WaveBench

WaveBench is a desktop app (Windows/macOS/Linux) that lets DSP engineers and audio developers benchmark signal-processing kernels and pipelines on their actual target machines. You import C/C++/Rust code or a compiled shared library, define inputs (sine sweeps, impulses, noise, real recordings), and run repeatable performance tests with automatic warmup, CPU pinning, and cache-control options. The app reports throughput, latency, jitter, CPU vectorization hints, denormal issues, and numerical error versus a reference implementation. It also generates shareable reports for teams and CI, so regressions are caught early. This is a traditional app with optional AI assistance: an on-device model can suggest micro-optimizations (e.g., SIMD-friendly loop transforms) and flag suspicious patterns, but it won’t pretend to replace real profiling. The goal is practical, reproducible performance data, not pretty plots.