jfochezato / midi-autofill

MIDI-Autofill

Background

All musicians know the frustration that comes with writer’s block. Whether the musician is making music for personal enjoyment or to meet a deadline, completing a piece of music seems to be the most difficult part of writing, even when the initial design is complete. Our solution to musical writer’s block is a stand-alone, easy to use, portable MIDI controller that can fill in melodic structures inputted by a user. Our concept relies heavily on artificial intelligence and machine learning that is supported with a well-designed MIDI controller and onboard processors. Since we do not expect any musician to be familiar with computer or electrical engineering, our MIDI controller will have an onboard display and a user interface similar to a Digital Audio Workstation (DAW) where the user can edit, playback, and quantize their entries to ensure their exact musical intent before processing the melody. Currently, we have technology that can process and continue audio through artificial intelligence such as Jukebox AI, but this requires an understanding of computer science that not all musicians would enjoy learning. The aim of this project, then, is to put all the technical aspects of this human/computer collaboration on the computer side. Using Machine Learning, we can train the computer to recognize music patterns and structures such that it could generate a comprehensive melody at the press of a button.

Diagrams