This is a sort of procedural noisy drone thing done in Pure Data.
My process was to listen to all of the files and then just sort of play around with them in PD until I found some sounds I liked. I ended up with 6 different voices using 5 of the provided samples as audio/data sources.
Voice 1
tools02.wav is run through a lo fi granular timestretcher. The grain reader (not sure what to call the fast phasor) is poorly locked to a static frequency via a phase locked loop. The reader’s own frequency as well as various PLL parameters are modulated by “probabalistic” gates with the probability increasing as the piece proceeds. The gates are actually deterministic because instead of using some random value, they are taking samples from a slowly traversed tools02.wav. The audio file is essentially acting as the “score” for its own modulations.
Voice 2
machine02.wav is run through a lo fi timestretcher and filtered.
Voice 3
ambience03.wav is set up in a similar configuration to tools02.wav with a timestrecher and a PLL but without the modulating gates.
Voice 4
work02.wav is run through a timestretcher. It is slowed and pitched down to a sort of low rumble which is then filtered a bit.
Voice 5
work02.wav is run through another timestretcher. This one is connected to another PLL and used for some noisy “pitched” stuff near the end.
Voice 6
tools01.wav is pitched up and used as sample data for a synchronous granular synth. The frequency of the synth is controlled by an envelope follower which is being driven by a slowly traversed tools01.wav. This is another voice whose sound source is also its “score”.
Here is the patch if anybody is interested : https://drive.google.com/file/d/1xYd9F-Ljp8oL7VRVK9Sv4LBD5YBPbRJT/view?usp=sharing
If you want to run it you will need to add the audio files to the /audio directory for things to work. (windows users might also need to change some file paths to have "/"s instead of "\"s in the patch itself)