W3C Audio WG meets at TPAC

The W3C Audio Working Group, the team responsible for developing the Web Audio and Web MIDI specifications, met recently at the W3C plenary meeting. Of particular interest to developers was some of the proposed changes and features expected to be added to the Web Audio API, including a simpler, reworked panner node, support for low-latency custom nodes using audio workers, side-chain compression in the DynamicsCompressor node, and support for suspending AudioContext for battery performance reasons. Although these changes have been agreed, it will take a while for them to specified and a little longer still before they start to appear in browsers, but they should make many more things possible when they arrive.

Meyda - audio feature extraction library

If you’re attempting to automatically detect the genre of a piece of music, want to separate ‘sound’ from ‘silence’, or build a Shazam-like music identification service, you’ll need to perform feature extraction on the audio. This library calculates many useful features including spectral shape, loudness and MFCC using the Web Audio API.

Headphone Surround Sound

The BBC recently broadcast a play which had been recorded for 5.1 surround sound. For those listeners who don’t have such a listening setup, they provided a “headphone surround sound” feature built using the Web Audio API. The ConvolverNode is provided with impulse responses recorded using a dummy head to process the sound so that it sounds like it is coming from a particular point in space even when listened to on headphones. Chris Pike’s post on the BBC R&D blog has more details.


Blip is an expressive sampling and looping API build on top of Web Audio. Its support for random scheduling of samples allows complex-sounding compositions to be created from simple building blocks.


I enjoyed playing with this synthesizer developed by Mitch Wells. It allows micro-tuning of the individual keys, and has a variety of interesting patches.