posted 01 October 2006 00:28            

If i take a given structure like SampleCloud -> SimplePitchShifter -> (MonoOutput)
and I want to transform this into a stereo version using two SampleCloud -> SimplePitchShifter chains feeding a stero output, how do I proceed?

I manage to build complex Sounds from scratch starting with a Prototype and deleting and adding elements but modifying an existing structure seems to be quite painful. But maybe I overlook something very obvious here...

Also, I wonder if there is a way to directly instert a class somewhere in the signal chain instead of grabbing a Prototype and deleting things?

Best, Robert

posted 01 October 2006 05:57            

you could start of by double clicking the stero output sound from the prototypes (to create a stereo output sound), copy your SimplePitchShifter structure, and paste it twice into the inputs of the stero output sound. (Just delete the original sounds feeding the stereo output).

You can also do a variation on the copy/paste whereby one instance of your SimplePitchShifter structure will feed both left & right channels. (Don't have Kyma up at the moment to confirm, but I think this is done using the Paste Special... command).

andrew

posted 01 October 2006 06:10            

Everything that you're asking is explained in detail in the manual. I would highly recommend that you have a look at it, it will save you a lot of time. RTFM!

To make a stereo version of your sound you can add a mixer at the end and then add another copy of your signal by copying the SimplePitchShifter and then pasting it in the input field of the mixer.

It's not clear to me what your question is about inserting a class in the signal chain, as you mention you've already added elements after starting with prototypes. But to insert a sound in the signal chain just drag it (from the prototypes, the soundbrowser or another kyma soundfile) and drop it on the line connecting the modules at the point where you want to insert it. If you drop it on the plus sign you will get a mixer with your old sound plus the one you just dragged as inputs, but if you want to just insert the new sound in the signal chain make sure you do it on the line itself. You may also need to pay attention to the default replaceable input in the sounds you drag, as it is at that point of the chain in the sound you're dragging that your original sound will be connected. Everything to the left of that will be lost when you drag it (replaced by the your signal). You can for example open a sound from the library or the prototypes, then instead of deleating what you don't need you can just choose the leftmost point that you want to keep and make it the default replaceable input (from the action menu) and when you drop it in the new signal it will be connected at that point.

Hope this helps.

Cheers,

posted 01 October 2006 13:22            

The structure should look like this:

SampleCloud > channeller (L) > SamplePitchShifter > ChannelJoin
> channeller (R) > SamplePitchShifter >

The same SampleCloud feed both the channeller sounds and the branches are feed into the ChannelJoin.

From scratch:

If you already have SampleCloud > SamplePitchShifter do the following:
From the prototype bar, drag Channeller and drop it on the line between SampleCloud and SamplePitchShifter. Then place a ChannelJoin after the SamplePitchShifter. It will ask you which input to replace (choose left). Now you will have one half of your structure. The other half will be some default left over from the prototype. Copy your structure into the the other side of the ChannelJoin. You can do this via key shortcuts or by drag the structure into the other side.

Now you should have two parallel structures feeeding into the ChannelJoin. Take one of SampleCloud and drag it into the input of the Channeller while holding down the Control key. This should put the one instances in both Channellers. Make sure you've matched up the Right Channeller to the Right side input of the ChannelJoin (there are check boxes in the Channeller sounds).

There are very few stereo sounds among the filter, delay, processing elements, so this type of stereo sound building becomes pretty second nature soon enough.

I tend to put a mixer between the channeller pair and the source (SampleCloud, in this case). That makes it easier to swap in and out sounds or add other sources to the mix.

One thing that saves time just swap out the processing on a pre-build sound. To do this grab one of the stereo processing sounds from the prototype and drop it after your source. This will replace the input of that sound with your input source (the prototypes are generally set up correctly with replaceable inputs). Expand the sound and you'll see a structure like the one above. You can then replace the effect sound in the stereo string with your own by either copy and paste or dragging the new processing object on top of the existing object icon.

Btw. If you haven't tooled through the book yet, the number lines indicates whether the sound is mono or stereo. All the sound icons themselves reference their type. It all becomes pretty easy and clear quickly enough.

posted 02 October 2006 11:23            

Cheers, Robert

posted 08 October 2006 09:45            

I wanted to create a polyphonic granular synthesizer which can be controled via MIDI notes and which allows to move back and forth in a sample and smear around an index point.

GrainCloud seemed to be perfect, but I was not able to figure out how to point to a specific location within a longer sample. SampleCloud offers the needed TimeIndex parameter, but it seems not possible to pitch a grain, since there is no Frequency parameter.

I assume I overlooked something very obvious, but reading the manual was not very helpfull for me: I have been reading all pages dealing with GrainCloud and all dealing with SampleCloud. I ended up with the Cloud Bank element but no success either.
I have to admit the whole logic of the Kyma language and the way it is described makes it quite hard for me to understand what is going on. Is there a reference for each class, which describes each parameter and the function of the object? If I look up the manual index and find e.g. the keyword "Parameters" for "GrainCloud" (page 33) I would expect to get a list of all parameters for this object but this is not the case.
From its parameters I can see that the SampleCloud deals with a sample of arbitrary length
( within the DSP card limit) and it _seems_ that GrainCloud uses a short sample as a wavetable, but where is this described?

I am getting quite frustrated here. I am very familliar with MAX/MSP/Jitter, Reaktor, and the usual studio soft- and hardware. However, with the Kyma I experience more difficulties then with any other program I ever touched.

Is it impossible to build the described granular instrument with the Kyma, or is it just my lack of understanding?

Thank you very much!

Robert

posted 08 October 2006 11:01            

Patience! The best way to learn Kyma is to go through the manual step by step. If you skip sections and only go to the specific places that interest you at that particular time you will miss much.

You can see a description of the basic prototypes in the kyma PDF documentation. Just go to documentation in the file menu and click on "Prototypes Reference". Only the basic classes are there, but it should help. You can also click on the name of each sound in the editor and you will get a list of its parameters with a brief explanation of what each one does.

As for the granualator you want to build, you need to use sampleClouds, and write the name of your soundfile in the sample parameter. Then you can use the TimeIndex to control it. -1 will be the beginning and 1 will be the end. If you want to play once through the soundfile at normal speed you can type: 1 repeatingFullRamp: (whatever the duration of your sample is) s.

You cannot control the pitch of the grains in the sampleCloud. They will be whatever they are on you sample. This is a feature that would be very useful, and some people have been requesting it, but it hasn't been implemented yet. So if you want to control frequency you have to find a way around it by using frequency shifters and so on.

Kyma is a little different from other programs, and it can be a bit tricky to learn it at first. But just hang in there. It is well worth it, and as soon as you become more comfortable with it you will be very happy with it, and you will find it much easier to do certain things in Kyma than in most other programs.

Cheers,

posted 08 October 2006 14:30            

What you'll find after a few weeks of hardcore immersion is that Kyma is very well thought out and allows you to be creative in unexpected ways. I also find Kyma much quicker to work with than similar software, I can usually build a specific idea using Kyma in half the time it would take me to do the same thing in Max/Reaktor.

As I mentioned earlier however, there are some quirks and omissions that make serious instrument building somewhat difficult in Kyma. Kyma is King when it comes to Sound Design (IMHO), but I would like to see some new Prototypes for more complex instrument building.

A Switcher of some sort becomes essential in these undertakings and its omission is a real killjoy for me. I also agree with Robert that a Pitch parameter is missing from the Sample Cloud.

Another big issue I've had this past year is the chore of keeping my patches stereo. I wish there was an easier way to work in stereo than copy/pasting parts of the signal flow, by way of a Mono/Stereo switch in relevant Prototypes for example?

posted 08 October 2006 17:05            

If you take a look over on the tweaky you'll see some basic grain synths I built a while back. They allow you to take the output of a sampleCloud and feed it into an oscillator and synth network. You can modulate the sampleCould to scan through the sample index. Of course, one issue is the 4096 buffer, but that can't really be seen has that big of an issue. A buffer is required of to do granular work of any sorts. You can shorten this buffer up if you like but does have knock on effects in the oscillator section. You can get around those effects by using some offsets in the frequency parameter.

I have some refined ones that incorporate the new wavetable oscillators and fixed some other aspects that I haven't posted and sadly given I am travelling for work current (and for the next year), my kyma and studio is packed up right now so I don't have the sounds around to posted up.

here is the link: http://www.symbolicsound.com/cgi-bin/bin/view/Share/Sounds#Synthesis

comgran and multiwavegrainsynth are the two examples.

I am very familiar with the output of your work (if you are the Henke I think you are) so I am very interested to see what you coax out of kyma in time. Please feel free to ask questions away, everyone seems to sync with kyma differently. Personally, it was the first and still the only of the contruction tools (Reaktor, MAX.MSP, etc) that I really enjoy working with.

posted 08 October 2006 17:29            

I have felt for a long time now there is a bit of a missing middle ground within the kyma prototypes as well as community share of sounds. What I see a lot of is either basic building blocks or more fully blown sounds. There are not a lot of semi-complex sounds packaged up. From the analog world, I guess the analogy would be the oberheim sem or the like. It could be standalone sound device or it could be patched into a larger system (mind you most of my experience with the sem is one that was modded out so perhaps it is not the best example).

It becomes a bit unwiedly at time to deal with very complex sounds. In that way it limits the ablity to create at times. I find that if I collapse sub-sections down with the class editor it certainly helps build up a library of these more complex building blocks. Making new sounds with class editor is time consuming, however.