stereo to binaural using 14 DSPs (i've uploaded Samples!!!) (foobar2000, realtime!)

[Artemus GF]

short overview:
http://upload.paul-doege.de/files/61...53a7d6ce_1.PNG
more information:
http://www.head-fi.org/forums/3417398-post8.html

Sample "sweet dreams": (setup m6F)
-FLAC 24bit/96kHz: RapidShare: 1-Click Webhosting
-WAV downsampled to 44.1kHz/16 bit: RapidShare: 1-Click Webhosting
(you can upload this files to other websites and post link here if you want, i know some people have to wait at rapidshare)

the filename shows what HRTF-file i have used, "stereo.*" is the original file with adjusted loudness.

somebody want more samples?

 

[Alleyman]

What is "mit kemar" sample?

I'd also like to request something. Can you please run the attached impulse wave file through foobar's converter and your own Stereo to binaural DSPs enabled. This is to check if convolution of stereo music with the impulse response file gives the same binaural result. If it does, it would probably mean savings in computing time.

Link to foobar convolver DSP: foobar2000 components

I like the end result of your experiment.

 

[Artemus GF]

Quote:

Originally Posted by Alleyman /img/forum/go_quote.gif What is "mit kemar" sample?

...
Binaural recording technique:
With a simple recording method, two microphones are placed seven inches (18cm) apart facing away from each other. This method will not create a real binaural recording. The distance and placement roughly approximates the position of an average human's ear canals, but that is not all that is needed. More elaborate techniques exist in pre-packaged forms. A typical binaural recording unit has two high-fidelity microphones mounted in a dummy head, inset in ear-shaped molds to fully capture all of the audio frequency adjustments (known as head-related transfer functions (HRTFs) in the psychoacoustic research community) that happen naturally as sound wraps around the human head and is "shaped" by the form of the outer and inner ear. The Neumann KU-81, and KU-100 are the most commonly used binaural packages. The KEMAR system is another alternative. The more expensive Aachen Head Acoustics unit does automatic equalization and processing to create a more enveloping experience....
Binaural recording - Wikipedia, the free encyclopedia
i think "MIT KEMAR" is the HRTF data recorded by this KEMAR system.


Quote:

Originally Posted by Alleyman /img/forum/go_quote.gif I'd also like to request something. Can you please run the attached impulse wave file through foobar's converter and your own Stereo to binaural DSPs enabled. This is to check if convolution of stereo music with the impulse response file gives the same binaural result. If it does, it would probably mean savings in computing time. Link to foobar convolver DSP: http://www.foobar2000.org/components/foo_convolve.zip I like the end result of your experiment.

i hope i've done the rigth:
RapidShare: 1-Click Webhosting

 

[Alleyman]

Thank you for the explanation.

Rapidshare has a long wait on me right now since I downloaded your previous 97 MB file. Although I'm surprised the impulse response file came out to be 32 MB. (I'm guessing you have upsampled the impulse file to 24/96)

In any case, you can check if the impulse response works correctly by using the convolver DSP.

1. Add the convolver DSP in foobar's DSP chain.
   Remove other DSP.
   Configure the convolver and load the impulse response file.
   Play your stereo file as normal
   Check to see if it gives you the same binaural effect.

 

[Alleyman]

Just checked the second file. Sorry but it's not what I wanted. This is how you can get an impulse response of your DSP.

-Load the unitpulse2K.wav in a foobar playlist.
-Right click on the file, select convert > Convert to...
-Select WAV as the encoder
-Uncheck replaygain processing
-Check DSP processing. Open the DSP configuration page by clicking the button on the right. Make sure the same VST configuration (any HRTF) is selected as used for your samples. Also remove any resamplers from the DSP chain (no upsampling to 96khz). Click Ok and exit out.
-Click OK again to start conversion. It will ask you for a location to save the impulse response. This file will be quite small in size.
-This impulse response will now simulate the whole DSP chain in one go when used with the convolver plugin.

I hope I was clear with my instructions.

 

[Artemus GF]

Now its clear what you want but it dont works, the music sounds different to the original but it has not the same binaural effect like my DSP chain.

 

[Alleyman]

Thanks Artemus_GF.

I figured it would be like so. If you are not already tired of this I have one more suggestion. Foobar also has a stereo convolve plugin that was developed specifically for HRTF modelling. I will keep my blabbering to minimum and link you to the thread discussing the plugin.

New plugin (05/18/07): Stereo Convolver 1.0.1 - Hydrogenaudio Forums

Thus if your DSP chain allows you to process left and right channels separately, you might be able to simulate it using stereo convolution (or not). What is there to lose huh?! I wasnt sure of what kind of impulse file it would need to the get the desired response so I didnt suggest it earlier.

In any case, can you write a guide as to how you got the effect to work. I see the screenshots in your other thread but you didnt mention what VST plugins you used. If I can replicate your settings on my PC I can try do my own impulse response modelling.

 

[Artemus GF]

Quote:

Originally Posted by Alleyman /img/forum/go_quote.gif Thanks Artemus_GF. I figured it would be like so. If you are not already tired of this I have one more suggestion. Foobar also has a stereo convolve plugin that was developed specifically for HRTF modelling. I will keep my blabbering to minimum and link you to the thread discussing the plugin. New plugin (05/18/07): Stereo Convolver 1.0.1 - Hydrogenaudio Forums Thus if your DSP chain allows you to process left and right channels separately, you might be able to simulate it using stereo convolution (or not). What is there to lose huh?! I wasnt sure of what kind of impulse file it would need to the get the desired response so I didnt suggest it earlier. In any case, can you write a guide as to how you got the effect to work. I see the screenshots in your other thread but you didnt mention what VST plugins you used. If I can replicate your settings on my PC I can try do my own impulse response modelling.

i will have a look at this later but i think i will continue work with Panorama 5. To get a good and "audiophile" binaural effect you need more than just HRTF and Panorame offers some high configurable reflection and reverb settings. Some discreet, barely noticeable reflection add as mutch spatila-effect to the sound as a simple HRTF-processing. (Its relative difficulty to find good reflection setting, i did big improvements on this since setup 6b(screenshots), 6F is the newest version)

How my dsp chain basically works:

i am using the lowest "PSP Neon HR"-VST to filter the bass out of the signal, both bass-channels going into a Panorama 5 vst optimized for creating a very deep bass using relative loud reflections and some quiet reverb. the wall material for the reflection of all 6 walls is "Acoustic tile" witch IMO sound best on bass. the virtual speakers are at 70 degrees in front of you, the locating of bass dont works good (also in reality) so it dont make sens to reduce this angle.

the middle "PSP Neon HR"-VST filters out the mids. the two left "+matrix"-VSTs split the stereo channels into two channels, the first contains the center information of the signal and the second contains the side information (LR to MS = LeftRight to MidSide). the mid/center-channel is processed by the "BauerStereo2Binaural"-VST and the side-channel comes from a Panorama 5 vst with is optimized for mids (i'am using 3 diferrent wall materials for reflections and no reverb) . the rightest "+matrix"-VST is to combine and convert the center and side channels back to LR (MS to LR).
"CSL_MM_S4S1" is used as an bypass.

the upper PSP Neon HR and Panorama 5 is for highs, and CSL_MS_S4 is just a mixer and bypass.

used VSTs:
KVR: ART Teknika Console Sound Modular Studio - Virtual Instrument - Virtual Effect

KVR: PSPaudioware PSP Neon - Virtual Effect
(could be replaced by a lot of free DSPs)

KVR: Wave Arts Panorama - Virtual Effect

KVR: Soundhack +matrix - Virtual Effect

Bauer stereophonic-to-binaural DSP

and some other DSPs at some setups.



all my setups (33 files):

 

[Sarchi]

Quote:

Originally Posted by Artemus GF /img/forum/go_quote.gif short overview: http://upload.paul-doege.de/files/61...53a7d6ce_1.PNG more information: http://www.head-fi.org/forums/3417398-post8.html Sample "sweet dreams": (setup m6F) -FLAC 24bit/96kHz: RapidShare: 1-Click Webhosting -WAV downsampled to 44.1kHz/16 bit: RapidShare: 1-Click Webhosting (you can upload this files to other websites and post link here if you want, i know some people have to wait at rapidshare) the filename shows what HRTF-file i have used, "stereo.*" is the original file with adjusted loudness. somebody want more samples?

More samples would be great, especially some jazz or classical. I only have a lousy laptop 16/44 soundcard; and I used my KSC's....so take my comments with a grain of salt. The differences are pretty clear, and it doesn't seem like the processing is detracting from the music in any serious way. This is promising. Mind you, the different binaural codecs all sounded about the same, with Eurythmics at least.

The simplest way to describe it would be that the soundstage is taller, and a little wider. So the hard l/r effect of using headphones is reduced (maybe not unlike crossfeed in a way?).

For us non-foobar people, can you describe the process to do this conversion? What software is needed?

 

[bigshot]

The advantage of binaural recordings isn't the "thickness" of the phase or the realism of the reverb, it's the ability to pinpoint individual instruments in the soundstage. Reverberation and reflection aren't necessarily an element of binaural recordings. They're an element of a particular recording venue.

What you're doing here as I see it is making the music itself more mono and wrapping a binaural sounding reverb around it. The sound is like putting a single speaker on the stage of an auditorium and recording the way it sounds with all the room acoustics mixed in. That's not binaural. That's like putting chocolate icing on a ham and cheese sandwich and calling it a cake.

See ya
Steve

 

[Artemus GF]

Quote:

Originally Posted by Sarchi /img/forum/go_quote.gif For us non-foobar people, can you describe the process to do this conversion? What software is needed?

post #8 ?!

Quote:

Originally Posted by bigshot /img/forum/go_quote.gif What you're doing here as I see it is making the music itself more mono ...

binaural sound is always more "mono" than stereo because a sound can't reach only one ear the sound reach alway both ears.

Quote:

Originally Posted by bigshot /img/forum/go_quote.gif The sound is like putting a single speaker on the stage of an auditorium and recording the way it sounds with all the room acoustics mixed in. That's not binaural. That's like putting chocolate icing on a ham and cheese sandwich and calling it a cake.

What?

Quote:

Originally Posted by Sarchi /img/forum/go_quote.gif More samples would be great, especially some jazz or classical.

i will upload more samples if im back home but im dont have jazz or classical, maybe you like Bush - Gglycerin?

Quote:

Originally Posted by bigshot /img/forum/go_quote.gif The advantage of binaural recordings isn't the "thickness" of the phase or the realism of the reverb, it's the ability to pinpoint individual instruments in the soundstage.

remember: i have just stereo-material to work with, you schould'nt mess my samples with real binaural recordings you schould mess it with crossfeeds and things like that.

 

[HFat]

Quote:

Originally Posted by Artemus GF /img/forum/go_quote.gif What?

bigshot is a bit weird and aggressive but he often makes good points.
I understood that what you're doing is similar to virtualization (such as Dolby Headphone). In my opinion, such processing is only good for close-miked recordings that contain no room ambiance.

Quote:

Originally Posted by Artemus GF /img/forum/go_quote.gif i will upload more samples if im back home but im dont have jazz or classical, maybe you like Bush - Gglycerin?

Google returned this on its first page for "\"creative commons\" classical":
OnClassical: copyright
Opsound: genre classical
Obviously, this is only scratching the surface of free music. Please note that not all creative commons licenses authorize you to modify the works and that this forum doesn't condone copyright infringement... I'm not a lawyer and I don't know if that applies to the sample you posted but you may want to consult with a mod before posting large samples of non-free music in the future.

 

[bigshot]

Quote:

Originally Posted by Artemus GF /img/forum/go_quote.gif binaural sound is always more "mono" than stereo because a sound can't reach only one ear the sound reach alway both ears.

The accuracy of stereo separation is MUCH more critical in a binaural recording than in a regular stereo recording. If you switch on crossfeed in a binaural recording, you can kiss the soundstage goodbye.

As I said before, you aren't creating binaural sound. You're creating mono sound with a realistic sounding digital ambience wrapped around it. That isn't creating soundstage- it's killing it.

See ya
Steve

 

[HFat]

Quote:

Originally Posted by bigshot /img/forum/go_quote.gif you aren't creating binaural sound. You're creating mono sound

Do you get off being argumentative or what?
Yeah, he isn't really creating binaural... he said so himself. But he isn't really creating mono either. You wanna defend the actual meaning of binaural, fine but don't turn around and use mono loosely!

You've got to reckon that there are many recordings with precious little soundstage to lose and even less natural soundstage (as opposed to mixing tricks).
The way I see it, when you apply processing like that to many studio recordings, you're remixing something artifical to be more pleasant on headphones than on speakers. It's not actual soundstage but it may well sound better than hard pans, shrill highs, having everything in your face and so on.

 

[musicmind]

hi Alleyman

Thanks for the link.

I gave that a try, and while I was impressed with the "out of head" effect it acheives, the reverb is just too much and makes vocals sound very nasal and metallic to me. Instruments also moved to completely different areas of the soundstage. Perhaps with different left and right config files, this can be improved, but so far there dont seem to be any other alternatives to choose from.

Cheers

Quote:

Originally Posted by Alleyman /img/forum/go_quote.gif Thanks Artemus_GF. I figured it would be like so. If you are not already tired of this I have one more suggestion. Foobar also has a stereo convolve plugin that was developed specifically for HRTF modelling. I will keep my blabbering to minimum and link you to the thread discussing the plugin. New plugin (05/18/07): Stereo Convolver 1.0.1 - Hydrogenaudio Forums Thus if your DSP chain allows you to process left and right channels separately, you might be able to simulate it using stereo convolution (or not). What is there to lose huh?! I wasnt sure of what kind of impulse file it would need to the get the desired response so I didnt suggest it earlier. In any case, can you write a guide as to how you got the effect to work. I see the screenshots in your other thread but you didnt mention what VST plugins you used. If I can replicate your settings on my PC I can try do my own impulse response modelling.