Factorsynth is a Max for Live device that uses machine learning to decompose any audio clip into a set of elements. You can rearrange and modify these elements to remix clips, remove notes, randomize patterns, create complex textures...
C.L.E.E.S.E. (Combinatorial Expressive Speech Engine) is an open-source Python toolbox for performing random or deterministic pitch, timescale, filtering and gain transformations on an input sound. Developed in collaboration with the CREAM team at IRCAM, it has been used to generate sounds for experiments studying how social judgements are conveyed in speech.