Compile differentiable audio models to real-time DSP

**Brief Summary** This open‑source library compiles differentiable audio models—typically written in Python‑based machine‑learning frameworks—into low‑latency,