Music Generation with Magenta - Using Machine Learning in the Arts
Overview
Explore machine learning for music generation in this 46-minute Devoxx conference talk. Discover how Magenta, a Tensorflow-based library, can assist in musical creation. Learn about specific neural network topologies like RNN, LSTM, and VAE and their applications in music generation. Gain insights into training models on personal styles to generate new rhythms, melodies, and audio clips. Delve into topics such as MIDI, audio processing, and network architectures. Experience live coding demonstrations, sound selection, and Ableton sound synthesis. Investigate practical applications with Magenta JS and Magenta Studio, and explore various projects showcasing the potential of AI in music composition.
Syllabus
Introduction
Agenda
Definition
Making music without being a musician
Monica Dean Coolidge
Brian Eno
Machine Learning in Music Generation
MIDI
Audio
Network Architecture
LongTerm Memory
Variational Auto Encoder
Wavenet
Network List
Live Code
Sound Selection
Coding
Consent
Results
Sequence
Ableton Sound Synthesis
Recap
Magenta Jazz
Magenta Gain Harp
Using Magenta JS
Using Magenta Studio
Projects
Taught by
Devoxx