Call for Papers
IEEE JSTSP Special Issue on

Reconstruction of Audio from Incomplete
or Highly Degraded Observations

The restoration of audio content, in particular speech and music from degraded observations, is a challenging and long-standing problem in audio processing. In particular this holds for severe degradations and incomplete observations. Traditional restoration techniques are often not applicable or perform poorly in this case. The advent of sparse signal processing in the beginning of this century and, even more recently, of (deep) machine learning has opened wide new research and design opportunities for audio restoration, among many other signal processing problems. With the aid of such contemporary tools, researchers have recently been able to achieve unprecedented success in recovering or significantly improving quality of severely degraded audio. As the field advances very quickly, the potential for improvement, as well as exploration, is hardly exhausted.

Audio restoration addresses a large number of important degradation scenarios. Further, audio restoration can be performed with varied tools. The proposed issue will serve both as a comprehensive primer on the state-of-the-art, and a showcase of current developments within the field, targeting newcomers as well as already experienced researchers.

Topics of interest in this special issue include (but are not limited to):

• Restoration problems:
  packet loss concealment; inpainting; declipping; dequantization; phase recovery; bandwidth extension; coding artifact removal; compressive sampling recovery; dynamic range decompression; reconstructing audio signals from features,
• Methodological frameworks:
  time-frequency representations; (non-)convex optimization; operator and dictionary learning; nonnegative matrix/tensor factorization; (end-to-end) artificial neural networks; generative networks (e.g., generative adversarial nets and variational autoencoders); graph signal processing; psychoacoustics.

Excellent articles that cannot be accommodated in the special issue will be automatically transferred (without re-submission) and considered for regular publication in IEEE/ACM TASLP.
 
This special issue encourages reproducible research: authors are invited to provide their code and data, to use available material for benchmarking (e.g. SMALL dataset), and to contribute by any means (e.g., high-quality datasets and code, challenges) to the sustainability, the reproducibility and the reliability of the research works in the proposed topics.

Accepted articles are immediately published as Early Access and do not wait until the entire special issue is closed.

Prospective authors should follow the instructions given on the IEEE JSTSP webpages and submit their manuscript to the web submission system.

Important Dates

• Manuscript submission:  May 1, 2020 (Extended)

• Decision notification:  July 1, 2020

• Revised manuscript due:  August 1, 2020

• Second reviews completed:  September 15, 2020

• Final manuscript due:  November 1, 2020

Guest Editors

• Pavel Rajmic  (Brno University of Technology, Czech Republic), rajmic@vutbr.cz, lead guest editor

• Valentin Emiya  (Aix-Marseille University, France), valentin.emiya@univ-amu.fr

• Nicki Holighaus  (Acoustics Research Institute, Austrian Academy of Sciences, Austria), nicki.holighaus@oeaw.ac.at

• Nancy Bertin  (Univ Rennes, Inria, CNRS, IRISA, France), nancy.bertin@irisa.fr

• Alexey Ozerov  (InterDigital, Cesson Sevigne, France), alexey.ozerov@interdigital.com