Recording Impulse Responses for Speaker Virtualization

[Alamar5]

I created it myself. We can make it directly using tools like REW or Audacity.
While it’s possible to use actual BRIRs or HRIRs, the current discussion is focused on input/output matching. And Hesuvi means just
Therefore, I applied Dirac impulses, as well as positive and negative filters, to each channel for intuitive visualization.

Can you describe your loopback measurement method more detailed? I see your graphs made in REW (i.e. the result), but your method desciption is quite short. Isn't it a scientific approach when the result must be reproducible by another person?

 

[Lion3]

It was my mistake. I’m sorry

 

[musicreo]

Is it not enough to know that the channel sequence in the copy command must match the channel sequence in the convolution file? So 1->1, 2->2, 3->3...

In my case:
copy: L L R R C C....
wav file: FL-l, Fl-r, FR-l FR-r, C-l, C-r....

L->FL-l, L-> Fl-r, R-Fl-r...

Hence, this two codes have the same output :

 

[Alamar5]

Lion3, I checked your measurement hesuvi.wav in Audacity. First, seems like you are using different Dirac impulses for the left and right ears. Second, for some reason (or you just forgot) you did not swapped left-right ears IRs for the channels 8-9. After I swapped these channels, as must be for proper HeSuVi wav, I've got both flat measurements graphs for the Left and Right speakers.

 

[Alamar5]

Is it not enough to know that the channel sequence in the copy command must match the channel sequence in the convolution file? So 1->1, 2->2, 3->3...

Seem like it's quite enough and how it actually works

 

[Alamar5]

Lion3, additionally, I checked measurement hesuvi.wav with surround voice audio test from Fraunhofer IIS
https://www2.iis.fraunhofer.de/AAC/multichannel.html
https://www2.iis.fraunhofer.de/AAC/7.1auditionOutLeader v2.wav

I highly recommend to make the very first test of any HRIRs with this audio test for evaluation of proper speakers positioning and other acoustic things. Improper made or of mediocre quality HRIR will be cleary distinguished in headphones.

So, the uploaded hesuvi.wav used in measurement, have a flaw that can be spotted even in the listening test - the right speaker is mostly shifted to the left speaker. But after swapping channels 8-9, it will be on its proper position on the right side.
That's why your measurement has such strange result.

 

[Lion3]

Hello, I would like to start with an apology.

Last night, during the loopback process, I mistakenly reversed the concepts of in/out, which led to incorrect measurements. As a result, the default Hesuvi appeared distorted, and this was entirely my error. (It seems there were more logical inconsistencies as I tried to reason based on the results of this mistake and fit them into my understanding.)

This morning, I intended to post about the revised results, but I returned to the original loopback process as it should have been done. Even then, I could not understand the discrepancies in today's results.

That is why I wrote the earlier post, and now I am writing this. Although this misunderstanding began with my own confusion, I will not make excuses for it.

The fault is entirely mine.
I sincerely apologize to everyone involved in this thread. I am truly sorry.

Additionally, if the information were correct, it would be appropriate to leave the post as it is. However, since the premise itself was wrong, I will edit my recent post to replace it with, "It was my mistake. I’m sorry."

Once again, I apologize and hope that this clears up any misunderstandings.

 

[Alamar5]

Lion3, as a result of this discussion, we have found about EqAPO undocumented Copy command behavior. It has not been in vain

 

[morgin]

Lion3 we all make mistakes and learn from them. But keep testing and don’t be discouraged to post new findings. We need discussion and testing.

But saying this I still won’t stop telling people Dolby access needs to be tested more with a Dolby atmos device because it’s so immersive. We don’t need the hack I posted earlier for driver. Just Dolby access, voicemeter, cru hack for 8ch to the Dolby device, hesuvi and eqapo.

 

[castleofargh]

@Lion3
Looking for your own possible mistake, finding something, and letting others know, that's the coolest behavior possible. You have no idea how many people cannot or will not do that. I appreciate you maybe even more for this than for all the cool stuff you've shared so far in the thread.

shiiit happens.

 

[Iohfcasa]

Has anyone tried this?
https://pub.dega-akustik.de/DAGA_2024/files/upload/paper/428.pdf
https://www.researchgate.net/public..._Transfer_functions_at_the_Ear_canal_entrance
https://github.com/AudioGroupCologne/pirate-extensions

 

[morgin]

Introducing Eclipsa Audio: immersive audio for everyone​

https://aomediacodec.github.io/iamf-tools/web_demo/

Not tried it yet but thought you guys might like to test it out.

 

[morgin]

Hey guys been a while hope you’re all doing great. Wanted to ask if you think these are good measurements. Also I gotta do it again but a big Thankyou to Jakko and for everyone (big shout out to musicreo) who helped me learn about this awesome software.