Publication detail

Revisiting synthesis model in Sparse Audio Declipper

ZÁVIŠKA, P. RAJMIC, P. PRŮŠA, Z. VESELÝ, V.

Original Title

Revisiting synthesis model in Sparse Audio Declipper

Czech Title

Revize syntezujícího modelu v algoritmu Sparse Audio Declipper

English Title

Revisiting synthesis model in Sparse Audio Declipper

Type

conference paper

Language

en

Original Abstract

The state of the art in audio declipping has currently been achieved by SPADE (SParse Audio DEclipper) algorithm by Kitić et al. Until now, the synthesis/sparse variant, S-SPADE, has been considered significantly slower than its analysis/cosparse counterpart, A-SPADE. It turns out that the opposite is true: by exploiting a recent projection lemma, individual iterations of both algorithms can be made equally computationally expensive, while S-SPADE tends to require considerably fewer iterations to converge. In this paper, the two algorithms are compared across a range of parameters such as the window length, window overlap and redundancy of the transform. The experiments show that although S-SPADE typically converges faster, the average performance in terms of restoration quality is not superior to A-SPADE.

Czech abstract

Nejúspěšnější metodou pro úlohu audio declipping se v nedávné době stal algoritmus SPADE (SParse Audio DEclipper) od Kitić et al. Doposud byla syntezující/sparse varianta, S-SPADE, považována za podstatně pomalejší než její analyzující/cosparse protějšek, A-SPADE. Ukazuje se však, že opak je pravdou: využitím nedávného projekčního lemmatu lze jednotlivé iterace obou algoritmů udělat stejně výpočetně náročné, zatímco S-SPADE má tendenci vyžadovat podstatně méně iterací ke konvergenci. V tomto příspěvku jsou oba algoritmy porovnávány podle celé řady parametrů, jako je délka okna, překrytí okna a redundance transformace. Experimenty ukazují, že ačkoli S-SPADE obvykle konverguje rychleji, průměrný výkon z hlediska kvality restaurece není lepší než A-SPADE.

English abstract

The state of the art in audio declipping has currently been achieved by SPADE (SParse Audio DEclipper) algorithm by Kitić et al. Until now, the synthesis/sparse variant, S-SPADE, has been considered significantly slower than its analysis/cosparse counterpart, A-SPADE. It turns out that the opposite is true: by exploiting a recent projection lemma, individual iterations of both algorithms can be made equally computationally expensive, while S-SPADE tends to require considerably fewer iterations to converge. In this paper, the two algorithms are compared across a range of parameters such as the window length, window overlap and redundancy of the transform. The experiments show that although S-SPADE typically converges faster, the average performance in terms of restoration quality is not superior to A-SPADE.

Keywords

Clipping, Declipping, Audio, Sparse, Cosparse, SPADE, Projection, Restoration

Released

02.07.2018

Publisher

Springer

Location

Cham

ISBN

978-3-319-93764-9

Book

Latent Variable Analysis and Signal Separation, 14th International Conference, LVA/ICA 2018 Proceedings

Edition

Lecture Notes in Computer Science

Edition number

10891

Pages from

429

Pages to

445

Pages count

17

URL

BibTex


@inproceedings{BUT146951,
  author="Pavel {Záviška} and Pavel {Rajmic} and Zdeněk {Průša} and Vítězslav {Veselý}",
  title="Revisiting synthesis model in Sparse Audio Declipper",
  annote="The state of the art in audio declipping has currently been achieved by SPADE (SParse Audio DEclipper) algorithm by Kitić et al. Until now, the synthesis/sparse variant, S-SPADE, has been considered significantly slower than its analysis/cosparse counterpart, A-SPADE. It turns out that the opposite is true: by exploiting a recent projection lemma, individual iterations of both algorithms can be made equally computationally expensive, while S-SPADE tends to require considerably fewer iterations to converge. In this paper, the two algorithms are compared across a range of parameters such as the window length, window overlap and redundancy of the transform. The experiments show that although S-SPADE typically converges faster, the average performance in terms of restoration quality is not superior to A-SPADE.",
  address="Springer",
  booktitle="Latent Variable Analysis and Signal Separation, 14th International Conference, LVA/ICA 2018 Proceedings",
  chapter="146951",
  doi="10.1007/978-3-319-93764-9_40",
  edition="Lecture Notes in Computer Science",
  howpublished="online",
  institution="Springer",
  year="2018",
  month="july",
  pages="429--445",
  publisher="Springer",
  type="conference paper"
}