posted 20 April 2002 12:49            

circular_panning.kym

This sound file is a crude first cut at incorporating all of these effects. The amplitude is modulated front to back to mimmick head shading effects. The ITD is modulated according to sound position, being a maximum differential delay when full right or full left. The spectral modifications are performed on the left and right channels to provide a 6 dB notch at 8 KHz when the sound is in front, 0 dB when at +/- 90 degrees, and 6 dB notch at 4 KHz when the sound is in back.

There is an lfo to provide continuous circular panning for this example, and a noise source to furnish a nice broad spectrum on top of a Kyma granular resynthesis of some speech.

When I close my eyes, listening through headphones, it appears quite realistic. But then, I originally had the spectral shaping backwards and it also sounded quite realistic. So just how important is the spectral shaping? Or do I need to deepen the notching and make it more narrow?

posted 13 August 2003 08:32            

QuadHeadphoneMix.kym

I hope you don't mind, but I've been fiddling with with your circular panning sound. Instead of panning the sounds, I thought I would have a go at placing them in certain locations within your stereo headphone mix, for example as speakers might be set up for a quad mix. This sound therefore is my first attempt at trying to recreate a quad mix for headphone listening (a la Lake Technology TheaterPhone HSM6240). The result actually isn't too successful - it sounds a little flat to me; however some of the circular panning does appear more realistic than, say, just listening to a stereo version of the timeline.

posted 17 August 2003 13:44            

No, I don't mind people taking my sounds and modifying them for their own needs! By all means, that's why I post them here!!

I looked at your Sound and it appears that you are using the Graphic EQ's to spectrally distribute the channels to each ear. Is that all that Lake Tech does? Personally, splitting the 4 and 8 KHz channels would have little effect with my own hearing, since I don't hear very well at those high frequencies without a lot of assistance.

But it would seem that if you want spatialization then you might have a look at the MIT Head Related Transfer Functions (freely available). These HRTF's can be used in the FIR Sound to produce what should be apparent for sounds arriving from different directions. Filters based on these HRTF's attempt to model the effects of the ear's pinnae on sounds arriving from different directions. (Perhaps that's what your GEQ's were attempting to do?)

- DM

[Whoaa! I just looked up the HSM6240... $2200!! Yikes!
But the little I could find about what they are doing makes it appear that they are indeed using HRTF processing to simulate sound arrival from different directions.

One of the things I found when doing my own HRTF research a few months ago was that using HRTF directed filtering to make sounds appear to come from in front or in back of you often times really sound like they are coming from overhead -- depending a lot on the individual listener. Further research reveals other researchers have written that this is a common problem. At best the HRTF front/back listening only makes the sound appear to come from somewhere on the medial plane (midway between left and right). In my case they appear overhead, but other listeners have said my sounds appear to be coming from the front but elevated off the horizontal...

At any rate, HRTF processing is the way to proceed with spatialization in headphones. I suspected as much about Lake Tech, since I have studied them in the past and they are mainly convolution freaks (big FIR filters). Seems a bit like a waste of a good DSP to commit so many of them to doing just this menial task, but oh well... ]