Rethinking Non-Negative Matrix Factorization with Implicit Neural Representations

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Subramani, Krishna, Smaragdis, Paris, Higuchi, Takuya, Souden, Mehrez
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916833273053184
author Subramani, Krishna
Smaragdis, Paris
Higuchi, Takuya
Souden, Mehrez
author_facet Subramani, Krishna
Smaragdis, Paris
Higuchi, Takuya
Souden, Mehrez
contents Non-negative Matrix Factorization (NMF) is a powerful technique for analyzing regularly-sampled data, i.e., data that can be stored in a matrix. For audio, this has led to numerous applications using time-frequency (TF) representations like the Short-Time Fourier Transform. However extending these applications to irregularly-spaced TF representations, like the Constant-Q transform, wavelets, or sinusoidal analysis models, has not been possible since these representations cannot be directly stored in matrix form. In this paper, we formulate NMF in terms of learnable functions (instead of vectors) and show that NMF can be extended to a wider variety of signal classes that need not be regularly sampled.
format Preprint
id arxiv_https___arxiv_org_abs_2404_04439
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Rethinking Non-Negative Matrix Factorization with Implicit Neural Representations
Subramani, Krishna
Smaragdis, Paris
Higuchi, Takuya
Souden, Mehrez
Audio and Speech Processing
Machine Learning
Sound
Non-negative Matrix Factorization (NMF) is a powerful technique for analyzing regularly-sampled data, i.e., data that can be stored in a matrix. For audio, this has led to numerous applications using time-frequency (TF) representations like the Short-Time Fourier Transform. However extending these applications to irregularly-spaced TF representations, like the Constant-Q transform, wavelets, or sinusoidal analysis models, has not been possible since these representations cannot be directly stored in matrix form. In this paper, we formulate NMF in terms of learnable functions (instead of vectors) and show that NMF can be extended to a wider variety of signal classes that need not be regularly sampled.
title Rethinking Non-Negative Matrix Factorization with Implicit Neural Representations
topic Audio and Speech Processing
Machine Learning
Sound
url https://arxiv.org/abs/2404.04439