What if there was a sample pack format with metadata for SuperDirt?

Great samples @cleary! I was wondering about clicks in the samples until I realised your demo had cunningly added a pattern of clicks on top !

What have come to be known as the 'default' samples are of mixed and sometimes unknown provenance, and for a looong time it's been an aim to start again, with a decent set of CC licensed samples.

Here's a thread about it:

Basically I'm wondering if you'd be up for adding a bit of metadata in YAML or json that superdirt could understand, in a simple in extensible way? Then we could work on making them easily addable to SuperDirt via the quarks system or something like that.

1 Like

Yeah of course!
I'll have a read over the next few days and shout if I have any questions

1 Like

Yes I'd have thought so. I don't have much knowledge on sound engineering so no idea what is an ideal number.

We could also add something like replaygain metadata to existing samples. https://en.wikipedia.org/wiki/ReplayGain

1 Like

I have none - can anyone with some more knowledge chip in here?

That looks interesting, would that require manual adjustment (in the metadata) by developers to match perceived loudness relative to other samples?

They could potentially do that, or it could be done by some automatic psychoacoustic analysis. I guess I'm just thinking that to adjust loudness of samples, we could change the metadata rather than the sample files. The metadata could be stored in an individual .json or .yaml file for each sample, or one json file for each folder that has a section for each file.

1 Like

I think it's a great idea... I was debating whether I needed to normalise these flbass samples again, but I think I'll leave them as is. They've had a good run over a number of users and platforms by now and they seem to be ok


Apologies for this, but I'm going to need a more obvious pointer (shove) down the road here - is there any documentation or examples for the type of metadata or even just variables that superdirt is looking for?

I'm still not sure what I should be including (or how I should format it for inclusion) at this stage - things that seem obvious at my noobish high level are just the bits I've included in the README/file names already:

  1. License and copyright information
  2. per sample description
  3. general sample set description

I can pop them in a metadata sub-folder, or I could include a metadata file per sample - I still need some guide on what the var naming should look like for superdirt compatibility though?

Fair question! There was some discussion around e.g. https://github.com/tidalcycles/Dirt-Samples/issues/15 but it didn't get as far as agreeing a format.

I've started a new issue here: https://github.com/tidalcycles/Dirt-Samples/issues/20

We could the teh discussion there?