Zynthian AI CLI support feature

Development
1

Has anyone thought about getting the Zythian OS to support an AI CLI account to play the instruments?

It could easily setup instruments, or set a sequence and evolve it in a live setting? Most have aux in too, and it could adapt to inputs and itself.

Please don’t ban me. And sorry if it has already been said here, I didn’t think I would be suggesting Zynthian as the future of generative music when I woke up this morning. I’m not commenting on the use of AI in the code base :slight_smile: I see there are other chats on that.

2 Likes
2

Using an LLM for the task you’re describing feels akin to using a chainsaw on salad: the tool is too complex for the task.

Automated setup scripts/macros, on the other hand, sounds like a really nice quality-of-life feature.

Generative sequencing and parameter automation? :smiley: Routable LFOs and envelope followers? :star_struck:

Pardon me, I’m dreaming out loud. :sweat_smile:

All that to say, there’s other (and more efficient) ways to get what you’re asking for. Let’s try those first.

1 Like
3

I hear you. I was actually thinking it did match the complexity of some tasks, like setting up a sound or taking decisions in a live setting. The amount of parameters on offer is also so high it could be a good fit, and the device can be connected to the internet.

I was also thinking that there wouldn’t be enough push for the use cases, and I guess they need to be front and centre.

4

Please don’t ban me…

Dont worry, if we ever get near doing such a thing, i’ll be the first on the chopping block!

5

@dealing_with_doors

Use Suno or similar slop generators. :face_with_symbols_on_mouth:
And give your Zynthian to someone who actually makes music.

1 Like
6

We do, however, occasionally operate a Paddington hard stare…

And remember, functionality you may want is but a fork away.
The ctrldev environment allows considerable personal tinkering, I myself have about five or so goes at bell programmes of various sorts and i can get a couple of local people who’s view’s I dont share, to claim inspiration cames from higher sources whilst all i’ve really managed smells feintly of incomplete projects and sulphur and brimstone…

I’ve also provoked genuine emotion from war veterans with son et lumiere events
which used battlefield noise taken from video games..

I felt it best not to mention the subtefuge…

Quite how far our next endeavours on this sort of path will dip into available technology is probably a discussion for the pub after ringing.
People seem to be moved by a personal experience which given the number of sensors we could get involved might have a state machine designed by something silicon based.

I happen to find they cover edge cases rather more effectively in such situations.

Apropos of the discussions we’ve managed i have five different versions of different language versions of lillie marlene… as wav files…

How to choose between them…? And select ambiences against time of day and dusk and ongoing services over a month…?

I would prefer it to be something that responds originally rather than a random juke box…

Zynths are rather good at this sort of stuff.

1 Like
7

Thanks for the response. I also wonder if this also opens the doors for those with accessibility issues.

1 Like
8

Certainly we have tried have to address such areas of control, and have frequently used such limitations as motivation for user interface design.

A reasonably competent ai can probably be trained to interpret specific input to allow control and quite where such facilites would cross over from accesibility to overall feature enhancement would probably vary case by case…