DSP Robotics Support

The Quilcom Whirlipan is an effect that takes a mono input and converts it to a stereo signal to create the impression that the source sound is flying around the listener’s head in the horizontal plane. The result is realised on headphones only but on speakers there is still an interesting wide panning effect.

The processing makes use of binaural psychoacoustic phenomena. These are interauditory time difference, head shadowing, pinna shadowing, pinna-based spectrum modification, sound reflection and space reverb. All of these parameters can be adjusted to get a range of results which can be matched to the type of input signal.

The inspiration for this came from tester

viewtopic.php?f=2&t=4415&sid=d7d12013076290af4e01ffc0e2ad8256#p24956

This is my current attempt at synthesising a HRTF using basic modules and fundamental principles. I have more ideas in mind since this is a truly fascinating area.

While developing this I noticed that I was becoming immune to the effect somehow. Initially I listened to many binaural recordings but gradually the degree of effect kinda wore off. I guess my brain was correcting for errors and discrepancies between my own learned location sense and the artificial ones. For this reason I checked out the impression on 2 other people who were both impressed by the Whirlipan. So, you will have to judge for yourselves and please let me know.

If you wish to experiment I would recommend reading my included user guide first. The download incudes some test WAVs, 3 of which tie in with presets 7-9.

It would be great if you modify or enhance my schematic and get better results than me. I’m quite sure some or all functions could be improved. For example I’m not convinced that my transfer functions for the LFO output are ideal and it may be worth making the reflection function more sophisticated. And I bet Martin could come up with a proper HRTF convolution system but that’s beyond my capabilities.

Updated on 24.7.16 to version 1.13 with info in later post:

https://www.dropbox.com/s/k5rwzgq89r7c1 ... m.zip?dl=0

The VST plugin is on http://www.flowstoners.com under Quilcom

Have fun and throw me a morsel of feedback…please!

Cheers

Spogg

Hello Spogg,

Funny I was working on something simular in May, but it was for use with speakers and not ear phones. I ran into some problems with it so I have put it on hold for now. I played with yours for some time, and I do think that it works quite well! I have a good number of old mono stuff and I listened to them at night throgh your Whilipan with headphones. Very cool! I agree with your findings. I owned a SansuI pseudo quad simulator back in the late 60's. It worked great for a while, but like you the more I listened to it the effect got less and less, yet friends over that were hearing it for the first time thought it was great.

Later then, BobF.....

If I may...

1) Do you get "immune" when sound is static or in motion? // (relativisation issues)

2) Eyes open or closed? Sitting/standing or lying down? // (sensory conflict issues)

3) Experimental. Try to to add some random background ("clicks" or something that is in motion or jumps out of different locations) while targeting static sound in certain location // (relativisation issues)

Nice that we have starting point.

p.s.: sorry for late feedback, on some computers I'm outside this place.

tester wrote:If I may...

1) Do you get "immune" when sound is static or in motion? // (relativisation issues)

2) Eyes open or closed? Sitting/standing or lying down? // (sensory conflict issues)

3) Experimental. Try to to add some random background ("clicks" or something that is in motion or jumps out of different locations) while targeting static sound in certain location // (relativisation issues)

Nice that we have starting point.

p.s.: sorry for late feedback, on some computers I'm outside this place.

Thanks for your comments tester.

1) The immunity is more pronounced for static sounds but affects moving ones too. Oddly this has now extended to real sounds to some extent! I've been focussing on this perception so much lately that I can kinda overide the natural illusion and actually hear real sounds inside my head to some extent. I'm sure I'll be back to normal soon.

2) It's better with eyes closed. I test upright seated in my chair at the computer and I keep my head still.

3) I intend to experiment more using multiple instances and add different sounds as you say.

Version 1.1 is nearly ready for sharing. I've made some changes and enhancements and subjectively it's working better but I'll still need some feedback. The main thing is the ITD is now dynamically controlled and based on a graph I found of ITD Vs azimuth.

Cheers

Spogg

a.d.1.

For stationary sounds it's rather normal. Even if you listen to stationary sounds in real life (like waterfal from a distance) - it is you who moves and it is the rest of the sonic space that interferes with that sound. Maybe - just maybe - some sort of gentle (spatial and/or filter centering) modulation would help, not necessarily driven with sine, and perhaps a bit irregular.

For non-stationary sounds you may wish to try 2 tricks.

a) How loud do you listen? Usually you should try to replicate "natural loudness". Or less.

b) The way you listen. This part comes from sound engineering practice and is a bit tricky. You probably switch too much to so called "analytical perception", which is helpful when dealing with mixing/mastering. This is why you start to not recognize spatial location of real sounds. Try to listen to what you hear "as if you were" regular listener.

BTW, you may wish to add to recorded source sounds a little bit of brightness and clarity/details.

I’ve made quite a few improvements to the Quilcom Whirlipan which can be downloaded complete with the other files in the 1st post. Here’s the change list:

- The Intra-ear delay is now variable and based on the azimuth of the virtual sound location so is now a truer Interaural Time Difference (ITD). You can set the maximum delay time which is for when the source is directly in line with the left or right ear. The time reduces dynamically based on the azimuth. This is based on a graph I found of angle versus delay time which fortunately is pretty much linear.

- I’ve replaced the somewhat dodgy pinna filters with a comb filter to better mimic the outer ear and auditory canal’s peak and trough transfer function. You can adjust frequency (delay time) and resonance (feedback). The mix level of this filtered signal can be also be set.

- The pinna filter is now fed with a high pass filter to reduce additive amplitudes at lower frequencies.

- The pinna filter is no longer amplitude modulated.

- The reflection signal is now panned and phase inverted.

- Updated the user guide

- Updated all presets which now make use of the new pinna comb filter.

To my now jaded sense of sound location these improvements add up to a more realistic impression but please let me know what you think about the results.

The new version plugin is on http://www.flowstoners.com under Quilcom.

Cheers

Spogg

Schematic is missing.

tester wrote:Schematic is missing.

Dammit! I uploaded the wrong file.
I've corrected this in the first post and here it is also

https://www.dropbox.com/s/k5rwzgq89r7c1 ... m.zip?dl=0

Thanks tester!

Cheers

Spogg

You're welcome, thanks for effort on such project. I will look at it in a few days, now the weather is a bit difficult here.