Boundaries of musical DSP
-
- KVRian
- Topic Starter
- 1096 posts since 28 May, 2010 from Finland
I've been thinking lately.
Do you think there are boundaries for musical DSP?
The question stems from the fact that to me it seems like most of the stuff has been done already and that musical signal processing has some foundational limits that are caused by signals and musicality. That is, there's only a certain amount of musical DSP that can be done.
Or do you think the field still evolves?
Have you found any interesting "new steps" taken in the field that open to a totally new kind, before unseen musical DSP (an example would be how FFT-based processing seemed, when first plug-ins of that kind started to appear. Or e.g. Melodyne)?
Perhaps physical modelling?
I don't think UI design and that kind of stuff matters (it matters for the users), but it's not really the point of musical DSP. Eventhough that (GUI design) is surely making advances.
Do you think there are boundaries for musical DSP?
The question stems from the fact that to me it seems like most of the stuff has been done already and that musical signal processing has some foundational limits that are caused by signals and musicality. That is, there's only a certain amount of musical DSP that can be done.
Or do you think the field still evolves?
Have you found any interesting "new steps" taken in the field that open to a totally new kind, before unseen musical DSP (an example would be how FFT-based processing seemed, when first plug-ins of that kind started to appear. Or e.g. Melodyne)?
Perhaps physical modelling?
I don't think UI design and that kind of stuff matters (it matters for the users), but it's not really the point of musical DSP. Eventhough that (GUI design) is surely making advances.
Last edited by soundmodel on Wed Jul 29, 2015 7:24 pm, edited 1 time in total.
-
- KVRian
- 1379 posts since 26 Apr, 2004 from UK
I think the field can always evolve. There are not that many things that can be done in real-time with a decent latency, that's for sure, but there are many things that can still be explored, and not with FFT (which I usually hate). Just working in other dual spaces could be interesting, extracting and processing different kind of features...
-
- KVRian
- Topic Starter
- 1096 posts since 28 May, 2010 from Finland
Well one could say that if mathematics can evolve, then so can most other fields that depend on it.
- KVRAF
- 9576 posts since 16 Dec, 2002
Hardware restraints and developer knowledge are the main bottlenecks.
There's still plenty you can do with an old kyma system that isn't available\possible as native software on a desktop OS
There's still plenty you can do with an old kyma system that isn't available\possible as native software on a desktop OS
Amazon: why not use an alternative
-
- KVRian
- Topic Starter
- 1096 posts since 28 May, 2010 from Finland
I was playing with the idea of making a desktop Kyma the other day. Since the PC could soon be fast enough, unless it's some low level stuff e.g. on Windows that makes real-time desktop Kyma impossible.VariKusBrainZ wrote:Hardware restraints and developer knowledge are the main bottlenecks.
There's still plenty you can do with an old kyma system that isn't available\possible as native software on a desktop OS
It would be a huge project and it would be cool if it was open source and relied on open software, rather than closed source.
Kyma truly has some stuff that doesn't exist on the desktop. So there's one project. Maybe some day.
Last edited by soundmodel on Wed Jul 29, 2015 9:02 pm, edited 1 time in total.
-
- KVRAF
- 3080 posts since 17 Apr, 2005 from S.E. TN
It is disappointing that the rules of physics or math seem to mitigate against "ideal" things. Such as ideal brickwall filters with excellent time performance and zero phase shift or whatever.
I know less than nothing about it. Just seems that the universe allows many things but not all. Caught between heisenberg uncertainty and the laws of thermodynamics. As far as I can understand it (very little), thermodynamics and information theory seem to work out to nearly the same thing. Maybe exactly the same thing. Have read articles claiming that thermodynamics can be derived from information theory. Am not qualified to have an opinion.
If it is not wise to try to fool Mother Nature, then perhaps surely foolish to defy Father Entropy!
Nonrealtime processing makes some things more possible. Breaking causality somewhat. Perhaps modify the music happening NOW according to what will happen 1 minute in the future or whatever.
Maybe this has already happened but I never heard about it-- The first fella to make excellent effective software that reliably and convincingly makes the average wannabe vocalist sound like elvis-- That guy will get rich. Of course such software would in principle also be able to turn the average dweeb into any famous (or infamous) vocalist. However, if the box can only do elvis then it will sell into the millions.
I know less than nothing about it. Just seems that the universe allows many things but not all. Caught between heisenberg uncertainty and the laws of thermodynamics. As far as I can understand it (very little), thermodynamics and information theory seem to work out to nearly the same thing. Maybe exactly the same thing. Have read articles claiming that thermodynamics can be derived from information theory. Am not qualified to have an opinion.
If it is not wise to try to fool Mother Nature, then perhaps surely foolish to defy Father Entropy!
Nonrealtime processing makes some things more possible. Breaking causality somewhat. Perhaps modify the music happening NOW according to what will happen 1 minute in the future or whatever.
Maybe this has already happened but I never heard about it-- The first fella to make excellent effective software that reliably and convincingly makes the average wannabe vocalist sound like elvis-- That guy will get rich. Of course such software would in principle also be able to turn the average dweeb into any famous (or infamous) vocalist. However, if the box can only do elvis then it will sell into the millions.
-
- KVRian
- 1265 posts since 9 Sep, 2005 from Oulu, Finland
What evidence do you have that it's because of Kyma's hardware capabilities instead of the software they have? To me Kyma looks like an underpowered and overpriced thing that happens to have software written for it that developers on the desktop PC side haven't bothered developing yet. This is just the all too familiar "the hardware works as the copy protection device" thing going on with Kyma, IMHO.VariKusBrainZ wrote:Hardware restraints and developer knowledge are the main bottlenecks.
There's still plenty you can do with an old kyma system that isn't available\possible as native software on a desktop OS
Did you know by the way that some things in Kyma also rely on long offline processings that have to be done before you can hear anything...? That is a thing that isn't widely acknowledged.
-
- KVRian
- Topic Starter
- 1096 posts since 28 May, 2010 from Finland
Some sources claimed that a desktop OS such as Windows has some internal latency that makes Kyma-like real-time performance impossible. Probably why OS X generally has lower latency for real-time audio as well.Xenakios wrote: What evidence do you have that it's because of Kyma's hardware capabilities instead of the software they have? To me Kyma looks like an underpowered and overpriced thing that happens to have software written for it that developers on the desktop PC side haven't bothered developing yet. This is just the all too familiar "the hardware works as the copy protection device" thing going on with Kyma, IMHO.
But it doesn't seem reasonable that the Kyma hardware would be special. Now that it's Intel even and not Motorola DSP (which Digidesign used as well).
I'd really like to work on an open source desktop Kyma-clone. I guess Composer's Desktop Project comes kind of close, but it's an ancient software and closed source.
- KVRAF
- 9576 posts since 16 Dec, 2002
It's the only platform I have that I know is unique and the developers are experts in their fieldXenakios wrote:What evidence do you have that it's because of Kyma's hardware capabilities instead of the software they have? To me Kyma looks like an underpowered and overpriced thing that happens to have software written for it that developers on the desktop PC side haven't bothered developing yet. This is just the all too familiar "the hardware works as the copy protection device" thing going on with Kyma, IMHO.VariKusBrainZ wrote:Hardware restraints and developer knowledge are the main bottlenecks.
There's still plenty you can do with an old kyma system that isn't available\possible as native software on a desktop OS
Did you know by the way that some things in Kyma also rely on long offline processings that have to be done before you can hear anything...? That is a thing that isn't widely acknowledged.
Amazon: why not use an alternative
-
- KVRian
- 1265 posts since 9 Sep, 2005 from Oulu, Finland
It may be unique or not and the developers may be experts or not, but the Kyma hardware itself is hardly anything special. It's just a way for the company to ensure they get cash from every user of Kyma while they almost certainly could make everything run on Windows and OS-X without any specialty hardware involved. (Unless they have programmed their stuff so stupidly it just isn't portable...But of course we will never know since they prefer the cash that comes from the hardware.)VariKusBrainZ wrote:
It's the only platform I have that I know is unique and the developers are experts in their field
-
- KVRian
- 1265 posts since 9 Sep, 2005 from Oulu, Finland
Weird...How the hell then did I just last night compile the CDP programs on Linux from source code...?Fluky wrote:
I guess Composer's Desktop Project comes kind of close, but it's an ancient software and closed source.
So, CDP has been open source and free of charge since last year. But I have to warn you that the source code isn't exactly nice to look at...However, if you are interested, maybe we should work together to try to see what could be done with the CDP source code to make the CDP programs work better in a modern context? (For example many of the CDP processes could easily run in realtime, but the code isn't well designed to work in a callback manner like VST plugins or ASIO work...)
The new free of charge binaries for Windows and OS-X as well as the source codes can be found from :
http://www.unstablesound.net/cdp.html
Last edited by Xenakios on Thu Jul 30, 2015 2:31 pm, edited 1 time in total.
-
- KVRian
- Topic Starter
- 1096 posts since 28 May, 2010 from Finland
Hmm. That could be a very fruitful open source project even. I guess the users of CDP would definitely prefer if the GUI was modernized and maybe some stuff made real-time and it could even interface with VST, OSC or something. The whole project might become more approachable by a bigger audience than some group of esoteric musicians. It has some cool effects, but using the current old software is a bit too involved.Xenakios wrote:Weird...How the hell then did I just last night compile the CDP programs on Linux from source code...?Fluky wrote:
I guess Composer's Desktop Project comes kind of close, but it's an ancient software and closed source.
So, CDP has been open source and free of charge since last year. But I have to warn you that the source code isn't exactly nice to look at...However, if you are interested, maybe we should work together to try to see what could be done with the CDP source code to make the CDP programs work better in a modern context? (For example many of the CDP processes could easily run in realtime, but the code isn't well designed to work in a callback manner like VST plugins or ASIO work...)
But I guess it would be simpler if the "framework" was modernized somehow as well. E.g. build something that interfaces with the CDP code, kind of a wrapper.
I think another good attempt at "Kyma-likeness" in the desktop is Usine. Which also has an open source SDK (meant for extending Usine though, it's a "module SDK"). Usine might actually be a fruitful platform to build on as well. Eventhough it's also commercial software.
-
- KVRian
- 1265 posts since 9 Sep, 2005 from Oulu, Finland
Unfortunately many changes would need to be done in the CDP source code, likely for all the programs. Most, if not all, of them work so that they read a sound file that is to be processed in a loop that runs until the end of the file or a duration set by the user is reached, while writing to an output sound file. So it is different from how for example VST plugins work where the host asks the plugin to produce a smallish buffer of output at a time. I should maybe look at some simple CDP program to see if it could be made to be driven by callbacks instead of its own processing loop...Fluky wrote:something that interfaces with the CDP code, kind of a wrapper.