Written in Rust for Linux<p>Most visualizers rely on raw FFT data and linear scales that don't match how human biology experiences sound. The result is usually a twitchy, nervous flicker. I built Lookas to align the visuals with the math of hearing.<p>Here are some technical details (most of these are configurable):<p>- It uses a mel-scale filterbank to remap frequency bins so the visualization aligns with human loudness perception rather than linear frequency spacing.<p>- Animation is driven by a spring-damper model (zeta = 1.0 for critical damping) rather than raw amplitude changes. This gives the bars a sense of mass and weight.<p>- Energy diffuses laterally between neighboring bands to produce fluid motion and prevent jittery spikes.<p>- Input is windowed with a Hann function to reduce spectral leakage. Dynamic range is managed via continuous percentile tracking.<p>It runs at 60+ FPS using Unicode block characters and large contiguous terminal writes to avoid flicker.<p>It's configurable, but defaults work just fine.<p>It captures from mic, system loopback, or both.<p>Excited to hear what you guys think