Trying to re-create ambiophonics with convolution

2016-03-01 20:12:15

For those that don't know what ambiophonics is you can read about it here: http://www.ambiophonics.org/

It's basically a way to extend the soundstage to 180° and reduce intermodulation distortion between the ears and the speakers. It takes some tweaking, but when setup right it is amazing. One of the biggest improvements to my system that I ever heard. I can only high encourage you to try it out.

I used the fb2k VST wrapper together with the  ambio.one VST for testing.

Now, I would like to have a setup that works will all applications, regardless of VST capability. Since I run EqualizerAPO anyway and it can do convolution since version 1.0, I thought that could be a way to get system-wide ambiophonics.

I tried the Unitpulse2K.wav file from misc_impulses.zip and applied the VST effect to it with the same settings as used in foobar. I also resampled the impulse to the same frequency as my soundcards set default (48khz).

Problem is, this works somewhat in that it sounds pretty much identical but I loose the ambiophonics effect of the soundstage extending way beyond the speakers. The 'magic' is gone. I have some test-tracks that allow easy verification and I'm sure I'm not imagining things. For example Wumpscut's Bunkertor 7 has a tank in the beginning that is panned extremely to the left and right.

Next, tried creating impulse responses myself using rePhase but this gave the same result. I also tried to use foo_convolve instead of EqualizerAPO to rule out any issues with the software, but the result is the same.

So now I'm wondering if I'm doing something wrong, or if maybe what I want cannot be achieved by convolution?

Any help would be greately appreciated.

Re: Trying to re-create ambiophonics with convolution

Reply #1 – 2016-03-03 14:50:42

I havn't heard about ambiophinics. But I just following the link of yours to this paper
http://www.ambiophonics.org/papers/AES136IRIS.pdf

and it seems the processing is nonlinear. There is some dynamic parameter estimation going on that affects the processing. So, it's not possible to represent this as a convolution.

Re: Trying to re-create ambiophonics with convolution

Reply #2 – 2016-03-03 16:57:07

Yeah I'm coming to the same conclusion. Since audio data has to be copied between left and right channels and the algorithm is recursive, I think a normal convolver cannot do this.

I found a solution to apply this system-wide by combining Virtual Audio Cable, ASIO4ALL and SAVIHost. I basically have a virtual soundcard now that pipes all sound into my VST host which then outputs to my real card. It's great.

Re: Trying to re-create ambiophonics with convolution

Reply #3 – 2016-09-05 04:44:57

Sorry to revive this old topic. I am using Ambiophonics playback for a long time. My preferred VST is http://www.filmaker.com/products.htm which works perfectly with JRiver.  You can add convolution separately which I have not tried as my room is too small to add extra speakers for the convolution. You can hear the setup here . They were recorded using binaural microphone and meant to be listen with your headphones.  The recorded volume level was 80dB and peaks reaching 100dB.  The learning curve is rather steep but once you get it right you would never go back to stereophonics setup.