Rethinking Recurrent Latent Variable Model for Music Composition

TL;DR

This paper introduces a novel convolutional variational recurrent neural network for music generation, capturing musical features and structures more effectively than previous models, leading to more realistic and stylistically consistent musical sequences.

Contribution

The paper proposes a new neural network architecture combining convolutional, variational, and recurrent components for improved music sequence generation.

Findings

The model better captures musical structure than existing neural networks.

Generated music sequences statistically resemble training data.

The approach improves motif detection and information dynamics modeling.

Abstract

We present a model for capturing musical features and creating novel sequences of music, called the Convolutional Variational Recurrent Neural Network. To generate sequential data, the model uses an encoder-decoder architecture with latent probabilistic connections to capture the hidden structure of music. Using the sequence-to-sequence model, our generative model can exploit samples from a prior distribution and generate a longer sequence of music. We compare the performance of our proposed model with other types of Neural Networks using the criteria of Information Rate that is implemented by Variable Markov Oracle, a method that allows statistical characterization of musical information dynamics and detection of motifs in a song. Our results suggest that the proposed model has a better statistical resemblance to the musical structure of the training data, which improves the creation of new sequences of music in the style of the originals.