Recording Impulse Responses for Speaker Virtualization

[sander99]

the Smyth A16 lets you calibrate your headset with stereo microphones and use it directly with already made hrir if i'm not wrong

You are not wrong but this is far from optimal unless you are very lucky to have a somewhat matching HRTF with the person or dummy head used for the speaker measurement. Really the way to go is to do a personal measurement of speakers, otherwise you will probably just never know what a Smyth Realiser or Impulcifer/HeSuVi is really capable of...
The Smyth also has a manual procedure called manLOUD to improve a (measured) HPEQ (=Headphone Compensation) that can be used to try and compensate somewhat for a different HRTF but even that isn't guarenteed to give a good result (that is not possible because this compensation should in fact be done independently for all the individual impulces, not with one stereo EQ on the stereo headphone signal). (And another aim of the manLOUD is to compensate for issues related to the ear canal resonance and the closed ear canal measurement method.)

 

[Iohfcasa]

It's also not easy to determine the equal loudness of different frequency bands and they use a standard average curve for the hp>>Speaker loudness correction.

 

[reter]

If you don't record your BRIR and only want to compensate headphones, the results probably won't be better than using HeSuVi built-in headphone eqs. Because most common hrirs in HeSuVi are measured in artificial ears, and in that situation, compensating headphones with artificial ears provides better authenticity. If you want to use the "More HRIRs" recorded in human ears, firstly you can't get the same mics as they used and secondly these HRIRs don't include reverberation, thus they sound less pleasant.



I wasn't talking about the final compensation curve but the BRIR measurement curve, which equals the final curve plus hptf curve. So if your mics have the same frequency response as what was used in harman experiment and are inserted close to ear drums, a harman target curve is preferred. If the headphones are tuned to harman target already, then the BRIR curve should be close to the HPTF curve, therefore a flat final compensation curve is preferred. Nevertheless, what I really meant was that the harman target here is just a rough estimation for tuning the balance of high and bass, and no one should equalize their measurement to exact harman target

what about if i want to simulate something like a cinema with my own speakers? can i reproduce it or i will be limited only to my room hrtf?

i don't think my room does fit for what i want to achieve, also i don't think my cheap speaker would be enough to achieve good results

 

[sander99]

what about if i want to simulate something like a cinema with my own speakers?

I am not sure what you are asking:

1. Can you make your stereo pair of speakers (instead of headphones) sound like a multichannel cinema with the help of Impulcifer/HeSuVi?
The simple answer is: No, that is not what it is intended for. HeSuVi (using files produced by Impulcifer) creates a binaural signal intended for headphones. By using headphones it is easy to control what sound goes in to the left ear fully independently from what sound goes into the right ear. To create a similar effect using speakers is a more complicated matter.
However: There is a Smyth Realiser A16 Speaker Edition that seems to be able to do this.
(https://www.head-fi.org/threads/smyth-research-realiser-a16-speaker-edition.960912/)
(They say the A16 needs additional hardware to do this, but I "secretly" do wonder what would happen if you ran the Impulcifer sweeps through such a Realiser Speaker Edition (personalized to you) and created a HRIR from that, and then use HeSuVi with speakers instead of headphones...)

2. Can I use my stereo pair of speakers to measure a complete virtual (over headphones) multichannel loudspeaker system using Impulcifer/HeSuVi?
Yes. You can measure all channels one or two at a time by either repositioning the speakers or repositioning yourself relative to the speakers.

3. Something else?

can i reproduce it or i will be limited only to my room hrtf?

Again I am not fully sure what you are asking. With "room hrtf" do you mean the combination of your room sound with your personal hrtf? That is in fact what you get if you measure with with your own head in your own room with Impulcifer. But Impulcifer has some options to improve the results like with room correction/eq and reverberation management I think. (I don't know the details but they are described somewhere.)
Or do you think a room has a hrtf? That is not the case. Hrtf is short for head related transfer function, and it describes how sound from specific directions is altered by bending around your head and into your ears.

i don't think my room does fit for what i want to achieve, also i don't think my cheap speaker would be enough to achieve good results

Not all properties of the speaker are caputered by the measurement which means that in some ways the virtual speakers can sound better than the real speakers. And again: room correction and reverberation management can further improve the result.
And if your speakers are really bad: maybe borrow or rent a better speaker just for the measurements?

 

[reter]

I am not sure what you are asking:

1. Can you make your stereo pair of speakers (instead of headphones) sound like a multichannel cinema with the help of Impulcifer/HeSuVi?
The simple answer is: No, that is not what it is intended for. HeSuVi (using files produced by Impulcifer) creates a binaural signal intended for headphones. By using headphones it is easy to control what sound goes in to the left ear fully independently from what sound goes into the right ear. To create a similar effect using speakers is a more complicated matter.
However: There is a Smyth Realiser A16 Speaker Edition that seems to be able to do this.
(https://www.head-fi.org/threads/smyth-research-realiser-a16-speaker-edition.960912/)
(They say the A16 needs additional hardware to do this, but I "secretly" do wonder what would happen if you ran the Impulcifer sweeps through such a Realiser Speaker Edition (personalized to you) and created a HRIR from that, and then use HeSuVi with speakers instead of headphones...)

2. Can I use my stereo pair of speakers to measure a complete virtual (over headphones) multichannel loudspeaker system using Impulcifer/HeSuVi?
Yes. You can measure all channels one or two at a time by either repositioning the speakers or repositioning yourself relative to the speakers.

3. Something else?


Again I am not fully sure what you are asking. With "room hrtf" do you mean the combination of your room sound with your personal hrtf? That is in fact what you get if you measure with with your own head in your own room with Impulcifer. But Impulcifer has some options to improve the results like with room correction/eq and reverberation management I think. (I don't know the details but they are described somewhere.)
Or do you think a room has a hrtf? That is not the case. Hrtf is short for head related transfer function, and it describes how sound from specific directions is altered by bending around your head and into your ears.

Not all properties of the speaker are caputered by the measurement which means that in some ways the virtual speakers can sound better than the real speakers. And again: room correction and reverberation management can further improve the result.
And if your speakers are really bad: maybe borrow or rent a better speaker just for the measurements?

thx for the patience! I'll try to explain better my situation

I want to record with impulcifer the hrtf that like you explained i'll be using with my headphones... my room is a little sized one, so i was trying asking if i could indeed manage the reverberation so i can "simulate" something like a cinema instead of just my room with impulcifer

just like i said a post ago, i like the idea that the smyth let you use all the presets (bbc and stuff) so i want to try to do something similar with impulcifer in my room, in 7.1 of course

 

[morgin]

thx for the patience! I'll try to explain better my situation

I want to record with impulcifer the hrtf that like you explained i'll be using with my headphones... my room is a little sized one, so i was trying asking if i could indeed manage the reverberation so i can "simulate" something like a cinema instead of just my room with impulcifer

just like i said a post ago, i like the idea that the smyth let you use all the presets (bbc and stuff) so i want to try to do something similar with impulcifer in my room, in 7.1 of course

My experience is I have a small room too. Id suggest you get hold of a good speaker and do your measurements in a bigger room. Or measure with the speaker further away from you it will have the effect of a big room speaker set up.

I used only one speaker and tested with measurements close and far and they gave me different results. But the best was having the speaker around 6ft away from me. It still offers a 6ft surround setup effect.

Get a good speaker it’s very important and take a lot of measurements with different ear placement, speaker placement, volume etc

 

[Iohfcasa]

The realiser's manspkr method doesn't guarantee a matching result, but it's an approach to personalize alien hrirs.
You can try to replicate the manspkr method with a program like earful, but it's a difficult procedure with many imponderables.

 

[castleofargh]

thx for the patience! I'll try to explain better my situation

I want to record with impulcifer the hrtf that like you explained i'll be using with my headphones... my room is a little sized one, so i was trying asking if i could indeed manage the reverberation so i can "simulate" something like a cinema instead of just my room with impulcifer

just like i said a post ago, i like the idea that the smyth let you use all the presets (bbc and stuff) so i want to try to do something similar with impulcifer in my room, in 7.1 of course

A16 or impulcifier, you’ll get a convincing result when the measurements of the room are done at your own ears. A room measured with a dummy head or at somebody else’s ears, will have components of those HRTF instead of your own. Unless you're lucky and find measurements from someone with about the same head and ears as you, such measurements will not be anywhere as natural as something measured at your ears.
The A16 has some rooms by default, and apparently many users never go through with making their own measurements. All I can say is that it's their loss.
I much prefer my measurements from an acoustically really horrible bedroom, over some fancy opera house in OOYH that just wasn’t working for me. And while we’re talking personal preferences, I prefer little amounts of reverb, and while I can’t claim that you will, I did notice a trend about that for several other users.
I tried a bunch of measurements and ultimately I use one where my speakers were about 1.3meter away from me when I measured them. Again, I don’t have more than my own anecdotes, so they might not apply to anybody else.
I think impulcifier has some OOYH capture somewhere, so you can just try.

 

[reter]

A16 or impulcifier, you’ll get a convincing result when the measurements of the room are done at your own ears. A room measured with a dummy head or at somebody else’s ears, will have components of those HRTF instead of your own. Unless you're lucky and find measurements from someone with about the same head and ears as you, such measurements will not be anywhere as natural as something measured at your ears.
The A16 has some rooms by default, and apparently many users never go through with making their own measurements. All I can say is that it's their loss.
I much prefer my measurements from an acoustically really horrible bedroom, over some fancy opera house in OOYH that just wasn’t working for me. And while we’re talking personal preferences, I prefer little amounts of reverb, and while I can’t claim that you will, I did notice a trend about that for several other users.
I tried a bunch of measurements and ultimately I use one where my speakers were about 1.3meter away from me when I measured them. Again, I don’t have more than my own anecdotes, so they might not apply to anybody else.
I think impulcifier has some OOYH capture somewhere, so you can just try.

i need to ask a question to feed my curiosity: why all the videos that shown the Smyth comes with the sennheiser HD800? is this some sort of optimization thing? still i didn't find a video with a different pair of headphones...weird

anyway i'm starting to understand more what you guys are trying to tell me, there's no other way than to try and do some stuff, but first i need a good speaker, any suggestion? impulcifer does need only one speaker to record the hrir so what should be ideal for this kind of measurements?

 

[morgin]

I was suggested to get the jbl mk2 3 series by @musicero and for the price and quality cannot fault it. My headphones are sennheiser 560s and both sound amazing for my hrir.

 

[castleofargh]

i need to ask a question to feed my curiosity: why all the videos that shown the Smyth comes with the sennheiser HD800? is this some sort of optimization thing? still i didn't find a video with a different pair of headphones...weird

anyway i'm starting to understand more what you guys are trying to tell me, there's no other way than to try and do some stuff, but first i need a good speaker, any suggestion? impulcifer does need only one speaker to record the hrir so what should be ideal for this kind of measurements?

The HD800 has good fidelity and is light/comfy for most people, it also doesn't need hundred volts like a Stax and can be powered by the A16 itself. So it's a rather good candidate. Remember that they're trying to simulate and recreate a specific sound. An objectively good transducer can't hurt when trying to convince people that the simulation is accurate.

Now, because the localization cues are mainly about the right frequency response for a given virtual direction, and the right delay between channels, you only need to avoid headphones with a frequency response in the shape of Mordor and high distortions figures. So, audible stuff that basic EQ won't handle.

At some point you have to make your own choice. Are you happy with something that's already incomparably more natural, relaxing, and in some ways more accurate than typical headphone listening at any price? Or, is your inner maniac audiophile screaming at you for not using a great transducer?
I use a HD650. For the price range it has pretty good specs, except at low freqs which aren't all that important for spatial cues. and after bending the metal plates in the headband to loosen the clamp of death, I find it very comfy(and light), so I use that(not that audiophile of a reason, but that's all I have). I did feel noticeable changes with other headphones, but then again, it's extremely hard to get consistent measurements when any little change placing the mics or the headphone can lead to several dB differences here and there. I don't know for sure if the subjective changes I got came mostly from the headphones being different, from my personal biases about those headphones, or from large measurement tolerances.
@musicreo is able to get stupidly good repeatability, to the point where I wouldn't be surprised if around 1984 he was looking for Sarah Connor. ^_^ the point is; his take on the impact from different headphones is probably most interesting if he's able to stay as consistent with all his measurements.

 

[musicreo]

@musicreo is able to get stupidly good repeatability, to the point where I wouldn't be surprised if around 1984 he was looking for Sarah Connor. ^_^

Some of my measurement have the following filename ending: T800: channel balance Trend, decay time 800ms ^_^

the point is; his take on the impact from different headphones is probably most interesting if he's able to stay as consistent with all his measurements.

I read some years ago in a publication that headphones with small soundstage should be better as they need less compensation (so a HD800 would not be the best choice if this is true ). When I compare my HD600 to the AKG701 this could be true but my HD600 showed me some channel imbalance in my measurement that wasn't really bothering me with my AKG701 and HD555/595 but made listening over the HD600 really exhausting. I could fix that with some additional equalizing which improved also the sound for the two other headphones. Overall all three headphones sound fantastic with the measurement.

If I remember correct jaakkopasanen uses a HD800 so he can probably say more about that headphone.

​

 

[jaakkopasanen]

HD 800 is the most comfortable headphone ever. That's enough of a reason to pick it. For me specifically it's also good because of the very tight manufacturing tolerances and it's flaws are very well known so it's a good reference tool for developing these algorithms.

 

[reter]

Some of my measurement have the following filename ending: T800: channel balance Trend, decay time 800ms ^_^


I read some years ago in a publication that headphones with small soundstage should be better as they need less compensation (so a HD800 would not be the best choice if this is true ). When I compare my HD600 to the AKG701 this could be true but my HD600 showed me some channel imbalance in my measurement that wasn't really bothering me with my AKG701 and HD555/595 but made listening over the HD600 really exhausting. I could fix that with some additional equalizing which improved also the sound for the two other headphones. Overall all three headphones sound fantastic with the measurement.

If I remember correct jaakkopasanen uses a HD800 so he can probably say more about that headphone.

​

mhm, what about imaging then? i have the HD660s that some people state being "smaller in soundstage but with good imaging"... i personally don't know much, i never tried other expensive brands in my life yet, also i think the HD660s does have the same drive the HD800 has if i'm not mistaken

 

[musicreo]

The HD660s should work well. If you it is comfy for you you don't have to look for a other headphone.