the smoothing on the right side to unsmoothed. Refilter the measurement, then generate your filter.

Once the filter is created, up top click on Anaylisys > Simulation. At the bottom click on export simulation.

From there save the files then you want to open REW, then DRAG the files in to the REW screen.

From there you can smooth the simulations as needed.

Take note, it seems like REW will only accept 44.1 and 48k sample rate simulations. Because of this the frequencies will be off slightly by 10hz ( youll see what i mean when you open)

Hope this helps.

-e
Screenshot 2024-02-16 at 4.45.40 PM.png
Screenshot 2024-02-16 at 4.45.57 PM.png
Screenshot 2024-02-16 at 4.46.18 PM.png
Screenshot 2024-02-16 at 4.49.19 PM.png
 

moedra

Member
Joined
Mar 1, 2021
Messages
75
I'm assuming that if I record my measurements at 48kHz, they will not exhibit the offset? I see in your images what you're talking about. Looks like more than 10Hz, though!
 
Top Bottom