Getting started

Description

hachi machi (High-level And Controllable Human Interface for MACHine Improvisation) is an open-source tool for performing artists to quickly and easily prototype, train, and run their own improvisational models, with their own data, in their own computers.

How it works

hachi machi works as a command-line interface (CLI) written in Python, with PyTorch as the back-end for deep learning. It's designed to provide the most control to artists and technologists in the data-to-model pipeline while minimizing the amount of expertise required.

To train a model with hachi machi, all you need is a set of sequential data—that is, data that can be represented as a series of steps, each step consisting of set of numerical features or values. The role of a hachi machi model is to learn how to predict the next event, like so:

hachi machi's model architecture is, in principle, data agnostic. This means models can be trained on any kind of sequential data—such as musical data, sensor data, or any other type of data where the order of events is semantically meaningful.