New version of SIR out! But...

[Torben]

wait... so there is only one single way to code something that analyzes blocks, and only one way to code something that does time based convolution?
or did they (ms and lake industries) patend the whole idea of doing it, regardless of the method?

The way you code it is irrelevant for the patent, else it would be too easy to get around. What is patented is the combination of time based and frequency / spectral based convolution. As you can not make a spectral analyse on a few samples you need to use time based convolution to get zero latency, and this is patented by Lake (the combination that is).

Torben

[greendoor]

I would be more interested in resynthesised reverb. An algorithm reverb that uses patches like a synth. A standalone app could analyse sampled impulses and create patches. I would hate to see this idea patented and held to ransom.

[Alan]

the voxengo one has zero latency - doesn't it ?

Not really sure,but I can use it at at 2.9ms realistically which is fine.If SIR can at least perform as good as Pristine Space in that dept. that would be more than good enough,but 7ms lowest and realistically 11 or so is a bit much.

[Torben]

It's not a problem to have a low latency, you can just process small blocks. But that will eat the whole processing power. What you can do is to use non uniform block processing which will give you a low latency and reasonable low CPU usage. But that I'm afraid is patented

Torben

[Scr1pt3r]

It's not a problem to have a low latency, you can just process small blocks. But that will eat the whole processing power. What you can do is to use non uniform block processing which will give you a low latency and reasonable low CPU usage. But that I'm afraid is patented

Torben

Torben, what are the laws regarding patents ?
Are patents valid forever or are they valid for a certain time after which everybody can use them ?

[Pollux]

Oh crap, I was about to post a topic on this forum just to check if there weren't any free zero-latency convolution because I wanted to make an open-source implementation, only to find that the algorithm is patented

Basically what I intended to do is split the impulse into increasing size blocks (one 1-sample block (x[0]), one 2-sample block (x[1],x[2]), one 4-sample (x[3],x[4],x[5],x[6]), etc...), then convolve each 2^n-sample block with the last (2^(n+1)-1) samples to get a block of 2^n convolved samples, then I would sum up all the blocks to get the correct sample. Is it what the patent is all about?!? I mean, I figured it out myself without reading anything whatsoever about this particular algorithm, it's all but rocket science... You could even do it by hand, by splitting your waveform in 2^n-sized blocks and having several instances of the patent-free plugin, each having a latency equal to the size of the block minus one... Would they sue you for doing THIS? Would they sue you for having a patent-free plugin that allows you to load several blocks and sum them up? Would they sue you for having an impulse splitter that would generate the 2^n-sample blocks for use in this plugin?

And from a juridical standpoint, is it only a problem if you're living in the US? (I don't ) I mean, could I have problems distributing such a program?

[Torben]

Torben, what are the laws regarding patents ?
Are patents valid forever or are they valid for a certain time after which everybody can use them ?

Yes, you got them in 20-40 years I think and then they are free to use. The Lake patent is from 1996 so it takes a while before you can use it.

Oh crap, I was about to post a topic on this forum just to check if there weren't any free zero-latency convolution because I wanted to make an open-source implementation, only to find that the algorithm is patented

Basically what I intended to do is split the impulse into increasing size blocks (one 1-sample block (x[0]), one 2-sample block (x[1],x[2]), one 4-sample (x[3],x[4],x[5],x[6]), etc...), then convolve each 2^n-sample block with the last (2^(n+1)-1) samples to get a block of 2^n convolved samples, then I would sum up all the blocks to get the correct sample. Is it what the patent is all about?!? I mean, I figured it out myself without reading anything whatsoever about this particular algorithm, it's all but rocket science... You could even do it by hand, by splitting your waveform in 2^n-sized blocks and having several instances of the patent-free plugin, each having a latency equal to the size of the block minus one... Would they sue you for doing THIS? Would they sue you for having a patent-free plugin that allows you to load several blocks and sum them up? Would they sue you for having an impulse splitter that would generate the 2^n-sample blocks for use in this plugin?

Look here, I think they have covered it all: http://www.patentstorm.us/patents/5502747.html

And from a juridical standpoint, is it only a problem if you're living in the US? (I don't ) I mean, could I have problems distributing such a program?

Yes, If you sell or distribute in a country where it is patented and you are not hiding on the moon

Torben

[Z3R0T0N1N]

i say go for it. if yo are in the eu than you still have time to get it out as opensource. the us patent doesn't cover you. Anyway, you said you came up with one on your own, so write it. PLEASE, write it. At least you could get the jump on it being patented elswhere. (of course, they might have already done it in asia, there is only a small window of opportunity between disclosure and patent application in the us, and in asia (as of my last course in patent law) it is immediate, so if a company can afford it, they apply to both at the same time.
anyway, just get it out there.
Software patents are really (in my opinion) one of the most serious and immediate threats to human progress. Circumvent them whenever possible, and try not to support companies that hold software patents for basic operations (ie basic image processing, sound and media software, blah blah.)

[Aleksey Vaneev]

'Software patents' are basically 'method patents'. BTW, methods, ideas and algorithms cannot be patented in Russia. But some lobby do exist to make these things patentable. Right now we have computer program registration bureau. This should have been done in both US and EU. Otherwise sane software development may stop because most things in software are based on simple algorithms and methods: if they are patented nobody can use them. Lake convolution patent is totally insane in this respect.

[drinelli]

Though I like the idea of patents, I do not like it when it makes it impossible for new companys/people to further devellop an idea, or if it makes people who don`t want to make money from it suffer because of it. there are actually in a way the same problem with music what if all musical possibilyties are used, in terms of notation, then everybody who make music could be brought to court. If something kills creativity then that does. In the same way Patents could in some way kill science, maybe not now but somewhere in the future.

The Stoneage would be back

[Z3R0T0N1N]

they already do. That is a BIG reason to only use open source software. I use quick swif and J generator for flash (trying anyway... that is my goal.) Nvu for simple websites, Wamp 5 server and stuff for datbase, gimp for bitmap and inkscape for vektor (actually, inkscape has a lot of gui stuff that kicks illustrator in the pouch, IMHO, and gimp will soon be better than potatoshop) I am moving more and more stuff to linux. I use blender and wings for 3d and i do stuff with it that the proprietary stuff at school sucks at. I am working on getting my linux audio stuff worked out, and for that there is Rosegarden and Muse (especially muse, but rosegarden 4 is VERY good. not at all like that win 3.1 looking, weak-ass version 2. ) JACK is wicked, use qjackconnect and qjackctrl for EASY gui manipulation instead of command line. More top of the line open source stuff for linux/unix/bsd than I can shake a stick at, including two very stable and advanced modular synth packages. Three, actually. TerminatorX for the djs. Hydrogen for the FL studio types (although I hve nothing against image line, they arent perpetrators )

I don't want to keep ranting and hijacking, but i feel REALLY strongly about this, as do a lot of people, and getting the word out is the only way I can really contribute right now. So, here's the thing: If you want to help, and make some cash, you can legally sell compiled copies of open source software, perhaps in group packages on CD roms; This is INCOURAGED by the developers. Get a nice package together, tested and stuff, document where necessary, make a pretty cover, and sell it, get it out there. Just make sure to credit the authors and include whatever files they ask you to. People will be more likely to trust the quality if they pay for it anyway, and the more that actually USE it, the better.
That is one way to stop the madness, or at least slow it down.

[Z3R0T0N1N]

PS: there is a TON of opensource software that have versions for Windows; a lot of people don't bother looking because they might feel like it is all for Linux et al, but it isn't. I use all of the graphic stuff I mentioned on my windows installation as well as my Linux installation.
There is also WINE (Wine Is NOT Emulation) which implements the win32 thing on Linux (as the name would suggest, it is not one of those emu's, despit what some tell you - YOU DON"T NEED WINDOWS INSTALLED TO USE WINE) I have run quite a few progs, including Flash on Linux using wine.
cheerz.

[Guckli]

Welcome to the club, Spoonboiler. dssi-vst is your friend.

[kurasu]

Look here, I think they have covered it all: http://www.patentstorm.us/patents/5502747.html

just skimmed it, and yeah, it looks pretty solid..

meh.. fudgepackers..

[Z3R0T0N1N]

yeh.
I hate that.
packing fudge all the time. bastards. as if fudge needs to be packed....

er, what?