FPGA-accelerated Real-Time Audio in Pure Data

With the advent of fast ARM processors more audio products are running embedded Linux systems and high level languages to implement real-time signal processing.
Still, embedded Linux systems can't compete with the processing power of desktop computers to implement complex signal processing as needed for physical modeling algorithms.
This paper describes how to interface custom digital logic circuits in an Field Programmable Gate Array (FPGA) with the Linux operating system to speed up processing. The hardware used is a Terasic DE10-Nano development kit equipped with an Intel Cyclone V SoC, running Pure Data (Pd) on Linux and communicating with a mass-interaction network for physical modeling sound synthesis on the FPGA side. The code referenced in this publication is available online.