Table of Contents
I'm excited to share my new audio visualization tool built with my graphics engine, Cuneus. Spectrum and BPM analysis made in the CPU/Rust side, shader is WGSL.
Video Preview
I created an audio vis within my wgsl shader tool with Rust 🦀
— enes altun (@emportent) March 15, 2025
note: Spectrum analysis on the CPU side via Gstreamerhttps://t.co/XPeSaWMzPD#rustlang #dailycoding pic.twitter.com/tJE1HFOmmM
Download
You can download on here with different OS: (search for 'audiovis' in the github releases page)
Small Tech Details
On the CPU side, GStreamer processes the audio stream to extract real-time spectrum data across 128 frequency bands with specialized handling for bass, mid, and high ranges. The system also incorporates BPM detection algorithms with octave correction to accurately identify musical tempo regardless of genre. But to be honest, its not stable. While coding this,BPM detection turned out to be much more challenging than I expected. spectrum analysis was much easier. The following article I found was very helpful in this process:
https://www.ifs.tuwien.ac.at/~knees/publications/hoerschlaeger_etal_smc_2015.pdf
The WGSL shader pipeline transforms this audio data into dynamic visualizations with frequency-responsive equalizer bars, reactive waveform displays, and color-cycling effects. All rendering is GPU-accelerated through WebGPU, means smooth performance even with complex visual effects while maintaining synchronized audio-visual correlation. :-)