Solving Time Domain Audio Inverse Problems using Nonnegative Tensor Factorization

Abstract : Nonnegative matrix and tensor factorization (NMF and NTF) are important tools for modeling nonnegative data, which gained increasing popularity in various fields, a significant one of which is audio processing. However there are still many problems in audio processing, for which the NMF (or NTF) model has not been successfully utilized. In this work we propose a new algorithm based on NMF (and NTF) in the short-time Fourier domain for solving a large class of audio inverse problems with missing or corrupted time domain samples. The proposed approach overcomes the difficulty of employing a model in the frequency domain to recover time domain samples with the help of probabilistic modeling. Its performance is demonstrated for the applications such as audio declipping and declicking (never solved with NMF/NTF modeling prior to this work), joint audio declipping/declicking and source separation (never solved with NMF/NTF modeling or any other method prior to this work), compressive sampling recovery and informed source separation (a low complexity encoding scheme that is possible with the proposed approach and has never been proposed prior to this work).
Document type :
Journal articles
Complete list of metadatas

Cited literature [33 references]  Display  Hide  Download

https://hal.inria.fr/hal-01669825
Contributor : Alexey Ozerov <>
Submitted on : Thursday, December 21, 2017 - 1:06:44 AM
Last modification on : Wednesday, October 17, 2018 - 4:45:47 PM

File

AudioInverseNMF.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01669825, version 1

Collections

Citation

Cagdas Bilen, Alexey Ozerov, Patrick Pérez. Solving Time Domain Audio Inverse Problems using Nonnegative Tensor Factorization. IEEE Transactions on Signal Processing, Institute of Electrical and Electronics Engineers, 2018, 66 (21), pp.14. ⟨https://ieeexplore.ieee.org/document/8458165⟩. ⟨hal-01669825⟩

Share

Metrics

Record views

117

Files downloads

214