SampleRNN: An Unconditional End-to-End Neural Audio Generation Model (2016-11-04T00:00:00.000000Z)

TL;DR

It is shown that the model, which profits from combining memory-less modules, namely autoregressive multilayer perceptrons, and stateful recurrent neural networks in a hierarchical structure is able to capture underlying sources of variations in the temporal sequences over very long time spans, on three datasets of different nature.

Abstract

In this paper we propose a novel model for unconditional audio generation based on generating one audio sample at a time. We show that our model, which profits from combining memory-less modules, namely autoregressive multilayer perceptrons, and stateful recurrent neural networks in a hierarchical structure is able to capture underlying sources of variations in the temporal sequences over very long time spans, on three datasets of different nature. Human evaluation on the generated samples indicate that our model is preferred over competing models. We also show how each component of the model contributes to the exhibited performance.

Authors

Yoshua Bengio

69 papers

Aaron C. Courville

35 papers

Soroush Mehri

3 papers

TL;DR

Abstract

Authors

References31 items

WaveNet: A Generative Model for Raw Audio

Theano: A Python framework for fast computation of mathematical expressions

Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks

Pixel Recurrent Neural Networks

Multi-Scale Context Aggregation by Dilated Convolutions

Web Audio Evaluation Tool: A Browser-based Listening Test Environment

Building End-To-End Dialogue Systems Using Generative Hierarchical Neural Network Models

A Hierarchical Recurrent Encoder-Decoder for Generative Context-Aware Query Suggestion

An Empirical Exploration of Recurrent Network Architectures

A Recurrent Latent Variable Model for Sequential Data

Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification

Adam: A Method for Stochastic Optimization

Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling

Learning to Generate Chairs, Tables and Cars with Convolutional Networks

A Clockwork RNN

The Blizzard Challenge 2013 - Indian Language Task

Generating Sequences With Recurrent Neural Networks

Speech Synthesis Based on Hidden Markov Models

Random Search for Hyper-Parameter Optimization

The conference paper

Unsupervised feature learning for audio classification using convolutional deep belief networks

Unsupervised learning of auditory filter banks using non-negative matrix factorisation

Modeling High-Dimensional Discrete Data with Multi-Layer Neural Networks

Long Short-Term Memory

Hierarchical Recurrent Neural Networks for Long-Term Dependencies

Learning Complex, Extended Sequences Using the Principle of History Compression

Published as a conference paper at ICLR 2018 S IMULATING A CTION D YNAMICS WITH N EURAL P ROCESS N ETWORKS

Under Review as a Conference Paper at Iclr 2017 Delving into Transferable Adversarial Ex- Amples and Black-box Attacks

The Neural Autoregressive Distribution Estimator

Dating and Context of Three Middle Stone Age Sites with Bone Points in the Upper Semliki Valley, Zaire

Long short-term memory in recurrent neural networks

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names