posted 04 November 2006 17:53            

I'm about to install the AV/C beta (after almost a year since it's been available? But I wasn't on OS 10.4 until recently), but just having some last minute questions after having read the release notes.

First, let me describe what I'm trying to do, so you can advise me whether it's likely to be beneficial for me:

I want to save cpu on the capy by transferring some tasks to Max/MSP, such as file playback and amplitude and pitch tracking. The pitch and amp tracking would be sent to to the kyma via David Kiers' flame~ object (which I gather has been updated). I would use a custom global map to create intelligible hot value names for these and other control streams coming from Max. Maybe I would do some other mixing-type tasks in Max/MSP on audio before I send it to the Capy.

How much of a cpu hit should I expect to take for taking 6 channels of audio in through the SoundFromExternalSource prototypes? If it's more than running three stereo GenericSourceFromDisks and an amp and pitch follower, maybe it's not worth the trouble.

My other concern is the note in the release notes that say that midi mapping in the VCS is disabled. Actually I've recently gotten a motormix so it's not problem when using that, but I think I'll still be using other controllers from time to time (most of my old sounds have VCS mappings using my old controller). So will I lose that? Or can I somehow send the midi from Max internally to the capy? Right now I use Max as the interface between my midisport and the capy: midi comes into the midisport, max remaps it and sends it back out the midisport and the out of the midisport goes to the capy midi in. I currently use the motormix going through max as well, in case I want to use other controllers as well.

So maybe I'd keep the same setup, but wouldn't need the out from the midisport, since that would be sent internally through the firewire? That's my guess, but it's a little unclear.

Is it true that I would be unable to send midi from the capy back to Max, either via the firewire or the physical midi out on the capy? I've previously used that ability to send Max information about where in the Kyma timeline I was.

Finally, what's the current status of getting compatibility with AC Toolbox? I'm always thinking that I'll get around to trying it out when I'm feeling up to tackling LISP. It's not a make-or-break thing for me, but I thought you should know that at least one person is interested in that...

posted 04 November 2006 20:00            

I don't think my 3 year old laptop will be able to handle it very well, but interesting to know that it's possible.

PC

posted 05 November 2006 05:16            

Being the kind of person that never gives up without a second fight, I tried to get it all running again just a couple of weeks ago, on another machine, and I had even less luck this time.

It would be AMAZING if the Capy could be better integrated into OSX, as a standalone audio/MIDI interface (independent of Kyma). And in Kyma be able to send/receive audio/MIDI to other OSX applications, without the current bugs

If that were the case I would buy a four channel upgrade and use the Capy as my primary audio interface. The possibilities are very exciting. However, I tried and tried to get it working and I eventually moved on, and as someone else once said on another thread, using the Capy for what it is best at, Kyma!

From those earlier threads I seem to remember that the main problem for SSC is the Flame interface, which required its own custom protocol to interface with OSX, as opposed to using the standard chips that come in most audio interfaces (or something to that effect). SSC, have there been any developments on this front? It would of course be great if the AV/C could be fully completed in its current form, although personally I would be willing to purchase a new Flame interface if it allowed me to use the Capy as a full, standalone OSX audio/MIDI interface...

posted 05 November 2006 07:16            

quote:

---

Originally posted by JackRosete:
It would be AMAZING if the Capy could be better integrated into OSX, as a standalone audio/MIDI interface (independent of Kyma). And in Kyma be able to send/receive audio/MIDI to other OSX applications, without the current bugs

---

I recently integrated Kyma into an Ableton Live set which really worked a treat. Basically just using midi clips in Live to control the Kyma sound parameters - it solved the problem of syncing Kyma sounds to an external sequencer sync and also editing/arranging the control data in live is easier than using the Kyma Timeline.

(Was using the sound from the recent AutoBreakz thread, which I edited to make it possible to 'play' the Breaks over the keyboard octaves)

If anyone is interested I will post my edit and live set controlling AutoBreakz.

posted 05 November 2006 11:41            

Thanks for sharing your experiences. I think I will play with it for a few days and see. I do have a project that would benefit from doing the file playback in Ableton and using Kyma primarily like effects on the sends. It should take a lot of cpu off of the capy, which I can certainly use elsewhere in the sound. Anyway, I'll see how that goes.

Andrew, your idea of using clips in Live is good, and something I hadn't thought of. Did you also send audio?

posted 05 November 2006 22:36            

Cheers

posted 06 November 2006 01:58            

A few notes about my own experience with AV/C.
Until recently, I was using a PowerBook G4 1,67 GHz with 2 GB of RAM. I found impossible to use the AV/C firmware because the CPU usage was very high. That was what explained why there was these audio glitches, clips and application crashes.

There is also an annoying bug, unfortunately that is more related to Apple's implementation of FireWire Audio instead of our fellow mates at SSC: Even while not streaming audio, on the PowerBook, the kernel_task was using between 4% and 8% of CPU resources with strong CPU spikes from time to time. This is a bug everyone can observe on any audio interface that do not require audio driver (actually, an audio interface that is standard enough to use the standard implementation of bulk drivers provided by Apple).

Now I have a super computer at home, a Mac Pro, and I am happily using Kyma and Logic at the same time without a hitch, things are running smoothly, which means that AV/C concept works well but you need a powerful machine in order to get it right. I think a base of line dual processors G5 is the minimum to have decent results.

In my opinion, Kyma (the host application) needs some optimization in order to enable its usage on every computers, especially in the instructions where the application is idle. Kyma, is never really idle, but in the normal run Kyma uses roughly 12% cpu without doing anything and even more if there's many Hot Variable to track, and 100% cpu when there is an oscilloscope or a spectrum analyser. I hope these optimizations are still possible to realize because SSC folks usually write highly optimized code.

I'd like also to outline that the AV/C implementation is really really clever : you don't need any drivers on the mac to use it. And it is fully integrated with the CoreAudio and CoreMidi services, though the implementation is not yet complete.

Phil, I don't think the limitations you are experiencing are related to the hardware specifications of the Flame. The Flame is just a CPU with a FireWire interface and a flash memory chip. So the firmware flash is enough to expand its possibilities, but you first have to write the software that runs into the Flame which is very complicated. I have been thinking about the question of running the Capybara as a standalone audio interface, and yes I am also a buyer for a 4 channels extention if it was ever possible. But as long as you need Kyma in order to boot the Capybara and patch the I/O, it is not a viable solution. The Flame must be able to boot and initialize the Capybara what Kyma currently does, and I think it would require a lot of work to transpose it from Kyma to the internal Flame software. So the Capybara as stand-alone audio interface is not an experience we'll know soon.

posted 06 November 2006 08:57            

Kyma's CPU usage when in the background is not really what it may first appear. Updating an Oscilloscope display, for example, requires constant two-way communication between the host and the Capybara. Kyma will use the *available cycles* to do those updates as fast as possible. However, it does *not* steal those cycles from other programs running on the host; it uses only the cycles that are left over after all the programs running in foreground and background have gotten what they need. Thus, you can see the Kyma CPU usage vary widely when it is in the background--something that may look strange but which, in practical terms does not impact any other programs.

posted 07 November 2006 03:56            

Kyma-Live.zip

I ended up using a similar sound, AutoKronos (sorry, I can't remember who to credit for the original sound but it does use the Filters from Ben Phoenix) which mixes up samples quite well, albeit in a more chaotic manner.

You might need to change the expression in the SoundToGlobalController according to how many octaves your keyboard has:
((!KeyNumber - 28 / (103-28)) * 2 - 1)

((!KeyNumber - **LowestMidiKeyNumber** / (**HighestMidiKeyNumber** - **LowestMidiKeyNumber**)) * 2 - 1)

This expression scales the midi key input between -1 and 1.

Works nicely in the live gliss sound too!

Curiously, I had some problems trying to scale the midi key input between 0 and 1 (for use in the AutoBreakz) - I couldn't get it to work correctly!

Andrew

P.S. Sorry to get political, but **GO Hilary, GO!!**
It would be great to see Hilary as President of the USA in two years time

P.P.S If you're Republican, please don't let the previous comment put you off enjoying my edits - people of opposing political persuasions can still share specific values whilst debating disagreements ;-)

P.P.P.S Have placed the archive in kyma•tweaky.Share.Sounds.Midi

posted 07 November 2006 08:39            

quote:

---

Originally posted by JackRosete:
Phil, I spent some serious time trying to get the AV/C working a while back, and after a few weeks I gave up.

---

After some initial success, I'm having problems as well. Right now I'm able to send sounds via the SamplesToExternalDestination sound, but am not receiving via the SamplesFromExternalSource sound. Strange.

Phil

posted 07 November 2006 10:42            

P.S. Which version of OS X? (and please remind me which computer)

posted 08 November 2006 22:23            

I tried the test, and doesn't come through the SamplesFromExternalSource module. I'm still able to send audio to another program (Ableton Live) if I select the Capybara as the sound input (I don't think it worked with the built-in audio as the input, and the Capy selected as the default input in the Mac preferences.

I'm using the latest version of Kyma, OS 10.4 on a G5 1.8 dual processor.

thanks,
Phil

posted 10 November 2006 02:48            

I had something similar yesterday while using Logic Pro 7.2.3. Phil, you could perhaps tell if my problem has anything to do with yours?
I have created an aggregated device that shares my Metric Halo 2882 with the Cappybara, this device has 24 i/o's and is called Frankenstein.

I was going to use an effect of Kyma into Logic with the I/O plugin, but it has not been possible. The plugin would not work. If I switched between the CoreAudio input 3-4 (where no audio signal was sent) and CoreAudio input 1-2 I could hear something like a leftover of a sample. I believe this was the 512 samples buffer of Logic.

I have been able to send audio to the Capybara by sending the output of the track directly to the Capybara. I tried to setup a new track with the right CoreAudio input, but all I recorded was inaudible junk, with strong energy (-10 dBFS).

The only solution has been to record the effect directly into the Capybara, and then import it back into Logic.

Is there something specific to do about the buffer size?
I think I remember reading something about this subject. But CoreAudio should take care of that, right?

posted 11 November 2006 13:15            

Sometimes I do hear something that sounds like the buffer when launching the SamplesFromExternalSource sound. Usually after playing another sound (any sound that doesn't take audio in, but generates some sound), and then try to launch SamplesFromExternalSource in order to hear my audio that should be coming in. Instead, I hear a whatever's left over in the buffer looping over and over.