the algorithm is given a set limitation in terms of geometric expectations by the designer and goals concerning reverberation times inside the space. the workflow then uses a series of acoustic simulations and generative 3d neuralnets to create interiors that achive the desired acoustic goals.

the workflow was benchedmarked through a series of digital tests, planning a structure that is spatially continuous but acoustically seperate. this type of approach towards architectural acoustic software could be implemented in future scenarios to improve the efficiency of an architecture's effect on acoustic quality.