On 2017-06-27, Augustine Leudar wrote:
I know I've asked this before but maybe there's some new developments. HAs anyone any suggestions for anything up to a permanent 22 channel installation (could be two devices started at the same time and set to loop). [...]
Why not just have a couple of commodity PC's running the highest channel count external D/A converters you can muster within your budget? Given the existence of 7.1 home cinema, lots of implicitly synchronized converters already exist, at reasonable cost. Given the existence of newer USB, Firewire, HDMI and even Ethernet based transports, reliably feeding such multibank converters shouldn't be a problem.
The only problem is how to keep up synchronization between the converters. But then that ought to be a problem which is soluble by feedback and software.
First, even low-range modern converters keep pretty good time all by themselves. They don't drift too much, compared to what we can hear in spatial reproduction; compared to accidentally moving the listeners' heads ten centimetres or so in aggregate, even an average free running commercial converter will keep adequate time over minutes or tens of minutes by comparison.
And second, if you really want to make sure your separate converters run in time with each other, it's possible to insert an inaudible timing reference into the signal set which lets you do continuous feedback correction. If this sounds alarming, it should: lots of papers and trials exist which purported to make inaudible changes to the program material, yet proved to degrade playback. Especially on the digital watermarking side the results have been dismal.
But then, here the application would be rather different. First, unlike in watermarking, there would be no requirement for the added signal to be highly resilient. Instead it could be optimized to be highly inaudible. That means that if it was only added to aid in synchronization, it would need a very narrow bandwidth, amplitude and the resulting extremely low data rate, so that it could be buried *well* below even the perceptual noisefloor of an existing recording.
And secondly, it would only be present upon playback. It wouldn't need to be buried into the original program material. It could be optimized further, e.g. on psychoacoustical grounds, over just the one playback system, or switched off at will if it ever somehow annoyed any listener. It could also be different for each playback system, and each reproduction instance, so that the human hearing system's notorious capability of learning to recognize even low level noise signals, if repeated many enough times verbatim, could still be subverted.
As I said, I don't think such solutions exist as of now. But I also think the theory behind them is well developed enough to make their implementation for your use almost trivial. All we'd need is a) an inaudible reference signal to provide us with a relative delay reference (easily doable via MLS sequences or the like), 2) a self-acquiring servo loop to drive a set of relative delay estimates to zero (a simple exercise in first order control theory, with the driving measurement being derived from an FFT implemented autocorrelation measure), and 3) a high grade, capable of continuous variability delay resampler to be driven by such a measurement-correction loop (implementable utilizing part of said FFT machinery, or separately if e.g. oversampling, minimum phase characteristics or something such is required).
I'll attach Olli Niemitalo here, because he might be even more well versed than I am in this sort of thing, and certainly is the more adventuresome+productive of us two, in the signal processing department.
Must be bomb/cleaner/child/adult proof,
Tape one gigabit ethernet cable into the floor, leading to your converter bank in the next combustion compartment. Software gang it up with a wifi route leading to the same destination.
The only way it's proofer to the end of the world, or your kids, is a setup where you can't actually route back the signal to be heard, in the first place. ;)
-- Sampo Syreeni, aka decoy - de...@iki.fi, http://decoy.iki.fi/front +358-40-3255353, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2 _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit account or options, view archives and so on.
