Acoustic Simulator

[utilisateur]

Great i think we're on the same boat/path/whatever (continent aswell )

So i guess the first step would be to actualy get a decent aproximation for the IID ITD,
i'm going to acquire some miniature microphone capsules to make some measurements.

I guess there are many psychoacoustic papers that are worth reading aswell , i just dont know how to access them all?

And i think it would be beneficial to take the downsides of known acousitc simulation methods down to paper so one knows in what way exactly a certain misconception affects the presentation!?


NOTES: (i'll just update here for now)

So far i read that below 1,5kHz it's mostly the time difference that counts and above it's mostly the intensity difference.
I think the intensity difference might make the biggest impact here as it's most probably way lower on a standart stereorecording
With a stereo signal we have a problem as we can now only adjust the recorded differenes..
So we have to work with the given ITD as a clue for how to calculate IID?

I'm not sure what do you think can we do anything to the missing HRTF and HRIR ?
Wouldn't we need to extrapolate a sounds origin before we can adjust the signal to mimic HRTF and HRIR? (absurd of course..)

Aparently the interference with the Pinnae very important to get the sound "out of your head"

 

[EloyDark]

Ok, I have direct access to University networks so I can find a lot of papers for studying. I will do some more research within the next few days and get back for reports

 

[nikongod]

End user signal processing of any type (be it passive analog active analog, or digital) is against all of the current trends in "high quality audio." I would not worry as much as it sounds like you are about the meier or linkwitz systems.

It requires a VERY loose definition of "in the signal path" to say that the chu-moy linkwitz passive crossfeed (which the tangent is based on) and the jan-meier designs have caps IN the signal path.

You could always build the crosfeed circuit with inductors instead of caps if you dont like having any caps in your system.

 

[EloyDark]

I can see your point, and I definitelly agree with you. Right now my concern on psychoacoustic effects is not about building something that can add "a little depth in the music".

The topic has already taken a specific direction, this is about trying to achieve the best possible solution, something that can measure the HRTFs, IID and ITD and still give the option of adjustment so that anyone can get the result he wants. Even if in the end this proves to be really hard and is in need of major programming skills, still it will be a great project and a lot of knowledge on human hearing and Signal Processing will be gained.

Anyone can built a very simple crossfeed with just basic knowledge of passive filtering. Hearing however is not that simple, and with the help of computers and their current processing capabillities anything can be achieved! It might turn out a really long and hard project, but what can we do, research is about taking one small step at a time, not jumping to conclusions without really verifying them!

 

[utilisateur]

I got two pairs of miniature condenser capsules. one Panasonic with datasheet and unfortunately the bigger ones are not clearly identified , in case anyone has a clue ? : Link to description


Now i have to look into an apropriate amp, guess i'll start here

Still a few hurdles until measuring, the frequency response is said to be quite flat with the Panasonic capsules but i'm not sure i can rely on that.. or i need to get a verified SPL meter, we have to be sure the soundsource for the measurements is putting out tones at the same volume for all frequencies afterall..


On the theory side of thinks, still waiting for EnOYiN to chime in, he wanted to comment..
Personaly i have not encountered a way to circumvent the inherent problems with a stereo recording,looks like we'd need to simulate many parameters, we'll probably need to set an aproximated width and distance for every recording used.
But once we get to that point maybe we can implement 4/5 channel signals aswell, more fun

1. Process might look like this:
2. Hrtf ITD and IID are measured from different locations with a mono source (Measured once)
3. Frequency response of headphone is measured inear to acount for headphone equalization and placement independant transferfunction of the pinnae (measured for each headphone)
4. Aproximate Width and Distance for the presentation is selected
5. Signal is processed using the measured templates and fed to the normal playback chain

(I feel like i forgot something significant, haven't looked into this for the last few days ,so if my feeling is right whatever is missing here is not forgotten)

 

[EnOYiN]

Chiming in!

Sorry that it took so long, but I've been really busy lately.

The first thing I would like to say is that what you guys are planning on doing here is hard. Really hard. (if not impossible, but I would like to stay positive)

I think that you should start thinking about virtual speakers when you are trying to make something achievable. Depending on the recording it will be impossible to get from stereophonic to binaural. 9 out of 10 times the person who mastered the record just made a mess of it and I'm not just talking about the loudness wars issue. I'm talking about drums which are going from you right to your left ear and back. No way that drummer picked up his complete drum kit and ran from the right to the left within seconds. The realism of such recordings is gone forever and can not be restored.

I'm going to post an idea here tomorrow, because I'll go to school then.(which means I've got about 4-5 hours to spend on head-fi

) I'll also try to comment on the process you described.

 

[EnOYiN]

Right, after a few beers, very few hours of sleep and a lot of red bull. I'm back again.

Quote:

Originally Posted by utilisateur /img/forum/go_quote.gif [*]Hrtf ITD and IID are measured from different locations with a mono source (Measured once)

Main problem with the above is that you probably haven't got an anechoic chamber. If the room you are going to measure this in isn't extremely silent your results won't be any good. However, it doesn't have to be -5 dB either. More important is that it's your own HRTF and not someone else's.

I'm going to check for anechoic chambers in the netherlands and I'll try to find out how much it'll cost me to get my own measured professionally.

Quote:

Originally Posted by utilisateur /img/forum/go_quote.gif [*]Frequency response of headphone is measured inear to acount for headphone equalization and placement independant transferfunction of the pinnae (measured for each headphone)

I'm all for this and I think you can make this work well. Still thinking about how to do this without harming the signal too much. (the problem with equalization is the phase shift)

Quote:

Originally Posted by utilisateur /img/forum/go_quote.gif [*]Aproximate Width and Distance for the presentation is selected

This is too hard for me to make for sure. This requires insane programming skills which I haven't got.

Quote:

Originally Posted by utilisateur /img/forum/go_quote.gif [*] Signal is processed using the measured templates and fed to the normal playback chain

Simple.

Once again, this is way above my programming or math skills. I'm going to ask whether some of my friends are up to this.


I was also going to come up with an idea, but since it has been made already I'm just going to post the link:

HOLISTIKS - Home

They've got the same idea for the virtual speakers and you can even upload your own HRTF. Seems an interesting program.

 

[Crowbar]

N.B.: Acoustic simulator is a very poor term to use for this, because technically an acoustic simulator is a (normally finite boundary method) computer simulation of sound wavefront propagation. What you really mean is acoustic emulator.

 

[utilisateur]

Glad to have you with us

Quote:

Originally Posted by EnOYiN Main problem with the above is that you probably haven't got an anechoic chamber. If the room you are going to measure this in isn't extremely silent your results won't be any good. However, it doesn't have to be -5 dB either. More important is that it's your own HRTF and not someone else's.

I was hoping the measurements would be sufficient if made in rather large hopefuly soundproofed room so that the timewindow form signal to first reflection is long enough

Quote:

Originally Posted by EnOYiN Still thinking about how to do this without harming the signal too much. (the problem with equalization is the phase shift)

If done with a real FIR filter this problem should be rouled out


I'm willing to sacrifice as much time as i can to accumulate the needed knowledge

Filter itself should be rather easy with BruteFIR or other comparable aplication.
But everything around, i'm realy afraid of.Location dependance...

Quote:

Originally Posted by EnOYiN Quote: Originally Posted by utilisateur Aproximate Width and Distance for the presentation is selected This is too hard for me to make for sure. This requires insane programming skills which I haven't got.

For this part i thought we could just measure/calculate for discrete points and then choose the apropriate one

Quote:

Originally Posted by EnOYin Once again, this is way above my programming or math skills. I'm going to ask whether some of my friends are up to this.

sounds great, i guess it wont be that easy to get into this from pretty much zero (programming at least)




My Agenda:
finishing the measurement mics
finding out how a Goniometer works, i think this is what we would need to place the signals in the right place of the stereoimage,although this is still only for the frequency with the highest amplitude...

 

[Jelle Schrijver]

Quote:

Originally Posted by EnOYiN /img/forum/go_quote.gif I'm going to check for anechoic chambers in the netherlands ...

If memory serves me correctly Speakerland has such kind of room for measuring there speakers. Your quest sounds very interesting, would you mind to keep me posted?

 

[Jelle Schrijver]

Quote:

Originally Posted by EloyDark /img/forum/go_quote.gif What I was searching for is something with active components that could mix the channels in a controllable way. Something like an active DSP, like the digital equalisers. Any ideas?

Tension Lab EAP03 has all the goodies you want. No word on SQ yet...
And InEar have/had some beltpack with DSP power. It also included replacable wires for your inearmonitor wich included microphones, so you could mix ambience into your listeningfeed

Quote:

Originally Posted by EloyDark /img/forum/go_quote.gif Yes, kind of like that but not that simple, more like a Behringer DEQ2496 (I know that's too much to ask ) but with crossfeed options as well...

Behringer ultracurve has some kind of crossfeed proces.
And the ultimate gear in this regard is the studer BRS Binaural Room Scanning Processor.

Quote:

Originally Posted by Fixz8 /img/forum/go_quote.gif That's really contradictory. If a single capacitor in the signal path is going to wreck the integrity of your signal... using a DSP and having to go through a ADC and DAC is going to be much worse. So you're looking for a crossfeed with some sort of equalizer capability? So you would have a control for overall "crossfeed strength" and then individual controls for certain frequency ranges? Sort of like a volume knob and then low/mid/high settings on a car stereo?

Active

link

passive

link

link

Quote:

Originally Posted by EloyDark /img/forum/go_quote.gif I have just found out that Behringer has a product called VIRTUALIZER PRO DSP2024P that has psychoacoustic capabilities such as Stereo Imager and Binauralizer with some adjustability. Has anyone tried it? Although it's not exactly DIY what I am asking right now, the thread began with a DIY topic

Had a try once in my IEM setup. Didn't like it. Passive solutions provided better results.

Quote:

Originally Posted by EloyDark /img/forum/go_quote.gif By Rockbox you mean the firmware for the mp3 players? How would rockbox be helpful in what I want? :S (confused once more)

A lot. I've rockboxed my iPod 3G, you'll get a awesome EQ (parametric) AND a fully customisable crossfeed. But, it on my iPod it's a huge load for the processor. With EQ and Crossfeed engaged, It can only play lossless files, because there is no more resource for decoding MP3. So i've your gone go this route, get a iPod Video, it should coupe better. Sound is very good. I am thinking about modding the iPod to include a flashdrive and a nice headphone buffer to max out the performance and be ultra portable. Battery time is not so wonderfull with rockbox and the interface is needs some learning time to get used to.

Quote:

Originally Posted by EloyDark /img/forum/go_quote.gif I have tried some FIR filters ... Any other recommendations?

VST plugins: Crossfeed and EQ for headphones Canz3D

Edit" Oh, just thought about it. Why not build a IEM with several b.a.d's (balanced armature drivers) to simulate the crossfeed and the use of your pinnea. So for example 2 high b.a.d's with one firing into a long curled up tube to simulate the pinnea.

 

[utilisateur]

Looks like even the Studer is suboptimal.
Headtracking sounds great as aparently small headmovement are a normal and automatic human reaction to get better localisation, I guess with a DIY solution this will be above our heads..
From what i could find the Studer uses a standard HRTF though and not your own?
And imo simulating a measured room like many other products do, is not the ideal aproach, although it might be necessary in the end...


The software EnOYiN linked to seems quite promising, but it's some work to get the desired result as you have to measure your HRTF and the headphone IR.

We should have a meet with every single Crossfeed/Binauraliser/Virtualiser/Externalizer/whatever they might be called present, THAT would give us a great starting point

[size=xx-small]Beyer Headzone, Smyth SVS, Studer BSR,diverse Dolby Headphone implementations and the Software plugs..

[/size]

 

[Jelle Schrijver]

I thought the Studer had custom HRTF. That the visit your studio to record your HRTF in your room with your headphones.

I am all for a meat. And to raise the bar even further. Portable would be nice... also i was thinking about combining the schematic's i have giving here to 1 schematic, that would be interesting. And if we start building our own DSP solution, could we use a NOS multibitter like the TDA 1541 for the dac please...

How do you record your HRTF? And keep bodily noises to the minimum?

Last thought. If it does not have to be portable, why not get a mould of your ears and put those in a box with speakers. Add microphones in the moulded ears and feed that to your headamp.

 

[Jelle Schrijver]

What does IDD stand for?

AcronymDefinition
IDDI Don't Disagree
IDDIdeal De-Correlating Detector
IDDIdentity of Key to Be Discontinued (FIPS)
IDDImprovised Device Defeat (US TSWG)
IDDIndeed
IDDInderdaad (Dutch: Indeed)
IDDInformation Data Display
IDDInformation Delivery Device
IDDInfratools Desktop Discovery (Peregrine Systems)
IDDInstrument Deployment Device (NASA JPL Mars Exploration Rover)
IDDInstrument Description Document
IDDInsulin Dependant Diabetes
IDDIntegrated Data Dictionary (CA-IDMS)
IDDIntegrated Dell Desktop
IDDIntentional Deficit Disorder
IDDInterdirector Designation
IDDInterface Definition Document
IDDInterface Description Document
IDDInterface Design Description
IDDInterface Design Document
IDDInterface Designation Drawing
IDDInterim Dry-Docking
IDDIntermediate Design Document
IDDInternal Disc Decompression (lower back pain treatment)
IDDInternal Displacement Division
IDDInternational Direct Deposit
IDDInternational Direct Dialing
IDDInternet Data Distribution
IDDInterpacket Delay Difference
IDDIntervertebral Differential Dynamics
IDDIntervertebral Disc Degeneration
IDDIntervertebral Disc Disease
IDDInvatamint Deschis la Distanta (Romanian: Open Distance Education)
IDDInvestment Dealers' Digest
IDDIodine Deficiency Disorders
IDDIdentifier Dialog
IDDIntegrated Data Directory
IDDInteractive Digital Disc
IDDInterface Design Description Document

 

[utilisateur]

IDD must have been a typo
IID is InterauralIntensityDifference
ITD is InterauralTimeDifference

Although a standalone solution would be great i think if we ever get this far it is not within the next steps, and i think it would be best to just do it as a fully digital box so it's DAC independant

I think HRTF for one angle can be recorded with one Dirac impulse so body noises would hopefully be no big problem.
I will look further into this when the mics are done, unless someone has an aswer ready

Quote:

Last thought. If it does not have to be portable, why not get a mould of your ears and put those in a box with speakers. Add microphones in the moulded ears and feed that to your headamp.

That would be hilarious

But i think head and body play a role aswell for the HRIR[size=xx-small]HeadRelatedImpulseResponse[/size] and it might be beneficial to have more space for the measurement so the early reflections dont interfer with thereflections in the pinnae

UPDATE:

I may leave for some time before this gets done,
i hope i get the mics and amp done but it's not sure, i have to finish themy β22 first.
So this project might get on the backburner for some time

and

Theres another plugin to try by Refined audiometrics Laboratories


Also SPL just brought out a new headphone amp intended for monitoring and Mastering with phones.
It has adjustable Crossfeed with "Center Level" and "Speaker Angle" adjustments.
I'll try to check it out at a Musicstore when they get it