On 2012-04-03, Richard Dobson wrote:
Well, we don't need to get hyper-paranoid about it. Apple have defined channel IDs for WXYZ, which goes no further than make it possible to create a 1st-order CAF file.
Agreed. And whatever ambisonic related patents there are for first order, they will have run out by now.
CAF is not closed, the spec is fully open and documented.
On the other hand, Apple hasn't placed any of its coding related software patents into the open domain, here, and CAF is rather new. Most of the technology could be challenged because it's a derivative of EA IFF and then Microsoft RIFF (WAV) derived (even EBU's 64-bit WAV derivative is part of the open, prior art).
But at the same time, Apple put in some streaming related indexing into CAF which is new and not as easily contested. As a pirate and someone who criticises those kinds of patents, I don't think they should have been granted. But at the same tiem, I know they have *been* granted, and I know they are likely to stick even if challenged. (The relevant parts are the ones which hint a real time media server about how to deliver RTP-streams. If you filter them out, you're probably safe until Apple decides to sue you on the trivialities and proven art which should have been safe already.)
It is supported in libsndfile (along with AMB), among other things.
I haven't been following Eric's work as closely as I should have been. Of the two lists I'm on, he's mostly spoken on musicdsp, and not here.
Eric, could you tell us a little bit about the patent status of the CAF implementation within libsndfile? And while we're at it, what would be tha chance of getting some newer, purely open source format into the library, if coded by an outside agency? Just in case?
There is no indication they have any interest in providing an in-house codec for B-Format - which would nevertheless be a strong way to establish it in the "mainstream'.
As usual, I can't be relied upon for anything. But I've narrowed down a certain spherical harmonics toolset as something which could be utilized for further ambisonic work, without worrying about the order, library-wise. It comes with a numerical stability proof right upto order 2800, which is to say "quite enough". Unfortunately it's written in Fortran, but then it compiles with GCC, using portable libraries like FFTW, LAPACK and BLAS, which we'd need in any case.
If I ever get around to finishing the Motherlode, I'm thinking SHTools ( http://shtools.ipgp.fr/ ) and some example code against it would be a terrific addition in the practical, computational front. I mean, obviously having all of the knowledge isn't enough to spread ambisonic around. We do need open API's, libraries, idiot-libraries and all that.
"If you want people to adopt it, you must first make it idiot-proof."
Those who want Ambisonics to become more widely established (aka "mainstream") will need to talk to those who want it to remain a niche process for the cognoscenti.
The latter part is zilch. None of us who have learnt what the technology is about wants it to remain on the sidelines. Sure, it's nice to talk about it within a little circuit, but none of us, and I repeat *none*, want to have to cobble up ad hoc circuits to listen to the sound, none of us have ever purposefully hindered its mainstream adoption, and then *all* of us really just wonder, why-didn't-it-or-how-to-make-it catch fire for real.
No kidding. Ask anybody on-list. While some patent hassles do remain, those have *never* been about overt exploitation of the basic technology. They, too, even as I hate the thing, have been about making a living while developing and promoting the system further.
(Mind you, in my time on the list, I've never *ever* met as many helpful and altruistic folks as here. Even with the development of the first stages of the Motherlode. A number of folks have gone to the length of scanning countless boxes of carefully preserved physical documents. That sort of sustained effort doesn't come from profit-mindedness, but from pure love of the elegance of the sound architecture.)
To do the former will by definition require some company or other to support it and present some de-facto standard implementation.
Today, it might or it might not require that. Nowadays there is the open source circuit as well, you know. It isn't only about a limited number of companies or bureaucratically shelved out government subsidies -- like the National Research and Development Council quango which already burnt the tech once. Now we have other options besides.
If it is pitched on the basis that most of the speakers will just present subtle degree of ambience, which many listeners might not notice at all, any more than they do in the concert hall or rock venue, I suspect its commercial appeal will be negligible.
Have you ever heard what pantophonic ambisonic, decoded from two channels to four speakers, can do? Eero Aro was once kind enough to show me that, and it was downright eerie. Even as the very, very limited BHJ version. The setup was nowhere near perfect, the playback came from analog tape, and so on... Yet stuff seemed to come from the sides and behind me. It stayed there as well, when I turned my head.
I'm a discerning listener myself. I *will* contradict your hunches if the experiment doesn't sound the way you described it. But ambisonic, it sounded even better than I thought it ought to.
I suspect that if Dolby et al, rather than define a single 5.1 surround format, had proposed umpteen options, arbitrary speaker positions, multiple user options for encoding and decoding, etc, the format would very likely not have been taken up at all.
Of course it wouldn't have. The difference is that now every piece of real audio hardware has a signal processor inside it. Now, every piece of hardware *and* software can easily, effortlessly and cheaply adapt to the ambisonic viewpoint. First order, it's no more than 20-30 lines of code.
So why *not* do it, since it's really, really good even on the minimum four speakers?
[...] 5.1 is a shoo-in as there is just the one thing to implement, which everyone will use. Even 7.1 is a problem as there are a whopping two alternative layouts around.
We can do both of those better than the folks who do them now, discretely. I can promise you that even at first order. No kidding either. :)
(We can also do better active decoding. While ambisonic theory mostly goes with passive matrices, we can most easily revert to the kinds of active decodings which BBC tried with Matrix H. They will work better with 5.0 and 7.0 than they would with a rectangular four-channel BHJ decode. We even have some Trifield and Vienna kinda decoding now, to boost that from. Not to mention VBAP/DirAC.)
B-format has so many options and permutations available that the commercial enthusiasm factor will be down to 0.1% or less.
Why don't the commercial manufacturers do what the early ambisonic decoder makers did, and limit the choices to just two: aspect ratio of the (rectangular) rig, and its mean diameter? I mean, it works spectacularly well regardless of the number of speakers, it's intuitive, and it can be easily generalized to non-ambisonic modes of playback as well.
This ain't rocket surgery, you know.
So there is absolutely no danger at all of Apple "locking in" B-Format as it is all but un-lockable.
Not much, but there is some: if theirs is the only widely spread format which carries B-format, and its ancillary online features are held behind a patent wall, then de facto B-format's only viable distribution channel could be owned by Apple. That'd be a real shame.
What you might get, on the other hand, is a hardware-based turnkey system aimed at a very specific market, such as IOSONO or Immsound, where they tell you only the absolute minimum information required to run the system, and it is probably closed beyond the possibility of opening.
Thus, where is our open sourced hardware for ambisonic? We used to have something like that in the analogue age. Where is the counterpart of that for the DSP age? :)
Unless of course they publish a file format for it....
Want a minimal and purposely highly (even overtly) extensible one? That I can design. In fact I've meant to do something like this from teenage up. :)
-- Sampo Syreeni, aka decoy - de...@iki.fi, http://decoy.iki.fi/front +358-50-5756111, 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
