Recording Impulse Responses for Speaker Virtualization

[reter]

guys does the speaker position adjustment in Hesuvi affect the quality of the recorded hrir when used? i plan to give some gap between the two front speakers but i'm not sure if this adjustment in hesuvi can and will worsen the quality

 

[Spyartinc]

https://wavearts.com/products/plugins/mlstool
is this basically GUI alternative to Impulcifer?

 

[musicreo]

https://wavearts.com/products/plugins/mlstool
is this basically GUI alternative to Impulcifer?

You do not have the post processing options impulcifer provides (channel balance corrections, headphone compensation). It is limited to stereo. Impulcifer can be much more customized.

 

[Brandon7s]

There have been no experience reports for a while now? Is the measurement and post processing process to obtain good results to difficult and nobody has done any measurements in the last months?

I've been so satisfied with my favorite BRIRs with both my HD 8XX and my Anandas that I haven't bothered taking any new captures in over a year. I still use Impulcifer every single day with both my work computer and my personal computer, too. I have speakers but rarely use them because I live in an apartment and can't get my preferred listening volume without (potentially) being annoying to neighbors. I prefer my BRIRs over my speakers anyway due to the fact that I don't have subs, so the BRIRs have a much more satisfying low end.

There's two things that could change that would make me want to take some more captures with my current headphones:
1. if a new, higher quality pair of in-ear mics were released to the public at a not-ridiculous price.
2. if I move to a house and set up a sound system with sub woofers. I'm curious how much better bass realism would be with subs vs. no subs.

 

[morgin]

1. if a new, higher quality pair of in-ear mics were released to the public at a not-ridiculous price.

Same here. Mine sound so good that I can’t really imagine it any better. And would only try if there were better quality mics or maybe even a properly treated room.

One problem with the new windows 11update is volume2 doesn’t seem to work anymore. What do you guys use to change the volume on a pc?

 

[Brandon7s]

One problem with the new windows 11update is volume2 doesn’t seem to work anymore. What do you guys use to change the volume on a pc?

I just use media keys on my keyboard or my headphone amp's volume knob. Simple, but it works!

 

[morgin]

I just use media keys on my keyboard or my headphone amp's volume knob. Simple, but it works!

I use it with my tv, pc and a long cable for my headphones. My amp is too far to walk. (Not that far but lazy to walk over to change volume). Volume2 worked well with my keyboard volume buttons in the past but not anymore.

 

[Sanman99]

does anyone have any sample audio from smyth realiser or Hesuvi and Impulcifer? I had some trouble installing and barely notice much difference.
Got VB-Audio Cable installed. Hesuvi also. But Impulcifer is having some errors and I can't get the GUI either.



installation:

TypeError: CCompiler_spawn() got an unexpected keyword argument 'env'

error: metadata-generation-failed





(venv) C:\Windows\System32\Impulcifer>"C:\Users\x\AppData\Local\Programs\Python\Python311\python.exe" impulcifer.py --test_signal=data/sweep-6.15s-48000Hz-32bit-2.93Hz-24000Hz.pkl --dir_path=data/demo

Traceback (most recent call last):

File "C:\Windows\System32\Impulcifer\impulcifer.py", line 6, in <module>

from tabulate import tabulate

ModuleNotFoundError: No module named 'tabulate'

and trying to click the GUI

C:\Windows\System32>"C:\Windows\System32\Impulcifer\gui.py"

Traceback (most recent call last):

File "C:\Windows\System32\Impulcifer\gui.py", line 7, in <module>

import recorder, impulcifer

File "C:\Windows\System32\Impulcifer\recorder.py", line 5, in <module>

import sounddevice as sd

ModuleNotFoundError: No module named 'sounddevice'

Viper4Windows and PeaceEQ sound ok. I haven't managed to figure out much besides graphic/parametric EQ, and AutoEQ, and then whatever with all the graphs or to use microphone to calibrate somehow or just import AutoEQ.
I see the menu to toggle PeaceEQ and Hesuvi in EqualizerAPO.

Does anyone know of some kind of slidebars like a mixer to adjust EQ? or just get a mixer and hook-up a VST?
i have another couple ideas to maybe post a separate thread.

But I'll consider some surround effects or maybe pre-production physical or post-production.

 

[reter]

I've been so satisfied with my favorite BRIRs with both my HD 8XX and my Anandas that I haven't bothered taking any new captures in over a year. I still use Impulcifer every single day with both my work computer and my personal computer, too. I have speakers but rarely use them because I live in an apartment and can't get my preferred listening volume without (potentially) being annoying to neighbors. I prefer my BRIRs over my speakers anyway due to the fact that I don't have subs, so the BRIRs have a much more satisfying low end.

There's two things that could change that would make me want to take some more captures with my current headphones:
1. if a new, higher quality pair of in-ear mics were released to the public at a not-ridiculous price.
2. if I move to a house and set up a sound system with sub woofers. I'm curious how much better bass realism would be with subs vs. no subs.

i only wish i could record my own hrir in a room treated like a cinema room, i'd like that cinema reverb feeling i miss so much.

even now i feel that a bad reverb could hide subtle details and i don't want that

also i'd like to upgrade my 660s to the hd800s but man...i don't know how much of an upgrade could be, i'm scared could not be much of a difference for the price.

 

[castleofargh]

also i'd like to upgrade my 660s to the hd800s but man...i don't know how much of an upgrade could be, i'm scared could not be much of a difference for the price.

The feedback in the Realiser thread is that the HD800 is perceived as giving a bigger stage (one with the right distances), while the 6 series in general seems to feel smaller than where the speakers really were. It's one of those things where I might need to have both side by side to be able to tell because of how strongly my brain tends to anchor the sound to the speakers I see in front of me.
Personally, I tried a bunch of placements and measurements, but as I mostly use my headphone in front of my computer, I ended up favoring my near field setup with my speakers about 1.3m away from me on each side of the screens. Probably the worst sound in the worst room for spatial virtualization, but I enjoy it and use it all the time instead of the bigger and more reverberating impulses I recorded. I suspect that having the speakers so obviously in my field of view all the time is a big factor for me. Partly because of that(and because I'm cheap), I never bought a HD800. But truthfully, I haven't seen anybody owning a hd800 and deciding to use a 5 or 6 series for multichannel on headphone. So there is that.

I'm not helping, I know ^_^.

About big rooms and reverb, maybe you'd enjoy a "shaker" more than (or maybe along with) big room reverb? I like me some rumble in the subs, but I also noticed that I'm not a fan of big room reverb in my headphones. Maybe because I don't have a hd800, but I'm guessing it has more to do with my brain going, "you tell me there is that much reverb, but I know and can see I'm standing in that small room? Get out of here!".

 

[reter]

The feedback in the Realiser thread is that the HD800 is perceived as giving a bigger stage (one with the right distances), while the 6 series in general seems to feel smaller than where the speakers really were. It's one of those things where I might need to have both side by side to be able to tell because of how strongly my brain tends to anchor the sound to the speakers I see in front of me.
Personally, I tried a bunch of placements and measurements, but as I mostly use my headphone in front of my computer, I ended up favoring my near field setup with my speakers about 1.3m away from me on each side of the screens. Probably the worst sound in the worst room for spatial virtualization, but I enjoy it and use it all the time instead of the bigger and more reverberating impulses I recorded. I suspect that having the speakers so obviously in my field of view all the time is a big factor for me. Partly because of that(and because I'm cheap), I never bought a HD800. But truthfully, I haven't seen anybody owning a hd800 and deciding to use a 5 or 6 series for multichannel on headphone. So there is that.

I'm not helping, I know ^_^.

About big rooms and reverb, maybe you'd enjoy a "shaker" more than (or maybe along with) big room reverb? I like me some rumble in the subs, but I also noticed that I'm not a fan of big room reverb in my headphones. Maybe because I don't have a hd800, but I'm guessing it has more to do with my brain going, "you tell me there is that much reverb, but I know and can see I'm standing in that small room? Get out of here!".

nope, for me i can hear that bad reverb mostly on vocals which are the most annoying to hear, music is ok...i can't complain on that, but man vocals are horrible, it's like all the people are speaking in a tunnel or something like that, i don't like that

what i'd like to hear is a cinema reverb which doesn't ruin vocals, not so much

 

[musicreo]

I think a lot of reverb people hear is not a direct result of the room but has its origin in a mediocre measurement.

 

[SonOfMorwen]

Hi everyone,

First, I'm not a native english speaker, so sorry if some things are confusing. I will try my best to be clear but if I'm not ask me and I will rephrase what I want to say to make it more understandable.

Thank you Jaakko for Impulcifer, it seems really interesting and much cheaper and practical than the Smyth Realiser. I think I will give it a try in the coming times, but for now I never tried any surround sound virtualization.

But first, I have some theoretical questions to be sure that I understand things well, I hope someone will be able to answer to them.

When listening to a speaker, are we able to know the distance from where it sits just by ear, without seeing it ? For example, if I have two speakers, we suppose them to be acoustically transparent, delivering the same sound level at my ears, one behind the other and both oriented toward myself, can I differentiate them ? How does it works ? Is it possible that the feeling of the distance of a speaker is the result of the reverberation or the result of the ratio of the direct versus reverberated sound ? If this is the case, could we just adjust the decay time in-order to bring closer or move away the virtual speakers, keeping the angles identicals ? It would be useful to adapt to differents screen sizes but with the same viewing angles and so different viewing distances.

For what I understand, the convolution will make the frequency response of the headphone looks like the same as the frequency response of a speaker. Does it mean that if I do my speaker measurement with X dB recorded at the ears, I will hear X dB with the headphones after the convolution ? Or the speaker measurements and the resulting convolutions are normalized separately ?
I have another related question. Does the HRTF change with the sound level ? For example, if I increase the sound of my headphones playing through surround virtualization, will it be equivalent to increasing the sound of the speakers when listening to them ? Or the HRTF of me listening to the speakers will not be the same as before when increasing the sound level ?

Thank you everyone for this thread, it's very helpful and instructive.

 

[castleofargh]

When listening to a speaker, are we able to know the distance from where it sits just by ear, without seeing it ? For example, if I have two speakers, we suppose them to be acoustically transparent, delivering the same sound level at my ears, one behind the other and both oriented toward myself, can I differentiate them ? How does it works ? Is it possible that the feeling of the distance of a speaker is the result of the reverberation or the result of the ratio of the direct versus reverberated sound ? If this is the case, could we just adjust the decay time in-order to bring closer or move away the virtual speakers, keeping the angles identicals ? It would be useful to adapt to differents screen sizes but with the same viewing angles and so different viewing distances.

We can differentiate distances, but not precisely at all because we rely on cues like loudness(can change without the distance changing), FR(frequency response, pas France^_^) with how high frequencies get attenuated faster in the air than low freqs(but it requires big distance and knowledge of the source's FR), and indeed, delays between direct sound and reverb from a wall(requires knowing where the wall is relatively to the speakers, and most likely to have moved around in that room for the brain to get used to correlating audio and visual cues).
So nothing is a solid and definitive way to determine distance by ear, and because the brain loves replacing complicated questions with easy ones without telling us, if you see something telling you about the size of the room while listening to the sound of another room, it's perfectly possible for you to end up feeling a distance that agrees with the room you're in. At the same time, if you're immersed in a movie, maybe you'll feel distances beyond your room or the recorded distance in the PRIR. All in all, distance isn't going to be super accurate unless you can associate the sound with a well known experience/reference. I often talk about how having physical speakers in my field of vision makes me anchor the sounds to them, even though they're turned off, even if the angle is a little off, or the original distance is quite different. I'm listening with the headphone, but over time my brain decides that sound comes from the sound making boxes I can see with my own eyes.
Not everybody reacts that way. As the cues are never perfectly clear and some end up contradicting others, it becomes hard to tell how you in particular will react, what cues your brain will prioritize/trust. A lot of it is based on your own life experience, and what you can manage to imagine while knowing you have a headphone on your head. Some people describe experiences with traditional headphone sound, giving them a space and distances I don't even get when I'm using the Realiser. And given how mad some get when I questioned their description, I'm inclined to think some really feel a vast soundstage from basic headphone stereo nonsense. They just have that much of a power of imagination.

When 100% of the cues agree, we fall back to traditional psychoacoustic, and we then know very well how the average human will react. For example, sound engineers make convincing "room sound" for movies where the original audio wasn't recorded at the scene. So there are tools and knowledge related to your idea, for sure. But they traditionally go with speaker playback, and a movie to watch, telling us why the sound is how it is.
For actual expert information on messing with the sound professionally to create an ambiance and some distance, you can go harass our local @gregorio .

For what I understand, the convolution will make the frequency response of the headphone looks like the same as the frequency response of a speaker.

Almost. There is still the possible FR impact of the mic moving between measurements, and the FR impact of it recording the sound instead of your eardrum maybe 2cm deeper in the ear canal. But yes, the principle it to match (at the microphone!!) the FR of the speakers and headphone, so you should get fairly close to that(still need something for the ear canal, so you perceive it as the same sound FR you get when listening to the speakers).

Does it mean that if I do my speaker measurement with X dB recorded at the ears, I will hear X dB with the headphones after the convolution ? Or the speaker measurements and the resulting convolutions are normalized separately ?

Loudness is handled by your amplifier. The app and you will have to care more about not clipping the signal (avoid getting the digital amplitude above 0dB due to EQ).

Does the HRTF change with the sound level ? For example, if I increase the sound of my headphones playing through surround virtualization, will it be equivalent to increasing the sound of the speakers when listening to them ? Or the HRTF of me listening to the speakers will not be the same as before when increasing the sound level ?

Yes and no. What can noticeably change when playing music louder is the amount of distortions coming out of the speakers and headphone. In general, speakers have a lot more distortions than headphones. It gets massive when you try to push them really loud (even more so if they weren't designed to handle that level. For example, my active speakers are rated for about 90dB, which might not seem like much for the headphone world). So recording the speakers at a loud but reasonable level and then playing that through your headphone with the right convolution at a really loud level, is likely to sound different from the actual speakers at that level. Because you won't simulate the extra distortions from the speakers, stuff in the room that might start shaking. Instead, you're getting the possibly audible distortions of the headphone itself, which have a very different distortion content (usually just some bass distortions, until you also reach the limit for that particular headphone)
So strictly speaking, the sound could be audibly different. In practice, though, it often means that you're getting a better sound quality because headphones tend to go loud with less distortion than speakers.
Otherwise, speaking strictly about HRTF, the content of an impulse is about FR and time delays, if making your speakers louder doesn't affect either, then the HRTF can be said to be independent of loudness.
In practice, you don't record the impulses too loud because the speakers or maybe the binaural mics will distort too much(you also might not enjoy really loud sweeps). And you won't record quietly either because then you'll have ambient noises and gear noises ruining everything. I think most of us found the right playback level through trial and error, and we're then very glad to stick to that PRIR for all our listening levels.

 

[bigshot]

The simulation of distance cues and sound fields is definitely the next step in recorded sound. It's still in its infancy and the stuff advertised to do it ranges from sorta good to completely fraudulent. The easiest and most effective way to do surround sound is with a speaker system. But that requires a room and costs a bit more ultimately. Simulations aren't the same thing yet, but in the future, it's possible. A lot of the simulations succeed or fail in the precision of the calibration to an individual's unique HRTF. I think that's the area where more development is needed.