[jaakkopasanen]

Hey guys... I'm already at my personal endgame with headphones, and I have the EQ settings I want to stick with forever (or until aging drastically changes my hearing). Meanwhile, I don't know the first thing about speakers or the speaker market. I don't own any speakers except for a passable bluetooth speaker to connect to my phone. I don't have any treated rooms or anything like that. Could anyone give me an idea what it would cost and how much trouble it would be for me to try to do the measurements for Impulcifer on my own? Would there be enough people like me to justify some kind of speaker loaner tour? Are people finding any noticeable difference between the different Sound Professionals microphones; is there good enough reason to go above the $20 model? Or would anyone happen to be near USA - SC? I'd gladly pay a few bucks to use someone else's setup.

I'm also wondering whether this can be used to make sense of how and why peoples' preferences in headphone FR differ from the Harman Curve. I've dialed my midrange preference in very clearly: it begins rising gradually in the lower mids, it doesn't begin the steep rise seen on the Harman Curve until past 2kHz, and it looks like Harman with a few dB less peak from 3kHz onwards. Might measurements like these help clarify whether I have different preferences from people that like the Harman Curve, or need a different headphone FR to have the same FR delivered to my ear drum that fans of the standard Harman Curve are getting delivered to theirs? Might binaural measurements together with Harman Curve FR calibrations be used to quickly dial in any given person's most preferred signature?

You can get convincing externalization with any half-decent speaker but if you want timbral accuracy, nothing can really replace good speakers in a good room. An affordable option would be to get a single JBL 305P MkII speaker which costs about 120€ in Europe.

The sound professionals SP-TFB-2 mics work well. A cheaper option is two Primo EM258 mono modules from FEL Communications with the added benefit of being able to connect these directly to two RODE VXLR+ adapters. Just make sure to buy mics that fit at the entrance of your ear canal. There are models on the market which are too large for this.

Room treatment isn't necessarily so important since it's the speakers which dominate sound above ~300 Hz and Impulcifer can get the low frequencies in control with room correction and reverb management. You will need a measurement mic for this but it's a lot cheaper option than room treatment. Of course many people listen to speakers without any room treatment or EQ and enjoy the music just fine.

 

[johnn29]

The used market is also a great place to pickup good speakers. If you check audiosciencereview.com you'll see some great speakers and their preference score based on Harman research.

I never thought that some people have grown up with no stereo loud speakers as a reference. It's going to be really interesting to get your thoughts on what that sounds like vs headphones.

Of interest to the thread: https://www.aes.org/tmpFiles/elib/20200825/20868.pdf

Summary of Publication:

When spatial audio content is presented over headphones, the audio signal is typically filtered with binaural room impulse responses (BRIRs). An accurate virtual auditory space presentation can be achieved by flattening the headphones’ frequency response. However, when presenting stereo music over headphones, previous studies have shown that listeners prefer headphones with a frequency response that simulates loudspeakers in a listening room. It is as yet unclear if headphones that are calibrated in such a way will be preferred by listeners in the context of spatial audio content as well. This study investigates how listeners’ preferences for headphone frequency response may differ between stereo audio content and spatial audio content, which was rendered by convolving the same stereo content with in-situ-measured BRIRs of loudspeakers in a room.

Summary (from reddit)

It was shown that an individually calibrated flat response was the preferred choice for spatial audio (binaural) content, but not for stereo content, for which the Harman target was rated significantly higher. This explicitly confirms that content-dependent HpEQ would be beneficial for devices designed to reproduce both spatial and non-spatial audio, such as head-tracked headphones, VR headsets and AR glasses.

Interestingly enough I preferred using AutoEQ to get to flat for any headphones I used (especially IEM's) for Out of your Head use.

 

[phoenixdogfan]

I have a Smyth A16 Realiser and would like to use the binaural mics for another project which involves measuring my hrtf on my existing speakers and sending it to a company so they can generate room and crosstalk cancellation filters similar to those created by the $5000 Baach system, only this software costs only $300.

Question is how can I do this. Do the Smyth mics need a phantom power source, and what about how would I create a mic calibration curve?

This is the software if anyone is interested.

https://www.homeaudiofidelity.com/english/home/

 

[sander99]

I have a Smyth A16 Realiser and would like to use the binaural mics for another project which involves measuring my hrtf on my existing speakers and sending it to a company so they can generate room and crosstalk cancellation filters similar to those created by the $5000 Baach system, only this software costs only $300.

Question is how can I do this. Do the Smyth mics need a phantom power source, and what about how would I create a mic calibration curve?

This is the software if anyone is interested.

https://www.homeaudiofidelity.com/english/home/

About using the A16 mics for other measurements with in-ear mics, for example with Impulcifer:
The A16 has a mode in which the signal of the mic inputs is played back as-is over the user B headphone outputs (analog RCA and digital outputs also).
This how you activate it:
Main menu -> Apps -> Listen to microphones on HPB.

I have a cheap Behringer usb audio interface with which I could get the signal from the analog RCA outputs (of the borrowed A16 I had a while back) back into my pc. (Unfortunately it didn't have spdif in so I had to take the analog detour, but I don't think that has any practical consequences for these kind of measurements).


About the $300 dollar software: Interesting, but I have one advice for you before you invest in this:
With the A16 you can simulate reduced cross talk or eliminated cross talk with a trick. (And with Impulcifer it may even be more easy to do this, only then you won't have headtracking).

And it could very well be that you don't like the effect at all! Because it is not at all obvious that it is an objective improvement. Some people will like it, others don't.
If you try first to simulate it with the A16 or Impulcifer then you know more before spending $300.

Eliminate cross talk all together can be done by doing a manipulated PRIR measurement where you unplug the right mic while measuring the left speaker, and unplug the left mic while measuring the right speaker. (Although I didn't try this yet. But people did it succesfully with the A8.)

Just reducing cross talk with the A16 is a little bit more complicated. I can try to work out the details later, start reading here to get the idea for how to do it with the A16 (only there is a little problem because the individual channels can not be attenuated, let alone per user, which is a condition for the "dual user mode" method I propose but maybe I can think of way around that):
https://www.head-fi.org/threads/correcting-for-soundstage.865212/page-2#post-13906823

 

[musicreo]

Is there a solution for the plotting issue?
Only the pre folder is plotted. I still get following error since the last impulcifer update.

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...
** On entry to DLASCLS parameter number 4 had an illegal value
** On entry to DLASCLS parameter number 4 had an illegal value
Traceback (most recent call last):
File "impulcifer.py", line 513, in <module>
main(**create_cli())
File "impulcifer.py", line 130, in main
fr.smoothen_heavy_light()
File "\Impulcifer\venv\lib\site-packages\autoeq\frequency_response.py", line 1240, in smoothen_heavy_light
treble_iterations=1
File "\Impulcifer\venv\lib\site-packages\autoeq\frequency_response.py", line 1197, in smoothen_fractional_octave
treble_f_upper=treble_f_upper
File "\Impulcifer\venv\lib\site-packages\autoeq\frequency_response.py", line 1155, in _smoothen_fractional_octave
y_normal = savgol_filter(y_normal, self._window_size(window_size), 2)
File "\Impulcifer\venv\lib\site-packages\scipy\signal\_savitzky_golay.py", line 337, in savgol_filter
coeffs = savgol_coeffs(window_length, polyorder, deriv=deriv, delta=delta)
File "\Impulcifer\venv\lib\site-packages\scipy\signal\_savitzky_golay.py", line 139, in savgol_coeffs
coeffs, _, _, _ = lstsq(A, y)
File "\Impulcifer\venv\lib\site-packages\scipy\linalg\basic.py", line 1218, in lstsq
raise LinAlgError("SVD did not converge in Linear Least Squares")
numpy.linalg.LinAlgError: SVD did not converge in Linear Least Squares

 

[jaakkopasanen]

Is there a solution for the plotting issue?
Only the pre folder is plotted. I still get following error since the last impulcifer update.

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...
** On entry to DLASCLS parameter number 4 had an illegal value
** On entry to DLASCLS parameter number 4 had an illegal value
Traceback (most recent call last):
File "impulcifer.py", line 513, in <module>
main(**create_cli())
File "impulcifer.py", line 130, in main
fr.smoothen_heavy_light()
File "\Impulcifer\venv\lib\site-packages\autoeq\frequency_response.py", line 1240, in smoothen_heavy_light
treble_iterations=1
File "\Impulcifer\venv\lib\site-packages\autoeq\frequency_response.py", line 1197, in smoothen_fractional_octave
treble_f_upper=treble_f_upper
File "\Impulcifer\venv\lib\site-packages\autoeq\frequency_response.py", line 1155, in _smoothen_fractional_octave
y_normal = savgol_filter(y_normal, self._window_size(window_size), 2)
File "\Impulcifer\venv\lib\site-packages\scipy\signal\_savitzky_golay.py", line 337, in savgol_filter
coeffs = savgol_coeffs(window_length, polyorder, deriv=deriv, delta=delta)
File "\Impulcifer\venv\lib\site-packages\scipy\signal\_savitzky_golay.py", line 139, in savgol_coeffs
coeffs, _, _, _ = lstsq(A, y)
File "\Impulcifer\venv\lib\site-packages\scipy\linalg\basic.py", line 1218, in lstsq
raise LinAlgError("SVD did not converge in Linear Least Squares")
numpy.linalg.LinAlgError: SVD did not converge in Linear Least Squares

I'm afraid I haven't managed to fix it because I can't reproduce the problem on my computer. I have one idea that would probably fix it but I'm not too keen on taking that route.

 

[kalstein]

I'm afraid I haven't managed to fix it because I can't reproduce the problem on my computer. I have one idea that would probably fix it but I'm not too keen on taking that route.

I have same error. I think you can test it new os system. For example virtualbox.

 

[johnn29]

Been using my AfterShokz Aeropex for a lot of spoken word stuff on my phone. Decided to throw one of my BRIR's from Impulcifer for watching talking head YouTube videos at my desk. Pretty amazing how realistic the BRIR is when your ear is completely unobstructed and you forgot you're wearing headphones. Shame bone conduction isn't yet Hi-Fi.

 

[Crema]

I just finished working in the worst environment. Here's my data.

https://gall.dcinside.com/mgallery/board/view?id=speakers&no=121168

The microphone I used was cut off and repaired, but the balance of the left microphone was broken, resulting in a roll-off in the high frequency range.

So I corrected the response of the left microphone with a Convolver and gave the right microphone a delay for measurement.

However, the measurement results were ringing, and the overall quality was slightly reduced.

I was frustrated. I did the measurement in tears and mustard, but I wasn't actually expecting it.

But the sound is good. It's worth listening to. Excellent calibration technology! Thank you for making this great impulcifier!

I'm now listening very well to the HRIR made with HD800S and B2031A. The sound is amazing!

Dummy head Comparison

Here's my dummy head response. There are three types: original, speaker, and impulcifer headphones. The sound quality has dropped a little to remove Humnoise, but I think it will be possible to compare. In fact, there's not much difference between headphones and speakers. Haha;

There are various songs. From pop songs to classical music to game music.

Originally, I was going to make a YouTube video, but copyright became a problem. So we share the original. If you're interested, download it and compare it.

Once again, thank you for making this wonderful work. Cheers!

 

[lowdown]

Once again, thank you for making this wonderful work. Cheers!

Yes, from me as well, once again and every day, thank you Jaacko!

My only "regret" about Impulcifer is no one else can hear what I'm hearing as they could with actual speakers. Yes, each can do their own measurements, but are they hearing what I am? It is so good I think if $10,000 speakers sounded like this they'd be considered an amazing bargain.

 

[jaakkopasanen]

I'm super glad to hear more people find Impulcifer and are liking it!

 

[amanieux]

this project seems really interesting, thanks for sharing your work, is what android apps like wavesnx is doing ? (do they use your open source code ) ?

 

[halcyon]

Somebody with an access to 7.1 full range speaker setup and an anechoic chamber, would do the whole field & community a huge service, by recording an HRIR using their own head (or even a B&K stereo mic dummy head). This would make a great generic starting point for a 7.1 Virtualization HRTF setup for HeSuVi, without all the added room reflections and artificial echo. The acoustically driest starting point, so to speak.




When I was studying (back in the ages), my engineer friends had routine access to a proper anechoic chamber along with multi-channel speaker setups. Unfortunately, no more...

 

[musicreo]

This would make a great generic starting point for a 7.1 Virtualization HRTF setup for HeSuVi, without all the added room reflections and artificial echo. The acoustically driest starting point, so to speak.

There a databases that provide such measurements. HESUVI uses already some of these measurements.
But the whole idea of impulcifer is to use your own measurements as measurements done by someone else will not give you a proper speaker virtualization.

 

[halcyon]

There a databases that provide such measurements. HESUVI uses already some of these measurements.
But the whole idea of impulcifer is to use your own measurements as measurements done by someone else will not give you a proper speaker virtualization.

Could you point to a 7.1 loudspeaker impulcifier measurements file done with stereo ear canal microphones in an anechoic chamber? Real person or dummy head, doesn't matter.
I have checked the latest HeSuVi 2.0.0.1 installer and I can only find pre-baked 3D VSS impulse HRIRs and some room based measurements, not anechoic. I don't have access to an anechoic chamber myself, and would like to compare a dead space HRIRs.
Thanks!