[Tero Parviainen] Music and AI in the Browser with TensorFlow.js and Magenta.js

Content level: Intermediate

Slides are at: https://teropa.info/gdgua2018/

2018 is the year deep neural networks have arrived in the web browser. With technologies like TensorFlow.js we no longer need special machine learning setups in order to tap into this thriving branch of AI. Everything you need runs on any web browser.

One of the exciting prospects of this development is that it brings musical AI tools into web apps. We can now take generative musical models designed and trained by people such as Google's Magenta team, and run them in the web browser. Into this we can incorporate other neural net models such as PoseNet, as well as everything else the web platform has to offer: Web Audio, Web MIDI, HTML & SVG, CSS, and WebGL.

In this talk we'll get familiar with some of the current deep learning techniques for making music. We will see how to apply them to generate music in the web browser with TensorFlow.js, Magenta.js, and Web Audio. We will also discuss ways to combine all of this with UI and visual design, to create interactive applications that help people engage with music in new ways.

Tero Parviainen
creative.ai / Finland

Transcript