A Deep Context-Aware Jazz Prediction Model
Welcome to Jasher, an ongoing research project inspired by David MacKay's Dasher, exploring the application of deep Markov models (PPM-C) to jazz harmony. Below you will find interactive tools developed to test, visualize, and eventually generate context-aware chord progressions.
An interactive, real-time prediction interface. Play chord sequences and observe the model's probabilistic suggestions for the next logical harmonic movement based on deep context.
Launch JasherAn autonomous composition engine based on the Jasher models, designed to independently generate and voice complete jazz tunes.
Launch composerA diagnostic utility allowing users and algorithm specialists to explore the underlying PPM-C context tree, analyze node entropy, and visualize the prediction model's raw logic.
Launch ExplorerA continuous-gesture version of Jasher that operates as a harmonic steering wheel. It is a direct nod to David MacKay's revolutionary Dasher interface, allowing you to fluidly navigate deep chord progressions using probabilities as spatial targets.
Launch JDasherAn autonomous composition engine based on the Jasher models, combined with the Dasher-like interface of Jasher Dasher.
Launch ConductorBased on James Cozens' Cycle of Fifths visualizer. Allows users to navigate a dynamic graph where edge widths adjust based on deep multi-chord context.
In Progress