The class uses the same algorithm as in the coreaudio and alsa backends and should probably go into libardour at some point
