electro-music.com   Dedicated to experimental electro-acoustic
and electronic music
 
    Front Page  |  Radio
 |  Media  |  Forum  |  Wiki  |  Links
Forum with support of Syndicator RSS
 FAQFAQ   CalendarCalendar   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   LinksLinks
 RegisterRegister   ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in  Chat RoomChat Room 
go to the radio page Live at electro-music.com radio 1 Please visit the chat
poster
 Forum index » Instruments and Equipment » Kyma
A few Kyma questions
Post new topic   Reply to topic
Page 2 of 2 [46 Posts]
View unread posts
View new posts in the last week
Mark the topic unread :: View previous topic :: View next topic
Goto page: Previous 1, 2
Author Message
DrJustice



Joined: Sep 13, 2004
Posts: 2114
Location: Morokulien
Audio files: 4

PostPosted: Sun May 10, 2009 8:14 am    Post subject: Reply with quote  Mark this post and the followings unread

Andy, Bill, bphenix , thank you for this very informative exchange!

Kyma/Capy/Paca used to be a bit an enigma. Now I'm starting to get a feel for how it works. I appreciate that both pros and cons are surfacing here. Hopefully this thread is useful to others as well. Please keep the discussion going among you users!

bphenix wrote:

2. How does it handle feedback (also relates to the previous question)?

You write to a buffer and then read it out. As the entire system is evaluated once per sample that sorta informs your minimum buffer length, though in practice it tends to be around 10 samples at minimum (at least on the capy320, not sure about the pacarana as mine is backordered).

Does this mean that you have to explicitly insert a feedback buffer in your patch? How does this fit in with the way objects place themselves in the graph, when you want a feedback path from anywhere to anywhere else? There's got to be a wire looping back into the signal flow somewhere(?)

bphenix wrote:
There are lots of ways to achieve feedback systems. One of the drawbacks of the Capy320 is that many feedback systems were confined to a single DSP chip (in many, though not all cases). I am not sure if that is still true but even if so, the new chips are so much more powerful, it won't likely be an issue.

Having a patch span several DSPs is not trivial from a developers view. I can see that feedback, or even straight signal flow, across DSPs is a problem.

On other points of interest, I note that the hierarchal design facilities are not implemented as a 'real' hierarchal editor and that there are some complexities involved.

Even though I don't expect that many users do it, the ability to go all the way down and write DSP code is good (TBI in Pacarana, I understand). I assume that this is the same facility that Symbolic Sound use themselves - that would make sense. Big plus for openness there.

MIDI Clock and Song Position Pointer would've been nice to have for the timeline. I'm not sure how widespread MIDI Time Code is.

So far, I'd say there is a bit more "Research Institution Flavour" than "MIDI Project Studio Flavour" to it. That's not necessarily a bad thing in itself, just a bit different than what most of us are used to I guess.

DJ
--
Back to top
View user's profile Send private message Visit poster's website
bphenix



Joined: Apr 20, 2009
Posts: 17
Location: austin, tx

PostPosted: Sun May 10, 2009 8:37 am    Post subject: Reply with quote  Mark this post and the followings unread

DrJustice wrote:
Does this mean that you have to explicitly insert a feedback buffer in your patch? How does this fit in with the way objects place themselves in the graph, when you want a feedback path from anywhere to anywhere else? There's got to be a wire looping back into the signal flow somewhere(?)




There are 2 primary ways to do feedback.

There are many places you can read a file from and pretty much all of them you can connect them to a memory writer buffer (including many which may not be labelled as 'read from memorywriter'). Writing to a buffer is as simple as inserting an sound object into your flow and giving it a name. It can be continuous or triggered. You can define the size as well if it is passive and lets the sound continue on the signal flow or it mutes the audio at that point.

It does not visually wire together to the place(s) that are reading from that buffer. I sometimes name the objects in the signal flow if it is useful.

The second way is to use a pair of objects called Feedback Input and Feedback Output. These are slightly different than the above method in that they actually create a virtual input and ouput within the signal flow. It also means that you can span across more than one DSP card (usually) since these are treated as IO rather than buffers. An aside, Kyma allows you to insert IO from physical (or in the capy's case core audio -- paca support is forthcoming) sources at any point in the signal flow.

You can pair the Feedback IO together via the same name. You can write once, read multiple and have multiple independent sets. It is a continuous stream unlike the memorywriter referenced above.

Again the signal flow does not wire the 2 sound objects together, you just see the point at which it writes and then reads by the object type (the icon image) + the arbitrary name you assign to the object.
Back to top
View user's profile Send private message AIM Address
BobTheDog



Joined: Feb 28, 2005
Posts: 4044
Location: England
Audio files: 32
G2 patch files: 15

PostPosted: Sun May 10, 2009 9:03 am    Post subject: Re: Kyma opinions
Subject description: Kyma and hierarchy
Reply with quote  Mark this post and the followings unread

mrbill wrote:
BobTheDog wrote:

Well I am not too sure that you need subroutines and encapsulation for a hierarchy myself. I was talking about the fact that a sound which can be represented by one icon can contain many other linked sounds, with each of these sounds containing many other sounds and so on. So you have a tree of sound graphs.


You can't push down into a prototype to see how it works or change it. So while it may contain other linked sounds, you can't know what they are or affect them in any way. The prototypes (icons) are more like atoms.

BobTheDog wrote:

I must admit I see sounds as this "box", they seem to encapsulate things to me.


Wait until you have created a program with 15-20 prototypes and you will understand why it would be nice to have a way to encapsulate functions and build an hierarchical structure. Kyma graphical programming (i.e.- with prototypes in the "Sound" editor) is flat, not hierarchical.

(You will also understand why it is frustrating to not be able to organize objects on the screen they way you want to see them.)


Hi Bill,

I can see your point on these issues, things could be better and already some parts of Kyma are beginning to annoy me.

But from another viewpoint it is very easy with kyma to start making things that sound pretty good, maybe after a time of using it I will start getting annoyed by the inadequacies that you mention and forget the magic of it.

To be quite honest for me I have managed to get more useful audio from Kyma in a week that I have from MAX/MSP in a couple of years, and I think the reason for this is that I can grab a load of library sounds and connect them together in a serendipitous way, I like this idea of things.

Cheers

Andy
Back to top
View user's profile Send private message
BobTheDog



Joined: Feb 28, 2005
Posts: 4044
Location: England
Audio files: 32
G2 patch files: 15

PostPosted: Sun May 10, 2009 9:13 am    Post subject: Reply with quote  Mark this post and the followings unread

DrJustice wrote:

On other points of interest, I note that the hierarchal design facilities are not implemented as a 'real' hierarchal editor and that there are some complexities involved.


Yep I now understand that when you add a sound into the graph of another sound it is a copy of that sound pasted in.

So one step up from a G2 and one down from say Max or Reaktor.

Even so Kyma is designed to easily allow you to hear what one sound would sound like with another sound as its input, this is built into the browser, you can choose a source sound and see what it is like played through another sound. In any sound you create you can set a single point in the structure where another sound could be inserted by the browser by auditioning.

Also in your sound structure it is easy to replace one sound with another, to place a sound between one sound and another in series and to place a sound between two sounds in parallel.


DrJustice wrote:

Even though I don't expect that many users do it, the ability to go all the way down and write DSP code is good (TBI in Pacarana, I understand). I assume that this is the same facility that Symbolic Sound use themselves - that would make sense. Big plus for openness there.


I am looking forward to this, it came as a bit of a surprise to me that this was not available already. They seem to be changing from a 56K assembler approach to a C approach for this, I have asked if 56K will be supported but have not had an answer yet.



DrJustice wrote:

MIDI Clock and Song Position Pointer would've been nice to have for the timeline. I'm not sure how widespread MIDI Time Code is.


This is an absolute pain, I am hoping there is a way around this, still looking into it.....

DrJustice wrote:

So far, I'd say there is a bit more "Research Institution Flavour" than "MIDI Project Studio Flavour" to it. That's not necessarily a bad thing in itself, just a bit different than what most of us are used to I guess.


Yep I have to agree, MIDI clock is a good example of this.


Cheers

Andy
Back to top
View user's profile Send private message
jksuperstar



Joined: Aug 20, 2004
Posts: 2503
Location: Denver
Audio files: 1
G2 patch files: 18

PostPosted: Sun May 10, 2009 11:23 am    Post subject: Reply with quote  Mark this post and the followings unread

I heard mention of a TI DSP chip, which would imply the 1GHz C6xxx series? In which case, the 56k machine code wouldn't help;) The Symbolic Sound site mentioned somewhere that the paca had 2 processors, while the pacarana had 4. There's no Freescale DSPs that offer this much power in 4 chips. Nor support that much ram!

But those TI ones are very nice, 1GHz, triple core choices. I wonder if the new Kyma hardware uses triple core, or floating point, etc, etc...
Back to top
View user's profile Send private message Visit poster's website
BobTheDog



Joined: Feb 28, 2005
Posts: 4044
Location: England
Audio files: 32
G2 patch files: 15

PostPosted: Sun May 10, 2009 11:49 am    Post subject: Reply with quote  Mark this post and the followings unread

I must admit I had been wondering about that, the only info I was going on was a press release on Harmony Central which said it was a Motorola DSP.

I will make a post on the forum.
Back to top
View user's profile Send private message
bphenix



Joined: Apr 20, 2009
Posts: 17
Location: austin, tx

PostPosted: Sun May 10, 2009 12:41 pm    Post subject: Reply with quote  Mark this post and the followings unread

you could just take the hood off yours and peak at the motherboard Wink
Back to top
View user's profile Send private message AIM Address
mrbill



Joined: May 08, 2009
Posts: 8
Location: USA

PostPosted: Sun May 10, 2009 1:13 pm    Post subject: Re: Kyma opinions
Subject description: Kyma and hierarchy
Reply with quote  Mark this post and the followings unread

BobTheDog wrote:

To be quite honest for me I have managed to get more useful audio from Kyma in a week that I have from MAX/MSP in a couple of years, and I think the reason for this is that I can grab a load of library sounds and connect them together in a serendipitous way, I like this idea of things.


Kyma does have a much higher "instant gratification" factor than MAX does. This is good in that it will keep you involved and pursuing your goals with it.

I continue to use mine after 10 years, despite how much it frustrates me sometimes.
Back to top
View user's profile Send private message
bphenix



Joined: Apr 20, 2009
Posts: 17
Location: austin, tx

PostPosted: Sun May 10, 2009 7:16 pm    Post subject: Reply with quote  Mark this post and the followings unread

if you want to see the signal flow in action, here is a little example video:

http://ssc-media.com/Movies/Tutorials/SignalFlowPt1.mov
Back to top
View user's profile Send private message AIM Address
robsol
Stream Operator


Joined: Apr 24, 2009
Posts: 2492
Location: Bristol UK
Audio files: 495

PostPosted: Tue May 12, 2009 6:22 am    Post subject: Reply with quote  Mark this post and the followings unread

For those of you interested in Smalltalk, here is an online collection of CC licensed books.

Not all of the links work, but there's still plenty of reading material for the curious.

_________________
Muied Lumens Sub Forum
Bandcamp
Back to top
View user's profile Send private message
blue hell
Site Admin


Joined: Apr 03, 2004
Posts: 24075
Location: The Netherlands, Enschede
Audio files: 277
G2 patch files: 320

PostPosted: Tue May 12, 2009 6:33 am    Post subject: Reply with quote  Mark this post and the followings unread

Muied Lumens wrote:
For those of you interested in Smalltalk, here is an online


Might be useful ... at the time I was at the university I didn't get too much of that Smalltalk stuff ... it was not very helpful of course that we were dropped into meta compilation immediately Laughing

_________________
Jan
also .. could someone please turn down the thermostat a bit.
Posted Image, might have been reduced in size. Click Image to view fullscreen.
Back to top
View user's profile Send private message Visit poster's website
BobTheDog



Joined: Feb 28, 2005
Posts: 4044
Location: England
Audio files: 32
G2 patch files: 15

PostPosted: Tue May 12, 2009 9:59 am    Post subject: Reply with quote  Mark this post and the followings unread

Also the smalltalk you can use in Kyma is limited, for instance you cannot create your own classes.

It is mostly just used for setting things up with smallish 'scripts'.

Andy
Back to top
View user's profile Send private message
BobTheDog



Joined: Feb 28, 2005
Posts: 4044
Location: England
Audio files: 32
G2 patch files: 15

PostPosted: Tue May 12, 2009 11:53 am    Post subject: Reply with quote  Mark this post and the followings unread

I asked about the DSP chips and they have said that they cannot comment yet as they are under a NDA.

interesting!
Back to top
View user's profile Send private message
DrJustice



Joined: Sep 13, 2004
Posts: 2114
Location: Morokulien
Audio files: 4

PostPosted: Tue May 12, 2009 12:28 pm    Post subject: Reply with quote  Mark this post and the followings unread

Hrmph... NDA... It's a bit essential for customers, existing and prospective alike, who'd like to write DSP code. Somebody please pop the lid off your unit and check...

DJ
--
Back to top
View user's profile Send private message Visit poster's website
BobTheDog



Joined: Feb 28, 2005
Posts: 4044
Location: England
Audio files: 32
G2 patch files: 15

PostPosted: Tue May 12, 2009 1:21 pm    Post subject: Reply with quote  Mark this post and the followings unread

Maybe thats why they are using C in the new SDK?

Big sticker saying warranty void if box is opened.
Back to top
View user's profile Send private message
bphenix



Joined: Apr 20, 2009
Posts: 17
Location: austin, tx

PostPosted: Wed May 13, 2009 6:44 pm    Post subject: Reply with quote  Mark this post and the followings unread

SSC has always been open about their architecture and provided access to the underpinnings so they wouldn't hide behind an NDA unless they actually were bound to. Given that there has been a few delays due to batches of chips coming in that were not up to spec, it points to a very new chip, perhaps not one widely or publicly available yet.

The prior chip make and model was publicly posted on the website and I am sure this will be as so as they are able to. Given only a small fraction of users develop their own DSP, it was a reasonable trade off to leave out for the initial release. SCC made it clear it was something they are working on so it will be there in time.
Back to top
View user's profile Send private message AIM Address
BobTheDog



Joined: Feb 28, 2005
Posts: 4044
Location: England
Audio files: 32
G2 patch files: 15

PostPosted: Wed May 13, 2009 10:30 pm    Post subject: Reply with quote  Mark this post and the followings unread

Well they were definitely waiting for a TI chip for my Paracarana. At the time I didn't think of DSP I must admit, but maybe...
Back to top
View user's profile Send private message
pdj



Joined: Jun 02, 2009
Posts: 4
Location: UK

PostPosted: Tue Jun 02, 2009 3:10 pm    Post subject: Reply with quote  Mark this post and the followings unread

Hi DrJustice

Re your very first questions-

1) The whole algo is calculated once every sample but each chip is one sample behind the other. Hence modules on the left will be scheduled on the first chip in the chain.

2) There are two ways to handle feed back first memory reader and writer pair but kyma will schedule this such that both are on the same chip. If you put to much processing between them it won't schedule. This gives you a delay of minimum one sample. The other method of feedback is a feedback in out pair. This sends the signal through to the last chip and passes it back to the first. This has a minimum delay of the number of chips you have in your system in samples, but you have no restriction on where they are.

3) Capytalk doesn't work at sample rate it is more for control signals but it's still fast. It can be placed anywhere in the hot param fields inside the modules. The modules (Interconnected) do work at sample and you have a large selection of arithmetic and logic modules as well as more complex ones that can be interconnected to allow you to make algos that work fully at sample rate.

4) You can make your own modules by encapsulating circuits of other modules which could then be encapsulated to make more. There is an SDK for DSP assembly programing but this does not come with the system. You have to ask SSC for this. At this stage it is only for the Capybara and I believe that there are only six of us that are currently using it. I think most find they don't need it or would rather someone else deal with that stuff. Here you make raw DSP modules which sit in with the rest of kyma.


5) You can wire what you want in the schematic but you can not leave it in a state of non working. i.e. you can insert modules on the wires between modules to get the schematic you want but you cannot have non connected modules sitting on the screen waiting to be connected. The GUI is normally referring to the VCS (Virtual Control Surface) which is another window that displays all the controls you have defined in capy talk. Note defining is as simple as !FRED typed in a hot parameter field and you have a control called FRED on the VCS.

Sorry if I've gone over what has already been covered but I hope it helps.

Pete
Back to top
View user's profile Send private message
BobTheDog



Joined: Feb 28, 2005
Posts: 4044
Location: England
Audio files: 32
G2 patch files: 15

PostPosted: Wed Jun 03, 2009 12:08 am    Post subject: Reply with quote  Mark this post and the followings unread

Hi Pete,

welcome to the forum and thanks for the informative post.

Cheers

Andy
Back to top
View user's profile Send private message
DrJustice



Joined: Sep 13, 2004
Posts: 2114
Location: Morokulien
Audio files: 4

PostPosted: Wed Jun 03, 2009 4:31 am    Post subject: Reply with quote  Mark this post and the followings unread

Hi Pete and welcome to EM!

Some of it has been covered, partly in other sub forums here at EM, but your posting certainly helps to further clarify things - thank you!

DJ
--
Back to top
View user's profile Send private message Visit poster's website
pdj



Joined: Jun 02, 2009
Posts: 4
Location: UK

PostPosted: Wed Jun 03, 2009 12:08 pm    Post subject: Reply with quote  Mark this post and the followings unread

Hi All

Sorry for covering old ground but I struggled to get logged in.

It kept asking me to verify the hidden code and it seemed to come up with an average of two zeros or "O"s , so there was a 4 to 1 chance of getting it wrong, but Murphy's law made sure that the odds were worse than that.

I kept running out of attempts and had to wait for the following day.

I eventually got round it by deleting some cookies to get more attempts in the same day attempts.

By this time most of the points had been covered but I decided to post it anyway.
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic
Page 2 of 2 [46 Posts]
View unread posts
View new posts in the last week
Goto page: Previous 1, 2
Mark the topic unread :: View previous topic :: View next topic
 Forum index » Instruments and Equipment » Kyma
Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum


Forum with support of Syndicator RSS
Powered by phpBB © 2001, 2005 phpBB Group
Copyright © 2003 through 2009 by electro-music.com - Conditions Of Use