Sounds Fun - Jake Archibald plays with the Web Audio API

I’ve read a lot of blog posts from people playing with the Web Audio API for the first time. They always make great introductions to the API for new developers. This one is a little different however, coming as it does from prolific open source developer and Google developer advocate Jake Archibald. Jake introduces some basic concepts from the API such as playing and looping sounds, but also uncovers some edge cases and gotchas that caught him (and no doubt all of us) out. One of his explorations into gaps at the start of decoded compressed audio resulted in him raising this issue against the spec. A great read.

Music Eclipticalis - generate music from star maps

Taking inspiration from John Cage’s Atlas Eclipticalis this interactive musical score superimposes a stave over a map of the sky to generate a continuous musical piece from the position of stars in the sky. Brian Foo, the creator of this beautiful and meditative piece, has documented his work and approach in an accompanying blog post.

Web Audio Weekly has a new home

I’ve been developing a new website to hold the archives of this newsletter. It’s pretty generic at the moment, but I’d like to add some tagging and searching functionality at some point to help you find interesting links from the archive.

I’ve been pretty quiet in this newsletter for the last few months but I’d like to say thank you to all of you who have emailed me to encourage me to keep writing, it’s much appreciated. I did find time recently to give an interview to the Art + Music + Technology podcast if you’d like to learn more about how I got involved with Web Audio and where I think we’re going in the future.

The infinite drum machine

The AI experiments project from Google attempts to explain some of the approaches used in modern artificial intelligence and machine learning using sonic and visual demonstrations. In the infinite drum machine, Web Audio developer Yotam Mann (creator of the Tone.js library) and his collaborators explore how unsupervised clustering algorithms work by creating an interactive map of similar sounds.

Web Audio Font - Soundfont for Web Audio API

Those of you who have been creating computer music for a long time may remember the Soundfont concept introduced in the Soundblaster sound cards popular in the ’90s. A soundfont is essentially a file format for building wavetable synthesisers - digital instruments that use samples of real instruments as the basis of their synthesis.

Srgy Surkv has used this concept to create a detailed sampled instruments that can be used with the Web Audio API. The library includes a full set of “general MIDI” style instruments - from acoustic grand pianos to drums via pan flutes, oboes and synths. A series of examples shows you how to use these instruments in your own applications.

React and Web Audio

If you are familiar with the React JavaScript framework and have been wondering how to use it for Web Audio applications, Joe Sullivan outlines his approach in this blog post.

Soundtracking a chess game using programmatic audio

Ben Griffiths wanted to learn more about how music in computer games can be programmatically generated based on the action in the game itself. To learn more he wrote a music generator for Chess which uses a positional-evaluator to modify the harmonies of the music. The cacophony of explosions that proceed the eventual defeat of one player is a rare treat.

WaveNet: A Generative Model for Raw Audio

A fascinating article from the DeepMind team that explores how audio can be generated using neural networks. Typical speech synthesis applications, for example, generate speech using a pre-prepared library of composite sounds. This approach is different - it generates sounds sample-by-sample at the level of the waveform itself having been trained on a vast library of examples. The paper explores how this can be extended to generative music.

The Great Animal Orchestra

An interactive application to accompany an exhibition by musician and bioacoustician Bernie Krause at the Fondation Cartier pour l’art contemporain. Explore the beautifully recorded sounds of nature arranged by landscape and frequency content.