From Music Hack Day
 free assembly
Adapt a collection of sounds to a target using echonest Remix and freesound.org APIs
graham boyes (gr[dot]ed[dot]boyes[at]gmail[dot]com)
 About the hack
Free assembly renders an approximation of a target sound using a database of sound samples. First, using the echnonest Remix API, the target is segmented into different structural units, e.g. tatums, beats, and bars. Second, using tags and filters supplied by the user, sound samples are obtained from freesound.org using their API. This collection of sounds is then combined using adaptive analysis techniques found in my python library for dictionary-based methods (pydbm) to construct a model of the target.
 What APIs, tools or kit did you use?
echonest Remix and freesound.org APIs
Some examples using the amen break can be found here