Cosparse Denoising: The Importance of Being Social

Abstract : This work investigates the performance of cosparse vs. social cosparse regularizations in addressing the audio denoising problem. Beyond the cosparse (also known as sparse analysis) model, results show that exploiting structures in the time-frequency domain is beneficial to audio signal restoration for high degradation levels.
Document type :
Documents associated with scientific events
Complete list of metadatas

Cited literature [8 references]  Display  Hide  Download

https://hal.inria.fr/hal-01510710
Contributor : Clément Gaultier <>
Submitted on : Wednesday, April 19, 2017 - 4:57:30 PM
Last modification on : Friday, September 13, 2019 - 9:50:02 AM

File

GAULTIERSpars17_CR.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01510710, version 1

Citation

Clément Gaultier, Srđan Kitić, Nancy Bertin, Rémi Gribonval. Cosparse Denoising: The Importance of Being Social. The Signal Processing with Adaptive Sparse Structured Representations (SPARS) workshop, Jun 2017, Lisbon, Portugal. ⟨hal-01510710⟩

Share

Metrics

Record views

1804

Files downloads

228