# Music generation with variational recurrent autoencoder supported by history

## TL;DR

This paper introduces a novel neural network architecture combining a variational autoencoder with a recurrent highway gated network, enhanced by filtering heuristics, to generate longer, melodically diverse, and pleasing music sequences.

## Contribution

The paper presents a new architecture that integrates a variational autoencoder with a recurrent highway gated network, enabling improved music generation.

## Key findings

- Generated music is acoustically pleasing and melodically diverse.
- The architecture successfully produces longer melodic patterns.
- Filtering heuristics enhance the quality of generated music.

## Abstract

A new architecture of an artificial neural network that helps to generate longer melodic patterns is introduced alongside with methods for post-generation filtering. The proposed approach called variational autoencoder supported by history is based on a recurrent highway gated network combined with a variational autoencoder. Combination of this architecture with filtering heuristics allows generating pseudo-live acoustically pleasing and melodically diverse music.