Self-Attention for Audio Super-Resolution

Published in

International Workshop on Machine Learning for ...(2021)

External Links:

Generate Graph DownloadPDF

TL;DR

A network architecture for audio super-resolution that combines convolution and self-attention is proposed that outperforms existing approaches on standard benchmarks and allows for more parallelization resulting in significantly faster training.

Abstract

Convolutions operate only locally, thus failing to model global interactions. Self-attention is, however, able to learn representations that capture long-range dependencies in sequences. We propose a network architecture for audio super-resolution that combines convolution and self-attention. Attention-based Feature-Wise Linear Modulation (AFiLM) uses self-attention mechanism instead of recurrent neural networks to modulate the activations of the convolutional model. Extensive experiments show that our model outperforms existing approaches on standard benchmarks. Moreover, it allows for more parallelization resulting in significantly faster training.

Authors

Nathanaël Carraz Rakotonirina

1 papers

References43 items

Generative Adversarial Networks

NU-GAN: High resolution neural upsampling with GAN

Phase-Aware Music Super-Resolution Using Generative Adversarial Networks

Conformer: Convolution-augmented Transformer for Speech Recognition

Speech Enhancement Using Self-Adaptation and Multi-Head Self-Attention

Self-Attention for Audio Super-Resolution

Published in

International Workshop on Machine Learning for ...(2021)

External Links:

Generate Graph DownloadPDF

TL;DR

Abstract

Authors

Nathanaël Carraz Rakotonirina

1 papers

References43 items

Generative Adversarial Networks

NU-GAN: High resolution neural upsampling with GAN

Phase-Aware Music Super-Resolution Using Generative Adversarial Networks

Conformer: Convolution-augmented Transformer for Speech Recognition

Speech Enhancement Using Self-Adaptation and Multi-Head Self-Attention

Attention Wave-U-Net for Speech Enhancement

Speech Audio Super-Resolution for Speech Recognition

Temporal FiLM: Capturing Long-Range Sequence Dependencies with Feature-Wise Modulations

Speech Super Resolution Generative Adversarial Network

An Attention-based Neural Network Approach for Single Channel Speech Enhancement

Attention Augmented Convolutional Networks

Improved Speech Enhancement with the Wave-U-Net

Speech Super-Resolution Using Parallel WaveNet

Wave-U-Net: A Multi-Scale Neural Network for End-to-End Audio Source Separation

Time-Frequency Networks for Audio Super-Resolution

Speech Bandwidth Extension Using Generative Adversarial Networks

FiLM: Visual Reasoning with a General Conditioning Layer

Attention is All you Need

Neural Combinatorial Optimization with Reinforcement Learning

SampleRNN: An Unconditional End-to-End Neural Audio Generation Model

SUPERSEDED - CSTR VCTK Corpus: English Multi-speaker Corpus for CSTR Voice Cloning Toolkit

WaveNet: A Generative Model for Raw Audio

Layer Normalization

Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network

Deep Residual Learning for Image Recognition

Audio super-resolution using analysis dictionary learning

Adam: A Method for Stochastic Optimization

Neural Machine Translation by Jointly Learning to Align and Translate

Speech bandwidth extension using Gaussian mixture model-based estimation of the highband mel spectrum

Bandwidth expansion of narrowband speech using non-negative matrix factorization

Narrowband to wideband conversion of speech using GMM based transformation

Long Short-Term Memory

Learning long-term dependencies with gradient descent is difficult

Statistical recovery of wideband speech from narrowband speech

Finding Structure in Time

Backpropagation Applied to Handwritten Zip Code Recognition

Linear predictive coding

Distance measures for speech processing

Audio Bandwidth Extension Application Of Psychoacoustics Signal Processing And Loudspeaker Design

DNN-based speech bandwidth expansion and its application to adding high-frequency missing features for automatic speech recognition of narrowband speech

Dropout: a simple way to prevent neural networks from overfitting

Beta Process Sparse Nonnegative Matrix Factorization for Music

BANDWIDTH EXTENSION OF AUDIO SIGNALS BY SPECTRAL BAND REPLICATION

Field of Study

Computer ScienceEngineering

Journal Information

Name

2021 IEEE 31st International Workshop on Machine Learning for Signal Processing (MLSP)

Page

1-6

Venue Information

Name

International Workshop on Machine Learning for Signal Processing

Type

conference

URL

N/A

Alternate Names

Int Workshop Mach Learn Signal Process
MLSP