Forum rss-feed


Developers: Suggested agents

Most Recent

written by: NothanUmber

@0beron: On second thought: Your proposal to retune areas before you even touch them also works when done right and is probably even better from the standpoint of getting reusable subcomponents. Will rework the spec on the weekend.

written by: GoneCaving

Wed, 25 Jan 2012 16:12:02 +0000 GMT

Could someone post the list of suggested agents from last weekend? It might be good to track who's working on what, so if others want to collaborate they know who to talk to.


written by: john

Wed, 25 Jan 2012 16:15:56 +0000 GMT

Hi Duncan - here you go:

- Lua Agent – Difficult
- Wavetable oscillator - Easy, cfilter
- Noise generator – Easy, cfilter
- Vocoder - Medium (because some DSP needed)
- Arpeggiator – Difficult – events, timing
- Plucked string physics model - Medium but cfilter (DSP again)
- Chord generator – Difficult – events, files
- Voice input to interpreter - Difficult (probably - level unknown really)
- Http to light gateway - Easy/Medium – threading
- VU Meter with lights - Easy
- Audio level to single light - Easy

Mathematical agents (all fairly easy):

- Differentiator (Slope)
- Sample/Hold (including max, min and reset)
- Two way momentary/toggle switch
- Multi axis control – controlling value sets from one or two inputs

Worth noting that the metering ones look near to done now, and I think you and Robin were a good way through the noise oscillator as of Sunday.


written by: jsn

Wed, 25 Jan 2012 16:42:31 +0000 GMT

Maybe there should be a Wiki page with this on so we can tick off those we are working on or have done? And any others we're working on.

For example shouldn't John's Fingerer be listed ?

written by: GoneCaving

Wed, 25 Jan 2012 16:58:48 +0000 GMT

Thanks John. Yep, we're close. Robin's going to send me the code tomorrow so we should have something up on github in a day or two.

JSN's wiki is a good idea, and that might also help capture the requirements for new agents.

Dragica, how are you getting on with the wavetable?


written by: john

Wed, 25 Jan 2012 17:00:01 +0000 GMT

And here it is. This one is not versioned, so please feel free to keep it up to date as it is generally editable.

written by: john

Wed, 25 Jan 2012 17:04:19 +0000 GMT

Just to add to that last comment (and following on from Duncan's), yes please use that page to flesh out requirements/specs for them too - we'll try to add some substance to each one in the next few days, but feel free to add too.


written by: NothanUmber

Wed, 25 Jan 2012 19:05:52 +0000 GMT

Have added a preliminary description of the cycler agent I am currently thinking about. Could be a good practise to get a more in deepth understanding of how events are working - and it should be nice to have something like this especially for the Pico :)


written by: john

Thu, 26 Jan 2012 15:24:58 +0000 GMT

Hi Ferdinand

I've just had a read of your proposed Agent, and it looks interesting. I have no idea if it will be good or not, but if you've had good experiences with a related set of functionality in the past then that's hopeful. It's certainly one of those 'need to try that so see how cool it is' ideas.

A couple of things sprung immediately to mind - firstly (and boringly) we already have an Agent called Cycler so you'll need a different name. I really don't have a suggestion though as it's quite an odd piece of functionality, perhaps others here might have an idea?

Secondly, it would be very good if you can think of a way that this can work that doesn't need to care about musical intervals, so it can exist pre the Scalers. This would open up a lot of interesting possibilities if scales apart from chromatic were ever used (remember that in a micro tonal tuning the concept of a perfect 5th can be a bit irrelevant for example). If you assume that the Scaler is set to chromatic you can just use course and key position (which is floating point nowdays) as an equivalent to your chromatic scale, and retain the freedom to do subsequent strange musical mappings - the chromatic scale remains the degenerate case where key and course numbers can behave essentially like a straight scale. This could be particularly important in a system set up to use exclusively just temperament for example, where once again the intervals are different depending on the key you are in but the behaviour of your proposed Agent probably shouldn't change or even have to be aware of that fact (a fact that could be seriously complicated to implement as well).


written by: NothanUmber

Thu, 26 Jan 2012 15:39:27 +0000 GMT

Hi John,

thanks for the feedback! Will try to find a different name.
Working with key offsets instead of musical intervals is more flexible, indeed. Will do this! (The old concept was derived from my old concept that was working with MIDI events.)


written by: 0beron

Thu, 26 Jan 2012 16:57:06 +0000 GMT

So this agent is a way of splitting up the keyboard, such that as you play up a scale from one section into another, it remaps the section you just left in order to extend the range of the instrument, is that correct? Seems like a neat way to reach higher and lower ranges on the Pico without having to reach for the octave keys, but has the potential to be brain melting depending on the phrase you want to play?

written by: keyman

Thu, 26 Jan 2012 17:08:57 +0000 GMT

Hi John, Ferdinand...

We did talked briefly about this new "key approach" on yesterday's G+ hangout, and cannot forget the EVI (it's looking at me) touching the canister would lower/shift a 4th down, and there where also trill keys.

Idea of Agent names:
Shift // Deviation // Sidetrack // Deflexion ?!?

Just my 2 cents...

written by: jsn

Thu, 26 Jan 2012 17:16:38 +0000 GMT

TheFerdinandFunkyFingerer ?

written by: NothanUmber

Thu, 26 Jan 2012 17:32:42 +0000 GMT

Yes, brain melting autotuning :)
We'll see how this turns out in practise, could be worth the experiment.

written by: geert

Thu, 26 Jan 2012 17:48:01 +0000 GMT

The GeoSynth instrument on the iPad does something very crude that sounds like it's going into the same direction. Based on the pattern that you're playing, it automatically transposes to higher or lower octaves. I've never found it particularly useful as it does tend to jump when you don't want to and not always jump when you want. It sometimes feels like you have to play a certain way for the auto jumping behavior to work, which I personally find weird. However, it would be interesting to see if something more intelligent can be built that does this on the Eigenharp.

written by: NothanUmber

Thu, 26 Jan 2012 22:18:49 +0000 GMT

Hope this one is more predictable.
Other than what 0beron wrote it actually transposes the section you jump to, not the one you leave.
You play normally in x. If you touch a key in y then y and z are tuned above x and if you press z then y and z are tuned below x. (You already hear the transposed tuning of the y respectively the z key as soon as you press it) Now the scheme changes to
and so on. So as long as you play at least one note on each 4-key column it behaves exacly as if you'd work with manual transpose keys for up- and downwards scales.

Presumably it's easiest to just try it out when it works :)

written by: NothanUmber

Fri, 27 Jan 2012 12:28:49 +0000 GMT

@0beron: On second thought: Your proposal to retune areas before you even touch them also works when done right and is probably even better from the standpoint of getting reusable subcomponents. Will rework the spec on the weekend.

Please log in to join the discussions