@mosgaard

I'm doing this offline in a DAW. Basically adding sound/texture/stuff to a real recording of knocking on a door.

So I collected some stuff aka samples and put it into this Sitala drum sample player. Instead of copy-pasting snippets like a hundret of knocks… I let the samples be triggered. I think the basic EQing and stuff in Sitala should be enough for this.

@nielso if I combine sounds, I would place them on several tracks, EQ, nutch timing, effects (verb), and when everything was set, I would just bounce it into one sound, I would drag it into the sampler.

But that might be less fun :D

@mosgaard

Coming back to this: Yes, I like the fun of tweaking the sounds in Sitala. Basically I define a range in MixBus and play it as a loop and then I can tweak my layerd sounds.

Of course, I don't have much control over each invididual sample, but perhaps that's also a benefit. Sitala has basic tone, volume, panning and sample decay controls.

However, I guess my main reason for doing so is that the director of that play will show up here soon and have other plans again, so I can just tweak the thing "live" with him.

What I'm missing is randomization. Okay, Sitala comes with a 808 sampe set, so it's not about authentic real-world sounds.

Btw., the old 1.0 version which I am using also exists as a Linux binary for free: https://decomposer.de/sitala/ (For some reason, I am using a Windows VST version…)

I could do parameter automisation though.

TAL Sampler with layers would be an idea… but this goes too far for that project. https://tal-software.com/products/tal-sampler

@nielso I think your approach makes a lot of sense in this regard, when you have a director with changing ideas. I use samplers very often, I’m just used to DAWs with good build in samplers, which kind of is why I ended up in Ableton/Bitwig for this kind of work. Bitwig is really huge on modulation and randomization.

It’s very generous of you to be this flexible for the assignment. They are lucky to have you!

And thanks for the link. It’s a really nice take on a basic sampler.

@mosgaard

Hehe, what can I do… *g*

Yes perhaps MixBus/Ardour isn't the 100% ideal tool for the job, but it's the tool I know best.

Just learning to use it's Rhythm Ferret function, since this director guy wants the knocking on his bunker door slower (to bad he wasn't present during recording, he could have just said so…)

So, I can slice that door knock beat into regions automatigally, now I'd need some magic to re-distribute a selection of regions accordingly, like… fanning them up. And I bet there is a function for this, but I haven't found it yet? @PaulDavisTheFirst - @x42

(Still in MixBus 10 here. I'm a happy but late adaptor. )

@PaulDavisTheFirst @mosgaard @x42

Thanks for your reply, Paul.

Yet, it's not exactly my use case. I don't need to snap to a grid, I just have the request to "generally make it slower", which means after using the RF split, I'd put some "space" in between these regions somehow, ideally not manually.

The recording doesn't follow any beat or rhythm.

@PaulDavisTheFirst @mosgaard @x42

So, using #ChatGPT to vibe code #LUA for #Ardour… what shall I say? Instant karma's gonna get you!

AI karma got me right away, and it hit hard.

Such a PITA. This LLM thing doesn't know the API and it doesn't even notice contradicting itsself.

Still, with a lot of trial and error, I got code drafted that does what I need:

Spacing out regions by a given number of seconds (floating point numbers allowed).

Works with one exception: I wasn't able to properly commit the action to the undo history.

(EDIT: This got solved by doing what ChatGPT should have done properly, looking at other people's code on GitHub.)

So, here it is, this is… well I'm very uncertain about this, but perhaps one day it will actually be proper code that can help other people, too:

https://pastebin.com/q0QHag0Z

(I added TODO comments where there's issues.)

#vibeCoding