it is a flexible, multi-voiced, varispeed audio buffer-manipulation framework. it has many features including sample-accurate crossfaded looping and built-in filters. can be configured as an echo / looper effect, granulator, sample playback unit, weird “live wavetable” oscillator, tape emulation, chorusy thing, &c, and provides the audio backend for norns MLR implementation.
in norns 1.0 there was a rather limited implementation of this thing available as an Engine in the supercollider environment. in 2.0 it is available in the mixing process (pre-fx), and can affect an arbitrary mix of Engine output, ADC signals, and/or Tape playback. (sorry, my bad, i did not add that mixing route.) softcut voices can also be cross-patched.
(i know… i need to write up a good manual for this thing, not to mention a guide to the overall signal flow in the current norns architecture. as soon as i fix the Tape bug.)