Head-Fi.org › Forums › Summit-Fi (High-End Audio) › High-end Audio Forum › Smyth SVS Realiser - PRIR Exchange Thread
New Posts  All Forums:Forum Nav:

Smyth SVS Realiser - PRIR Exchange Thread - Page 21

post #301 of 391

There have been quite a few requests lately to join the DropBox folder.  Another option to keep the files safe an reduce chance of deletion is for folks to access via the following link.  Posted this up top at the beginning of the forum:


UPDATE: For those wishing access to the folder, the following link may be used (and this is the safest way to download files, no risk of deletion).  In order to contribute, just send an email to get an invite into the shared folder which allows for contributing files).   FOLDER LINK HERE

post #302 of 391
Originally Posted by symphonic View Post

I don't have a Realiser but I do like playing around with HRTFs and such.

How many of you guys have made a speaker set up outdoors? Off axis response of the speakers becomes much less important because of the lack of reflections. You would start with a very dry sound but I would suppose that adding some reverb should be easy enough. It would be interesting to try. Bass outdoors requires more power (no room gain) but this is not an issue with headphones.

There are several possible advantages. Of course no one wants to leave their audiophile speakers out in the weather, so listening rooms are indoors. An afternoon is something else, though...

That's a great idea! I have a DEQX preamp that does a speaker analysis outdoors, so I could do both at once. Because I live in Chicago, I may not be able to get it done till Spring.
post #303 of 391

Yes, I should have asked someone to try outdoor measurements in the summertime. One possible benefit of trying in winter, though, is that snow on the ground might attenuate the floor bounce even more. 


Something that might work rather well in this scenario (given a low enough noise floor/plenty of averaged measurements) would be to get your hands on a nice full range driver, EQed and played well within it's SPL limits. Headphones can add dB with ease. You might be able to get very close to an ideal point source as far as measurements go, and then increase volume on playback. 

post #304 of 391

Hi hekeli,

I took some audio probes and analysis from your PRIRs.

The very best for me is 2049.

It's center phase stability (L/R/C) and the sound is perfect in comparison to the other files. The distance of speakers is critical (1,37 m), because in nearfield the influences of BRTF changes dramatically (typ. < 1m).

The quality of the 12s takes decreases unfortunately heavily. Steady head position is necessary during 12 s ...

The 2049 recording shows some slightly abnormality:

LFE is out phase. It's roll off frequency is 80 Hz (-40dB), latency is 0 ms (I mean the true recorded latency in PRIR, not the delay shown in Realiser). No wonder, that changing latency has such effects in center location ...

Positions of surround channels have a stable 4 ms latency.

Surround leveling was done by pink (all channels, but C -6dB) / brownian (LFE -10dB) noise, I think, good tune.

The clarity caused by increasing 4-8 kHz range at +3 dB?

There are two gaps -6/-10 dB in low range at 52 H2 and 85-95 Hz (both shown at Hamming/Bartlett. all speakers).


post #305 of 391
Originally Posted by jlejle View Post

The quality of the 12s takes decreases unfortunately heavily. Steady head position is necessary during 12 s ...


Interesting. I think I was pretty steady and almost all diagnostic values were excellent. I wonder if it would make sense to run even 4 sweeps of 3 sec measurements, instead of few 12 sec ones..


Also I wonder how important is getting the speaker angles correct. Should one use the recommended HT mode for angle measurement or atleast carry a laser angle meter..




There are two gaps -6/-10 dB in low range at 52 H2 and 85-95 Hz (both shown at Hamming/Bartlett. all speakers).


Hmm do these imply nulls in the room?


I played around with Room EQ Wizard, actually it seems quite handy to make some measurements. Here are overlaid all 4 PRIRs I made over there. I just used my Xonar STX out -> Realiser with PRIR loaded (no HPEQ) -> STX in, measuring one channel at a time then moving RCA. There's lots of data, if I only knew how to read it correctly. wink.gif It's nice that you can also save the whole measurement package which includes impulse responses which can be exported etc..




You could probably use the Realiser mics to take normal room measurements (without HRTF) before putting them into ears.. this way you could easily verify how the room sounds.

Edited by hekeli - 11/20/12 at 11:34pm
post #306 of 391

Hi hekeli,


hard work answering in a few words:

Ok, you get better S/N in silent rooms with 12s, but ITD differences lock up better in 4s, much better with dirac impuls.

If more repetitions gives better results ... I think not ... better have a look on Realiser screen and the interaural differences before saving.

The angle may be a more important fact to generate the tracker movement signals. The PRIR recording is a "as if" room response. There are anyway differences between recordings of the individual channels doing in sequence. Keep same headposition (x,y,z) in mind.


If recording of Realiser is perfect in any case ... I don't no the base of detection procedure of ITD, phase problems of speakers may be a reason ...

The gaps: yes, I think the room is the causer, but it's only a assumption.

To understand BRTF, look into homepages of http://www.irt.de (multilingual) or http://www.hauptmikrofon.de (multilingual), http://www.acoustics.hut.fi/ (it's a link from  Institut für Technische Akustik (ITA) - RWTH Aachen)

Quick points (works for me):

< 300 Hz interaural level differences are irrelevant
> 300 Hz sound diffraction takes place, interaural level differences are critical, 30° = 0,26 ms ITD!
ca. 5 kHz pinna and ear reflections have maximum (overall frequency range)
reflection gaps at 1,5 kHz shoulder, 3 kHz earchannel, 8 kHz pinna, 12 kHz earchannel (individual varations)
note: virtual center channel has comb filter effects, caused by against each other delayed sonic dues of L/R channels, analysis is difficulty.

Good idea to record the mics, but using constant distance of perhaps 17-18 cm, like a head in A-B (mic's are pressure receiver). A second record with a normalised dummy or a turnable Jecklin float OSIS 321 would be very interesting ...



post #307 of 391

Ok I'm obsessed now to get perfect results.. going to visit a local studio which has a fully calibrated Genelec 3x1037/2x1032 set hopefully free of pesky nulls etc.


I will take MASSIVE amounts of 3 sec PRIRs and record all diagnostic screens with camera, so I can analyze everything later at home without rush. biggrin.gif


(Why aren't the diagnostics saved along with the PRIR? Or maybe they are and just hidden somewhere in the data..)

Edited by hekeli - 11/22/12 at 4:29am
post #308 of 391

Hi hekeli,

I appreciate your "obsession" (please don't carry it too far), it's the right way to get best results - one chance to record , spent much money - future disappointment prevented... good luck!

A word to the S/N data of your PRIRs: the 12 s take 2047 has -82dB (L), the 4 s takes 2046/2049 have  -91/-92 dB (L) !!! appraise yourself!


2048 won't work, I checked out hash, it's value is ok ...?

ITD: yes, pics are a good idea. I think record data is saved separately, because it's possible to eliminated the dues of a millisecond by menu/SVS/delay + CLR/REF global delay function (sweep spot). This function don't affect the ITD differences for each channel, only the first wavefront latency.

In brackets: good question, ask the doctors ... Symth peoples empire.


post #309 of 391

FYI, have any of you ever used the "Taper" function?


Lorr mentioned the Taper function and the reverb/window function too, but I never really played with it much.


Anyway, you can check page 89 of the manual.


Lorr had also mentioned that my own speaker system, the Cello system, in my super bright room was a perfect candidate for playing with the Taper function.


Basically you can "taper" the impulse responses in the PRIR which effectively "tapers" the room reverberations. The net result is that the "brightness" of my room can be reduced significantly making the Realiser output sound better than my actual room! By tapering the impulse response, the timbre and localization of the speakers is unaffected.


I have to say, it's not perfect, but it is interesting to take my really bright room and taper it all the way down from slightly reduced to virtually anechoic. (When tapering down, I still think there are some artifacts since the original PRIR has a lot of reverb and simply cutting or tapering off that info doesn't make it exactly the same as if I had the speakers in a much more damped room.)


Previously, my Cello PRIR was of limited use. It worked well on music, but was way to reverberant to work well with movies or any surround sources. Now with it tapered, it sounds really good. I love the way the speakers sound, it was just the room that was causing problems.


Anyway, if you are interested, download/load my Cello PRIR, and you the Taper function when loading the PRIR. I tried the following values: "S:100ms E:150ms" with pretty good results.


Here's the excerpt from the manual:


For PRIRs made in reverberant rooms, the user may wish to alter the reverberation decay. Upon
loading a PRIR into a preset, the Realiser allows the user to set the start and end points of a taper for
this purpose.

When a PRIR is to be loaded into a preset, the user presses MENU and then the target preset key. As
an example, for preset 2 the user would press MENU-P2, and then choose PRIR PRESET 2. The
screen will say:

  S:800ms E:850ms

Move the cursor to the second line. The < key will increment the start point of the taper by 50 ms, and
the > key will increment the end point by 50 ms. The default of 800 and 850 represents a 50 ms taper
on the full 850 ms capacity of the Realiser. The start and end points can be adjusted separately to
create tapers longer than 50 ms; the minimum taper is 50 ms. The controls are independent except that
the 50 ms minimum may cause the two values to move together, and the start point limits the end point

When a taper has been set, the preset window will show an indication as follows.

1   john doe
    studio a
A   20:01 10-OCT-08
  . .TPR.   .   .

The taper control is somewhat similar to the REVERB control on the MENU-ALL screen and the
WINDOW control in the virtual speaker setup menu 2. The differences are: (1) the taper is just that, a
linear taper providing a fade to zero, whereas the REVERB/WINDOW control truncates the reverberation
tail; and (2) the taper must be set upon loading the PRIR into the preset, whereas the
REVERB/WINDOW control can be varied while listening to a preset. One can initially adjust
REVERB/WINDOW to see if reverb reduction is desired; then use either control for the desired result.

Let me know what you think.


(BTW, Kiritz, you might find this especially useful since you have your PRIR done at my house.)

post #310 of 391

One of the R& D guys at Acoustic Zen, Alex, e-mailed me yesterday and asked me if I wanted to come measure his new prototype system. Umm, yes!


I uploaded the PRIR I made today. It's a very interesting system. You can see the photos in the DropBox folder, "darinf/Acoustic Zen Alex Ribbon".


Here's the system description file:

Acoustic Zen Alex Ribbon System - http://www.acousticzen.com
PRIR MEasurement done on November 23, 2012 by Darin Fong

2.0 Channel System:

This system uses prototype ribbon speakers and a subwoofer.

The most interesting thing about this system is that one of the R&D guys at Acoustic Zen named Alex, has been working on an audio processor that does EQ and room corrections. This processor is a prototype with a DSP processor that Alex has programmed himself. He would not divulge many details other than the fact that it does use microphones to measure the room.

So this system is not about the speakers or electronics other than the prototype processor.

Unfortunately he was not able to A/B the processor on and off so I could hear how much the processor was doing, but I thought the system sounded incredible.

As far as the speakers go, they are long ribbon drivers. Alex didn't even really know what drivers they were. I asked him and he said, "I don't know. I think they are from Taiwan or something."

Then he pointed out that the source was a 10 year old CD player that was being fed digitally to his processor.

The amplifier being used was a low end 12 channel home theater amp. I forgot the name of it, but he was only using two of the 12 channels.

You can see the subwoofer in the photos. I don't know what brand it was either, but it looked like an oversized ottoman. It had a down firing woofer.

The point being that his processor will compensate for sub-par sources and electronics and make them sound world class. (Do you believe it?)

He did say that the ribbon driver was key in allowing him to apply the processing accurately, much the same as Smyth recommending Stax for headphones since they are very detailed and fast.

Sorry about not having more details about the gear used.

I generated a 7.1 PRIR from their 2.0 channel system.




post #311 of 391
Hi Darin,

Some words to the tap function: the characteristic of the cutted delay tail is not linear, perhaps logarithmical. The fade out time to noise ground differs from value setting ca. +30 ms (I don't known about purpose, perhaps RT60 reverberation time calculation). Initial delay will be considered.

In real life reverb delay is a linear function ... see original Cello 3.1 file. Images were made by envelope curve function of Wavelab.

It could have a benefit, if the tap has a subtler value resolution, like 1 ms or 10 ms. In the important range of 100 - 300 ms 50 ms tap is not really usable, just exactly described in manual (soft windowing).

Your new PRIR from Alex:

Phase alignment and 5.0 localisation are perfect, but frequency response is not optimal, likewise room acoustics.

There are some large gaps at 110 Hz, 380 Hz and 650 Hz, the room sounds poor, some compact absorbers in height of head may kill the problem.

Decrease in bass range see spectrum plot.

A-B Audio probes (Acoustic Zen v2, hekeli siba) with cello, drums, bass and speech shown rasp tonality and a sligtly booming basement.

The sound is lean and rough but also balanced.

The virtual LFE can be used as back center in 6.0 configurations, very good localisation.

LS and RS have high damping ...?

Apparently a good construction to record a "PRIR for all", like eletrostatic systems, but at lot of work to a running system.

post #312 of 391





Let me know what you think.


(BTW, Kiritz, you might find this especially useful since you have your PRIR done at my house.)

Hey Darin,


Indeed, the taper function greatly improves my PRIR of your cello system.  I tried redoing it using your settings and got rid of most if not all the reverb, revealing a nice sound.  The Zen Ribbon system PRIR is good, however it is a bit harsh sounding to my ears, especially up high.  I may play around with it a bit, but suspect that the issue is that I am spoiled, now that I have Acoustic Zen and Magico PRIRs made to my own ears.  I find I am gravitating more to the Magico system, which seems to have more "heft," in the upper bass and a somewhat more realistic sound to my ears.  I still like the Acoustic Zen very much.  It has a wider and deeper sound stage.  Both of these genuine personalizations are much better for me than ones made by others.  However, the latter are really interesting to try out.


Take care...

post #313 of 391

Alright uploaded the new one. This is more near-field experience than Siba.


Hekeli/Fantom 5.0 (Genelec 1037C,1032A)



Managed to make 6 prirs in an hour. It's always 15-20 minutes to prepare everything, also the mandatory small talk. wink.gif


Realiser reported a bit too high level, but I could not touch the speakers to adjust (direct Realiser -> Genelec DI8A RCA->XLR balancer -> Speakers). But I think I heard 4 chirps so it's probably ok (manual says at least two should be heard).


Now for the tedious task of taking 30 channel measurements..

post #314 of 391

I made FR graphs from several places in Hekeli/_Graphs directory.. I don't know if they tell anything in reality, but it's always fun to look. biggrin.gif You can also see some HRTF differences (or maybe just peoples heads were in different measurement positions). Graphs from my own PRIRs are average of all 4-6 takes I've made, buy they are actually quite consistent even separately. I've adjusted channel/prir levels so they overlay nicely (dB is meaningless).


Seems Fantom left channel has a nasty drop at ~70-100hz.  I actually noticed this during the sweeps but the graph really confirms it. rolleyes.gif


The few drops in Siba doesn't look that bad anymore compared to other stuff. Also the response overall looks very flat. Probably still my favourite including the airy sound. Also the bass extension is impressive with 8260A. biggrin.gif

Edited by hekeli - 11/29/12 at 1:00am
post #315 of 391

Does anyone have an opportunity to do a PRIR on these?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: High-end Audio Forum
Head-Fi.org › Forums › Summit-Fi (High-End Audio) › High-end Audio Forum › Smyth SVS Realiser - PRIR Exchange Thread