JSConf US2013

Video recording and production done by JSConf

Digital Signals Processing (DSP) is inherently essential to any signals-based field, be it digital audio effects, computer graphics, or computer vision. The web has just begun it's foray into accommodating for DSP, becoming increasingly equipped to handle and modulate signals with the advent of technologies such as the Web Audio API and WebGL. With the basic DSP tools provided by these technologies, we are minimally bootstrapped and able to leverage these tools to implement powerful algorithms and incorporate powerful fields of research. One such application is the modulation and processing of music signals. In this talk I will focus on an application that is more complicated than it seems: automatic BPM detection of music. Due to the complexities of how we as humans perceive sound, parse harmonics, and "feel" beats within a song, BPM becomes a challenging and enthralling problem to tackle. I've just begun to embark on an endeavor to implement BPM detection in javascript, building upon budding DSP tools in JS and pulling from algorithms hardened by academic research. I have been digging into academic research papers in the field identifying and categorizing plausible algorithms from simple to complex, separated by effectiveness for differing applications. I hope to implement a number of these algorithms across this spectrum and compare their effectiveness for different kinds of music and applications. I believe exposing and providing robust algorithms for BPM detection would be a strong tool and asset towards many javascript-based web (and server-side ;D) applications, whether for categorization purposes, or for something more dynamic like music visualization.

Rated: Everyone
Viewed 2 times
Tags: There are no tags for this video.