Recording Impulse Responses for Speaker Virtualization
Aug 31, 2023 at 2:09 PM Post #1,636 of 1,914
@musicreo can you try cropping this one? i tried multiple times but i'm getting agan this weird warp when the right side stereo is playing

i'm mostly testing this music


all the other official hrir in hesuvi plays good, i just can't get this hrir working without this warping sound



probably the fact that this measurements are more bassy is due to the fact that i'm recording at full volume, prior ones weren't recorded full volume, anyway at the moment i'm more concerning about the warp problem, after that i could try recording more of the same hrir with less bass
 

Attachments

  • DH2 Exciter test.zip
    1.9 MB · Views: 0
Last edited:
Aug 31, 2023 at 2:42 PM Post #1,637 of 1,914
Attached is the cropped file for HeSuVi. I can't test it as I have not installed HeSuVi.
I optimized the cropping script. It is now possible to see the cropping before export. The exported files are found in the folder of your original file and it is possible to export single channels. It would be nice if you could try and tell me if it works for you.

Most of the packages should be installed. Perhaps you have to install tkinter with
python
pip install tk
in the command line.

This is how the script should look like:
Screenshot 2023-08-31 204048.png


I tried to record Dolby Headphone also from my Xonar U1 soundcard for testing but it seems with Win 11 the driver for the Xonar U1 is no working anymore.
 

Attachments

  • DH2 Exciter test_HesuVi.zip
    845 KB · Views: 0
  • croppingrecording.zip
    2.6 KB · Views: 0
Last edited:
Aug 31, 2023 at 4:43 PM Post #1,638 of 1,914
Attached is the cropped file for HeSuVi. I can't test it as I have not installed HeSuVi.
I optimized the cropping script. It is now possible to see the cropping before export. The exported files are found in the folder of your original file and it is possible to export single channels. It would be nice if you could try and tell me if it works for you.

Most of the packages should be installed. Perhaps you have to install tkinter with
python
pip install tk
in the command line.

This is how the script should look like:


I tried to record Dolby Headphone also from my Xonar U1 soundcard for testing but it seems with Win 11 the driver for the Xonar U1 is no working anymore.


Found it! now i know why there's warp in all these measurements... looks like having max volume to get better signal to noise ratio is under some levels,

i plotted two of the same recordings made with less volume from MPC and now the warp's gone completely!

i also used your awesome program to crop these measuremets and works great!


if is possible i would like to cut shorter the tails (it gives me an error when i try to do), i can't go below 500ms for crop time
1693514411154.png
 

Attachments

  • DH2 TEST 2 no warp.zip
    9.2 MB · Views: 0
Last edited:
Aug 31, 2023 at 5:24 PM Post #1,639 of 1,914
i also used your awesome program to crop these measuremets and works great!
Thank you for your testing.
if is possible i would like to cut shorter the tails (it gives me an error when i try to do), i can't go below 500ms for crop time
I found the wrong line in the code and replaced the file in the previous post. It should work now.
 
Sep 1, 2023 at 5:24 AM Post #1,640 of 1,914
ok now i would like to compensate my headphones for this hrir, just to experiment the result

for this matter do i need to do an inverse equalization? i would prefer eq my headphones measuring directly with my binaural mics, which program can i use?

i mean, i do also have the impulcifer headphones plot, is there a way to "convert" that .wav to the actual eq graph?
 
Last edited:
Sep 1, 2023 at 6:05 AM Post #1,641 of 1,914
ok now i would like to compensate my headphones for this hrir, just to experiment the result

for this matter do i need to do an inverse equalization? i would prefer eq my headphones measuring directly with my binaural mics, which program can i use?https://www.roomeqwizard.com/
I'm unsure if it is possible to just output the deconvoluted headphone wav with the actual Impulcifer code?
Alternative would be REW - Room EQ Wizard should work. The problem is that the generic hrirs and your own headphone measurement will probably not work well together as inaccuracies and imbalances in the measurement don't cancel each other out anymore.
I strongly suggest just to use left or right channel measurements of your headphone and apply it to both channels.
i mean, i do also have the impulcifer headphones plot, is there a way to "convert" that .wav to the actual eq graph?
Just from the png plot it is not possible. You would have to digitalize it first and could use REW to generate any compensation you want.
 
Sep 6, 2023 at 6:09 PM Post #1,642 of 1,914
Are there "normal" non-binauralized stereo mixing studio recordings available somewhere in the internet? The recordings I've found out there are already recorded using a dummy head, so it's already introducing the HRTF effect. I'm looking for a clean studio recording, like one recorded with a flat measurement microphone, for example. Currently, I'm trying out SPARTA Binauralizer with my ow SOFA file, but the sound has no ambience, like listening in an anechoic chamber, and I would like to just add the studio ambience, just to have a sense of space :)
 
Last edited:
Sep 6, 2023 at 7:16 PM Post #1,643 of 1,914
Are there "normal" non-binauralized stereo studio recordings available somewhere in the internet?
???
99.999% of all existing stereo recordings are "normal" non-binauralized stereo recordings.

I'm looking for a clean studio recording, like one recorded with a flat measurement microphone, for example.
You want to use a recording made with 2 microphones from audio played back over speakers in a studio? And then binauralize it using your personal HRTF? Theoretically that makes no sense at all.

I don't know SPARTA. Does it not emulate speakers in a room including reflexions? Maybe you should try VIRTUOSO, you can do a free tryal and it also can use a user provided SOFA file.

You have a SOFA file with your own personal HRTF? How was it created?
 
Last edited:
Sep 6, 2023 at 10:50 PM Post #1,644 of 1,914
???
99.999% of all existing stereo recordings are "normal" non-binauralized stereo recordings.
You want to use a recording made with 2 microphones from audio played back over speakers in a studio? And then binauralize it using your personal HRTF? Theoretically that makes no sense at all.
I'm using my own SOFA file with SPARTA Binauralizer. I just want to add a realistic mixing room ambience to it (not an ambience made with an algorithmic reverb, it sounds unrealistic) because SPARTA Binauralizer by itself is only doing the binauralization, it's not adding any room ambience, therefore delivering unusual results like if one is listening to speakers inside an anechoic chamber.
I suppose using BRIR's isn't going to work for what I want because these are recorded using a dummy head, therefore they already have their own HRTF built-in.

I don't know SPARTA. Does it not emulate speakers in a room including reflexions? Maybe you should try VIRTUOSO, you can do a free tryal and it also can use a user provided SOFA file.
SPARTA Binauralizer emulates the sound of speakers but inside an anechoic chamber, like Genelec Aural ID, for example. It does not add any room ambience, unfortunately.

Actually, I already have tried Virtuoso out but it was before I got my SOFA file, so I think I can't test it again because iLok registers my machine ID, so even if I create a new account, it probably won't let me try it again lol
By the way, do you know of any other software that allows you to import a custom SOFA file and that adds room ambience too?

You have a SOFA file with your own personal HRTF? How was it created?
Yes! Using Mesh2HRTF :)
 
Last edited:
Sep 7, 2023 at 5:45 AM Post #1,646 of 1,914
I'm looking for a clean studio recording, like one recorded with a flat measurement microphone, for example. Currently, I'm trying out SPARTA Binauralizer with my ow SOFA file, but the sound has no ambience, like listening in an anechoic chamber, and I would like to just add the studio ambience, just to have a sense of space
Like @sander99, I’m not clear about what you’re after. Recording studios record with recording studio mics, not flat measurement mics and finished (distributable) mixes/masters will contain considerable reverb. They obviously won’t contain the reverb of the control room/listening environment as that will be added by the consumers listening room.

There’s very possibly no stereo recordings with measurement mics of control/mix room playback and even if there were, it’s extremely unlikely they would be publicly available. So your chances of finding what I’m guessing you’re looking for are very remote.
I just want to add a realistic mixing room ambience to it (not an ambience made with an algorithmic reverb, it sounds unrealistic) because SPARTA Binauralizer by itself is only doing the binauralization, it's not adding any room ambience
Most commercial recordings already have algorithmic reverb added and those that don’t, that only use convolution or actual recording venue reverb, are also “unrealistic”. How “real” an algorithmic reverb sounds to you, is largely a function of how you adjust its parameters. So maybe you need an algorithmic reverb with more user adjustable parameters and/or to spend more time learning how to adjust them?

G
 
Last edited:
Sep 7, 2023 at 2:06 PM Post #1,647 of 1,914
I think he was asking for impulses from studios to add room reverb to his existing HRTF model from mesh2hrtf that lacks in room ambiance to his taste.
 
Sep 9, 2023 at 4:56 AM Post #1,648 of 1,914
Hi. Does anyone know about this warning?
Creating impulse response estimator...
Running room correction...
Running headphone compensation...
Creating headphone equalization...
Creating frequency response target...
Opening binaural measurements...
Plotting BRIR graphs before processing...
Cropping impulse responses...
Equalizing...
Adjusting decay time...
Traceback (most recent call last):
File "impulcifer.py", line 556, in <module>
main(**create_cli())
File "impulcifer.py", line 144, in main
ir.adjust_decay(decay[speaker])
File "C:\Windows\System32\Impulcifer\impulse_response.py", line 330, in adjust_decay
if target_slope > rt_slope:
TypeError: '>' not supported between instances of 'float' and 'NoneType'

I tried to work with Decay parameters but it didn't work on some files.

What I was trying to do was to lower the volume and erase the information on the opposite side of both ears in FL-L, FL-R/FR-L, FR-R. (In audacity)
The reason for doing it exactly is to imitate Crosstalk-Cancellation.
And actually it worked. But it doesn't work if you apply Decay parameters to the file.

1.png


2.png


Something like this.
I think it's also a problem with my recording file because some users said it translates well in this way. But I'm trying this and that with this file I have because I won't be able to record again for the next few months.
Also,
It worked well even if I just converted the original file of a small volume, and it worked well even when I amplified it randomly.
So is there a condition for Decay parameters that I don't know?

<What works>
1. For Crosstalk-Cancel, Decrease Volume FL-R / FR-L each channel in audacity. And Run Impulcifer - It's work.
2. Normal FL,FR file with Decay Parameter only in Impulcifer - it's work.

<Doesn't work for me>
Decrease Volume FL-R / FR-L + Decay Parameter - it's not work.

Can you check the audio file in my compressed file if anyone can help me?
 

Attachments

  • Helpme.zip
    5.7 MB · Views: 0
Last edited:
Sep 9, 2023 at 8:18 AM Post #1,649 of 1,914
What I was trying to do was to lower the volume and erase the information on the opposite side of both ears in FL-L, FL-R/FR-L, FR-R. (In audacity)
The reason for doing it exactly is to imitate Crosstalk-Cancellation.
And actually it worked. But it doesn't work if you apply Decay parameters to the file.

I don't see how this is related to crosstalk? Can you explain in more detail what you want to achieve with this procedure? For me it makes no sense at all.
 
Last edited:
Sep 9, 2023 at 9:49 AM Post #1,650 of 1,914
Hi. Does anyone know about this warning?
Creating impulse response estimator...
Running room correction...
Running headphone compensation...
Creating headphone equalization...
Creating frequency response target...
Opening binaural measurements...
Plotting BRIR graphs before processing...
Cropping impulse responses...
Equalizing...
Adjusting decay time...
Traceback (most recent call last):
File "impulcifer.py", line 556, in <module>
main(**create_cli())
File "impulcifer.py", line 144, in main
ir.adjust_decay(decay[speaker])
File "C:\Windows\System32\Impulcifer\impulse_response.py", line 330, in adjust_decay
if target_slope > rt_slope:
TypeError: '>' not supported between instances of 'float' and 'NoneType'

I tried to work with Decay parameters but it didn't work on some files.

What I was trying to do was to lower the volume and erase the information on the opposite side of both ears in FL-L, FL-R/FR-L, FR-R. (In audacity)
The reason for doing it exactly is to imitate Crosstalk-Cancellation.
And actually it worked. But it doesn't work if you apply Decay parameters to the file.





Something like this.
I think it's also a problem with my recording file because some users said it translates well in this way. But I'm trying this and that with this file I have because I won't be able to record again for the next few months.
Also,
It worked well even if I just converted the original file of a small volume, and it worked well even when I amplified it randomly.
So is there a condition for Decay parameters that I don't know?

<What works>
1. For Crosstalk-Cancel, Decrease Volume FL-R / FR-L each channel in audacity. And Run Impulcifer - It's work.
2. Normal FL,FR file with Decay Parameter only in Impulcifer - it's work.

<Doesn't work for me>
Decrease Volume FL-R / FR-L + Decay Parameter - it's not work.

Can you check the audio file in my compressed file if anyone can help me?
I'll check out the audio file later today, no time right now, but I'm curious about something. What happens when you run Impulcifer and use the Decay parameter as you wish, and then perform the volume change in Audacity to the final hesuvi.wav that Impulcifer generates?

I'm not entirely sure what the end-goal of your process is, but it seems to me that you'd get the same result by editing the completed file rather than editing each individual measurements before processing with Impulcifer. But if you're going to try to reduce the natural crossfeed from the measurements, then that would defeat the purpose of using Impulcifer in the first place.

If understand your process correctly, the final result would mean you would only be able to hear sounds with the ear closest to the speaker. So if you were to wipe out FL-R because the right ear is on the opposite of your left speaker, you'd only hear the front left speaker in your headphones' left driver. I don't see any advantage to using Impulcifer in this case outside of trying to use it entirely for EQ purposes.

Here's the default track order for the hesuvi.wav file, FYI:

Track 1: FL-left (LL) *
Track 2: FL-right (LR) *
Track 3: SL-left
Track 4: SL-right
Track 5: BL-left
Track 6: BL-right
Track 7: FC-left
Track 8: FR-right (RR) *
Track 9: FR-left (RL) *
Track 10: SR-right
Track 11: SR-left
Track 12: BR-right
Track 13: BR-left
Track 14: FC-right
 
Last edited:

Users who are viewing this thread

Back
Top