by Hiroki Nishino.
This paper describes a novel implementation technique for unit-generator-based sound synthesis, which allocates audio vectors on the stack frame. The computation is also performed on the stack and the same audio vectors can be shared and reused by sound synthesis modules (e.g., patches or instruments). As the number of audio vectors involved in sound synthesis can be significantly reduced in comparison with a straightforward implementation technique that allocates audio vectors for each unit generator in the heap area, our on-stack allocation technique is desirable for sound synthesis in a memory- limited embedded system. As the utilization of embedded systems is a topic of interest among the NIME (new interfaces for music expression) community, such a memory-efficient implementation technique for sound synthesis can benefit the research and practices in the field.
H. Nishino, On-stack Computation of Audio Vectors for Unit-generator-based Sound Synthesis, International Workshop on Computer Music and Audio Technology (WOCMAT), Hsinchu, Taiwan, 2018 Nov.