To what extent can headphones simulate a room?

[dshreter]

How well can in room mound be simulated through headphones?

I noticed Gelenec produces Aural ID which is meant to simulate loudspeaker based monitoring through headphones. They have a lot of credibility in my book, so I tend to expect this offering is effective.

Aural ID - Genelec.com

The ultimate in tailor-made headphone monitoring. Create your profile, download the plug-in, and enjoy a headphone listening experience you can finally trust.

www.genelec.com

Are these practices good enough that different rooms and room treatment could be simulated? Could the technology be used to test for preference of different acoustic treatments or speakers?

One nice attribute of headphones is if you could do effective simulation they would be an excellent vehicle for doing blind tests. How amazing would it be if you could test drive speakers through headphones and assess different configurations of absorption or diffusion?

 

[NTK]

Dr @Sean Olive post this on his blog 15 years ago. Wondering how far this technique has come. His AES papers are AFAIK behind the paywall, so I am not familiar with this technique.

Binaural Room Scanning - A Powerful Tool For Audio Research & Testing

Binaural Room Scanning (BRS) is a powerful audio technology being used by Harman scientists to conduct innovative psychoacoustic research a...

seanolive.blogspot.com

 

[dshreter]

Dr @Sean Olive post this on his blog 15 years ago. Wondering how far this technique has come. His AES papers are AFAIK behind the paywall, so I am not familiar with this technique.

Binaural Room Scanning - A Powerful Tool For Audio Research & Testing

Binaural Room Scanning (BRS) is a powerful audio technology being used by Harman scientists to conduct innovative psychoacoustic research a...

seanolive.blogspot.com

Thank you, this is spot on to what I was curious about. It seems this technique has see very little application on the consumer side.

That write up had a reminder of this product in the comments as well.

Realiser A16 the world's most advanced 3d audio processor

The Realiser A16 is the world’s most advanced personalised head-tracked 3D headphone processor, rendering 15.1.8ch Dolby Atmos with breath-taking realism.

smyth-research.com

Apple is doing positional simulations on their EarPods, just shows the potential if some companies with more scale get to work on this.

Perhaps we will see virtual showrooms one day if the traditional hifi industry contracts and dealers stop buying showroom samples.

 

[dasdoing]

it's not really rocket science.
But a perfect simulation of a room sounds very,very bad....because rooms sound bad. our brain just (partially) ignores it when we are in a room, because it expects it to be bad.
over headphones though all the problems of the room become very apparent

 

[somebodyelse]

You could try it with https://github.com/jaakkopasanen/Impulcifer, discussed in a few threads such as:
https://www.audiosciencereview.com/forum/index.php?threads/replicate-imagined-speaker-sound-to-headphones-with-impulcifer.49595/

 

[kemmler3D]

Are these practices good enough that different rooms and room treatment could be simulated?

Could the technology be used to test for preference of different acoustic treatments or speakers?

Short answer, yes and yes.

But to @dasdoing 's point, even simulations of good rooms IME sound very "room-y" because your brain is not (for lack of a better word) "calibrated" for it.

You can try something similar by using https://github.com/ShanonPearce/ASH-Listening-Set (free) these impulse responses with this convolver plugin https://impulserecord.com/convology-xt/ (free) with EQAPO (also free.)

It's not the same as Genelec's implementation by any means but it's the same basic concept. I found it really did help spatialization but it also colored the sound too much for my taste, and I went back to regular crossfeed.

How amazing would it be if you could test drive speakers through headphones and assess different configurations of absorption or diffusion?

Totally possible in theory, but the dispersion of the speaker needs to be completely modeled in the simulation, you can't do it with impulse responses like this, which have the speaker "baked in". You would probably need to synthesize the spatialization filters using a klippel scan, maybe even some kind of raytracing situation, but I don't see why not.

I am not sure who (apart from the consumer) has a big enough incentive to build such a system. But I think it could be built.

 

[dasdoing]

I used to try to "fix" IRs.
this guy does it and does a pretty good job: https://github.com/ShanonPearce/ASH-Listening-Set
you can use this dataset to hear the difference between reflective rooms and treated studios. but as we said already, room IRs sound very roomy.

we don't really want to simulate the room, we want to simulate the space.
algorithms for that used to be very bad, but lately they are MUCH better than what any real IR can do. I don't know the Gelelec one but this is what I use, and it sounds great and "spacy" without coloring the sound (depends on "room" and settings): https://www.dear-reality.com/products/dearvr-monitor

 

[dshreter]

I used to try to "fix" IRs.
this guy does it and does a pretty good job: https://github.com/ShanonPearce/ASH-Listening-Set
you can use this dataset to hear the difference between reflective rooms and treated studios. but as we said already, room IRs sound very roomy.

we don't really want to simulate the room, we want to simulate the space.
algorithms for that used to be very bad, but lately they are MUCH better than what any real IR can do. I don't know the Gelelec one but this is what I use, and it sounds great and "spacy" without coloring the sound (depends on "room" and settings): https://www.dear-reality.com/products/dearvr-monitor

Fantastic! I'm going to try the free trial now.

 

[Smaestro]

How well can in room mound be simulated through headphones?

I noticed Gelenec produces Aural ID which is meant to simulate loudspeaker based monitoring through headphones. They have a lot of credibility in my book, so I tend to expect this offering is effective.

Aural ID - Genelec.com

The ultimate in tailor-made headphone monitoring. Create your profile, download the plug-in, and enjoy a headphone listening experience you can finally trust.

www.genelec.com

Are these practices good enough that different rooms and room treatment could be simulated? Could the technology be used to test for preference of different acoustic treatments or speakers?

Anecdotally I’d say yes.
I use one of these headphone room sims (APL virtuoso). If I put on the “small room” preset and sit still, I can actually fully believe the sound comes my monitors. Even though I know better.
The illusion breaks when I move my head.

You can move speakers around in the room and change room sizes and it seems very real. Also different speaker FR’s can be chosen.

Expanding whatever physics modelling they use to try out different sorts of panelling seems only a matter of measuring and validating.

Though I’m sure that to simulate irregular room sizes or objects in a room, more physics work needs to be done. I can imagine math/physics/coding shortcuts are applied by assuming only rectangular rooms or specific audio type rooms.

 

[LIΟN]

With IEMs, you can have a nearly equal experience, while headphones tend to have a more pronounced coloration that varies from product to product, resulting in less opaque binaural reproduction compared to IEMs. However, it is still acceptable to some extent.
Of course, some sensations, like tactical bass, cannot be recorded and reproduced with an impulse. However, this can be supplemented with additional equipment and a subwoofer, and properly personalized impulses can actually provide an experience that is very close to reality.
Of course, some publicly available datasets or personalized spatial audio from companies like Apple or Samsung aren't bad. However, they're not great either. They do not fully replicate the experience of listening to real speakers. (While some head tracking may give the illusion of improved spatial representation, it seems that the truly mysterious responses beyond the secondary folds of the ear cannot be correctly implemented with the current scanning and SOFA methods.)

And yes, it becomes possible to make clearer comparisons across different spaces, such as aspects related to diffusion or sound absorption treatments. Especially when listening directly in reality, the noise floor cannot be ignored. However, with the incredible seal of custom CIEMs, there is no noise present, allowing everything to be heard more starkly and realistically. Additionally, digital manipulation is also possible to some extent. You can listen only to direct sound, only to early reflections, or only to late reflections, manipulate certain HRTFs in the time domain of reflections, absorb them, or XTC(Crosstalk Cancelation), or even manipulate the space itself. Yes, it is possible. Some users, including myself, are doing this. However, in Korea, where I live, and worldwide, there have been very few users discussing and manipulating these aspects.

Returning to the main topic, I personally don't have any particular thoughts about the Aural ID service.. Impulcifer is free.

 

[kemmler3D]

Though I’m sure that to simulate irregular room sizes or objects in a room, more physics work needs to be done.

Maybe not, you could download free room simulation raytracers like 10 years ago, so simulating an irregular room was not state of the art even then. And consider that we have FEM / BEM systems that can simulate acoustics in fine detail for speaker cabinets and such.

My point is just that the science is already there, implementing it in a given consumer software package is another question.

 

[1st.Virtual.Me]

Audio Modeling just announced Abiente plugin for use in a DAW

So, for example, Stem separation is now everywhere, so shifting around a sound stage could be very doable for a personal remix. And/or with something like Wave's Abbey Road Studio, or IK Multimedia Sunset Sound Studio.

I also note that Klippel NFS can create IR's from speaker cabinets as measured. It might be interesting to try a couple of Klippel generated IR's to see if a characteristic sound of a cabinet could be heard over headphones/IEMs. It might be fun to try.

 

[LIΟN]

Maybe not, you could download free room simulation raytracers like 10 years ago, so simulating an irregular room was not state of the art even then. And consider that we have FEM / BEM systems that can simulate acoustics in fine detail for speaker cabinets and such.

So, for example, Stem separation is now everywhere, so shifting around a sound stage could be very doable for a personal remix. And/or with something like Wave's Abbey Road Studio, or IK Multimedia Sunset Sound Studio.

Thank you both for sharing the information. It seems like most sound-related simulations were typically compatible only with Mac, but I’ll definitely take a look at this one as well.

I also note that Klippel NFS can create IR's from speaker cabinets as measured. It might be interesting to try a couple of Klippel generated IR's to see if a characteristic sound of a cabinet could be heard over headphones/IEMs. It might be fun to try.

I'm curious about this, but I also wonder if it might actually be challenging or even difficult to perceive the characteristics as they truly are. Characteristics like sound power and directivity, along with various other speaker traits, seem likely to change significantly depending on factors like listening distance and the size of the space.

 

[kemmler3D]

Characteristics like sound power and directivity, along with various other speaker traits, seem likely to change significantly depending on factors like listening distance and the size of the space.

Yes, a normal impulse response can't capture this, there is no spatial component. You would need several IRs or some other method entirely of representing speaker directivity in a room simulator.

 

[LIΟN]

Yes, a normal impulse response can't capture this, there is no spatial component. You would need several IRs or some other method entirely of representing speaker directivity in a room simulator.

Yes you are right.
and to truly "listen to it as if hearing it in reality," a vast amount of HRTF data would also be needed to account for the multiple reflection paths within the space.
That makes it quite challenging, indeed. =)
Of course, if someone were to go to any given space, acting as their own dummy head and recording each sound based on their own head, torso, and ears, then bringing those recordings home for A/B testing, it would work accurately. However, the most cumbersome part is that it has to be recorded with the listener's own head, torso, and ears.


That's why I think there's limited value in listening to public binaural IR data or spatial data recorded by someone else, as mentioned earlier in this thread. In the most extreme example, one of the spaces I recorded was a large 30m x 21.3m x 19.7m space. However, when I shared this with someone whose HRTF didn't match mine, they couldn't perceive it as intended, despite calibrating the headphone/IEM compensation curve as precisely as possible. The main issue was that the ITD cues based on basic ITD/ILD weren't aligned, so it wasn’t accurately perceived. Instead, they found that their own recording from a much smaller, approximately 2.7m space sounded more realistic to them.
While a mismatched HRTF can still offer some imaging or a faint phantom center effect (especially in True Stereo, due to how we perceive sound), it's far from fully capturing spatial realism.
In real-life listening, the brain applies natural compensation, but with binaural reproduction, it often quickly becomes confused. This can sometimes work as an advantage, but often, the brain identifies discrepancies faster than expected. It realizes that what it's hearing isn’t authentic but rather just a particular IR effect being played. The result is usually limited to an impression of a soundstage forming directly in front, without the depth of true realism.
However, if the recording is done accurately, capturing one’s own ears and torso without such errors, then it’s simply like being in the space, with the speaker sounds being authentic. But achieving this level of precision—where one could perform A/B testing without variables, comparing subtle nuances of speaker or room absorption differences—is not easy. It requires everything to be perfectly aligned, making this level of testing quite challenging.

 

[phoenixdogfan]

Yes you are right.
and to truly "listen to it as if hearing it in reality," a vast amount of HRTF data would also be needed to account for the multiple reflection paths within the space.
That makes it quite challenging, indeed. =)
Of course, if someone were to go to any given space, acting as their own dummy head and recording each sound based on their own head, torso, and ears, then bringing those recordings home for A/B testing, it would work accurately. However, the most cumbersome part is that it has to be recorded with the listener's own head, torso, and ears.

As I did when I went to a studio where John of 3D Soundshop, I, and another friend recorded 24 channels of our own HRTFs with Dutch & Dutch 8C speakers using the Smyth A16 Realizers with their binaural in ear microphones.

And yes, it does indeed capture the experience of speaker's playing in a room, so much so that it betters my own 5.2 system of Kef LS 50's + SVS SB 2000's +Dirac doing their home theater thing in my living room.

Been using that system as my go to home theater for the last 3 years now, and I'm very happy with it.

 

[LIΟN]

As I did when I went to a studio where John of 3D Soundshop, I, and another friend recorded 24 channels of our own HRTFs with Dutch & Dutch 8C speakers. And yes, it does indeed capture the experience of speaker's playing in a room, so much so that it betters my own 5.2 system of Kef LS 50's + SVS SB 2000's +Dirac doing their home theater thing in my living room.

Yes, the experience with BRIR is excellent in this regard, but when discussing it, I often get asked questions like, “That’s impossible with headphones or IEMs. Why use headphones/IEMs when you have speakers?”
HRTF represents the frequency response generated when the body (head, torso, ears) reacts to a sound from a specific angle. When represented in the time domain, it becomes HRIR, and if spatial reflections are included, it becomes BRIR. This aligns exactly with how we hear in reality (provided the proper headphone compensation curve is applied on the playback device, like headphones or IEMs). However, I’ve rarely seen people receive this concept positively or accept it as it is with an open mind.....

 

[phoenixdogfan]

Yes, the experience with BRIR is excellent in this regard, but when discussing it, I often get asked questions like, “That’s impossible with headphones or IEMs. Why use headphones/IEMs when you have speakers?”
HRTF represents the frequency response generated when the body (head, torso, ears) reacts to a sound from a specific angle. When represented in the time domain, it becomes HRIR, and if spatial reflections are included, it becomes BRIR. This aligns exactly with how we hear in reality (provided the proper headphone compensation curve is applied on the playback device, like headphones or IEMs). However, I’ve rarely seen people receive this concept positively or accept it as it is with an open mind.....

Thing is, not everyone can set aside the space and have a room with 24 speakers (especially D&D 8Cs) or, even if they can, be able to play them at any hour of the day or night without disturbing everyone else in the house and/or neighborhood. I could literally live in a van with a 50 inch OLED and my Realiser, and still have that experience.

 

[LIΟN]

Thing is, not everyone can set aside the space and have a room with 24 speakers (especially D&D 8Cs) or, even if they can, be able to play them at any hour of the day or night without disturbing everyone else in the house and/or neighborhood. I could literally live in a van with a 50 inch OLED and my Realiser, and still have that experience.

I agree with you.
Sometimes, I go to the forest to listen. (Especially in these cases, I bring my Apple EarPods instead of CIEMs, so the ambient outdoor noise can blend in naturally.) In this setting, the impulse of the forest, the real sound of my footsteps on the leaves, the birds chirping, and the visual image of the forest all align perfectly.