rvsp is a lightweight audio spectrum visualizer made in Rust, drawing inspiration from techniques used in WebAudio API. It is GPU accelerated through OpenGL, supporting Linux and Raspberry PIs.

This uses PipeWire and Wayland/EGL (via winit/glutin), which is not available anywhere except Linux. Since my primary workstation is Linux, I have no plans to port it to Windows or Mac.

PipeWire development headers must be installed through the package manager of your distribution, if it packages them separately.

$ sudo apt install libpipewire-0.3-dev
$ cargo install --git "https://github.com/cynthia2006/rvsp" --locked

It installs a binary rvsp in ~/.cargo/bin.

• ↑ to increase, and ↓ to decrease gain.
• ← to decrease, and → to increase smoothing time constant (0 < τ < 1).

This app has no mechanism for dynamic configuration; if you want to tweak something, you can tweak it by changing the constants defined in src/main.rs. One might say, it satirizes the aforementioned (cult-like) philosophy.

Suppose you want to broaden the frequency range, you open src/main.rs; modify MAX_FREQ to, for example, 12000.

-const MAX_FREQ: i32 = 10000;
+const MAX_FREQ: i32 = 12000;

Having saved the file, you may re-install the app (--path is the path where the repository is cloned).

$ cargo install --path . --locked

Or, you might execute cargo run to produce a debug build, and see if your changes fit; if not, re-edit, recompile, repeat.

Smoothing time constant (τ) is a parameter that controls how data is smoothed exponentially over time; a kind of lowpass filtering. In this context, it controls the responsiveness of the spectrum—higher the values, the less it is responsive. It defaults to 0.7, which in my personal experience, provides the best experience.

The spectrum is linear, both on the frequency and the amplitude axis. The frequency range is limited between 50 to 10000 Hz; chiefly considering visual aesthetics, because usually one generally won't be interested in the higher parts of the spectrum.

To understand the theory that underpins the mechanics of this program, i.e. STFT (Short-Time Fourier Transform), read this article; and, for the choice of normalization, read this (rather technical) article.