Tidal sound explorer

Resources
1

Hi everyone :slightly_smiling_face:
I want to share a project I've been working on.
It is a python tool that automatically analyze all of your samples, segment them, and plot the segments on a 2d plane such that similar segments will be placed near each other. Then you can explore this 2d space using tidal cycles patterns, see on real-time what samples are being played, and hopefully make some cool music!

The inspiration for this project came mainly from the Flucoma project.

See the repo for code and instructions (please let me know if something is not clear in the instructions and I'll help)

Enjoy your trip in the embedding space!

13 Likes
2

Having a lot of fun with this! As well as updating some paths in the source I had to pip install umap-learn.

I sent a dm with some more thoughts.

1 Like
3

Took the time to get this up and running on my machine today. Had to fiddle with a few things to get the plotter to work, but that comes with the territory when using matplotlib. This is an interesting way to visualize the samples, and I will be curious to see how the project develops. Tempted to add in more sample libraries that I use with software other than TidalCycles, and then to play around more with filtered data sets using the features that are implemented here.
Thanks for sharing this!

1 Like
4

Happy to hear you used it :smiley:
Do you think there is anything that can be improved in the code or README to make it easier to install?

I added some nice capabilities that are not shown in the README... I'll try to get to it if there is interest

5

Just getting back to this. The only thing that tripped me up during installation was that one of the dependencies requires an older version of numpy, so it requires downgrading numpy to work correctly. I ended up setting up a virtual environment and installing the dependencies there and things were fine. Also had to play with the matplotlib settings a bit, but that's a common problem I think.

I plan to revisit this sound explorer more before too long and will update here if I have any comments that might be helpful.

6

I'd forgotten about this, thanks for the reminder... it is super fun to play with :smiley:

...I have +24000 samples, many of which I've forgotten about (or never actually heard), it's actually a great way to do discovery too

7

Thank you! I am still surprised to hear that someone is using it :sweat_smile:
Regarding the python requirements, I updated the requirements.txt so it does not specify a version. It should be easier to install and I just check on a fresh python install and it worked well.

2 Likes