ftis 1.1.0

The finding things in stuff package.

ftis - finding things in stuff

ftis is a framework for data manipulation, management, creation and munging in Python 3.8+. It is designed for creative use, mainly for my own preoccupations with segmenting, analysing, organising, discovering structure within, and composing with audio corpora.

Architecture

The overall architecture can be conceived in two parts; 'analysers' and 'processes'. A ftis 'process' can house any number of 'analysers' that can be chained and connected in different ways. A python script can house any number of processes and therefore you can compose multiple processes. Inside of a process, ftis makes connections between a source, the analysers and an output.

Installation

You can install ftis using pip install ftis. This will pull down the necessary dependencies so that all of the analysers that ship with ftis work straight away.

You can also fork this repository and clone it to your machine.

Workflow

The simplest setup is to have a virtual environment setup with ftis installed as a module. cd to the clone of your fork of ftis and enter the module (the directory containing setup.py). Once there run pip install -e . to install ftis to your activated virtual environment. Once you've designed your script you can easily run it with python mycoolscript.py. Of course if you have used pip to install ftis then none of the previous advice applies.

There are some good examples of scripts in the examples directory of this repository. Otherwise the basic structure looks like this:

# import ftis modules that we need
from ftis.analyser.slicing import FluidNoveltyslice # novelty slicing
from ftis.process import FTISProcess # a ftis 'process'
from ftis.corpus import Corpus # a corpus object

src = Corpus("~/corpus-folder/corpus1") # corpus object collects audio files at this directory
out = "~/corpus-folder/slicing" # set an output folder

process = FTISProcess( # instantiate an instance of the process
    source=src, # provide the source and the output to the process
    folder=out
)

process.add( # now add different 'analysers' to the process
    FluidNoveltyslice(threshold=0.35, feature=1),
    ExplodeAudio()
)

if __name__ == "__main__":
    process.run() # and run (inside this block for multiprocessing to work)

and thats it! For more information read the full documentation.

Contributing

If you feel up to contributing plumbing code or your own analysers please feel free to do via github.

1. Fork this repository
2. Checkout a branch with your new feature
3. Implement feature
4. Make a pull request!