Activity Report from the AASP-TC

Published in TC News on 1 July 2017

by Sharon Gannot (AASP-TC Chair) and Patrick A. Naylor (AASP-TC past-Chair)

The AASP TC's mission is to support, nourish and lead scientific and technological development in all areas of audio and acoustic signal processing. Ranging from array processing for microphones and loudspeakers to music genre classification, from psychoacoustics to machine learning, from consumer electronics devices to blue-sky research, this remit encompasses countless technical challenges and many hot topics.

Honors

Under the leadership of Shoji Makino, the chair of the Nominations and Elections and the Fellows Subcommittees, the AASP TC had the following successful honors from IEEE.
1. IEEE Fellow: Gaël Richard, Telecom ParisTech, Paris, France for contributions to analysis, indexing and decomposition of audio and music signals.
2. 2015 IEEE Signal Processing Society Industrial Innovation Award: Gary Elko
3. 2016 Best Paper Awards: Cees H. Taal, Richard C. Hendriks, Richard Heusdens and Jesper Jensen, for “An Algorithm for Intelligibility Prediction of Time–Frequency Weighted Noisy Speech”, IEEE Transactions on Audio, Speech, and Language Processing, Volume 19, No. 7, September 2011.

New TC Members

In September 2016, AASP TC held its annual election of new members. We had a record number of 31 nominations. We welcome the following elected (or re-elected, after a period outside the TC) members:

1. Fabio Antonacci, Politecnico di Milano, Italy
2. Simon Doclo, University of Oldenburg, Germany
3. Dan Ellis, Columbia University, NY, USA
4. Hirokazu Kameoka, NTT Commumication Science Laboratories, Tokyo, Japan
5. Michael Mandel, Brooklyn College, NY, USA
6. Jan Skoglund, Google Research, CA, USA
We would also like to thank all retiring members of the committee for their contribution. Now the committee consists of 1 Chair, 1 Past Chair, 33 members, 19 associate members, well represented world-wide with roughly 60% of the members from academia and 40% from industry.

IEEE/ACM Transactions on Audio, Speech and Language Processing: Tutorial Paper

A tutorial-style survey of the field of speech enhancement and speaker separation, titled “A Consolidated Perspective on Multimicrophone Speech Enhancement and Source Separation” was recently published in IEEE/ACM Transactions on Audio, Speech and Language Processing. Research in this domain has followed two convergent paths, starting with microphone array processing and blind source separation, respectively. These two communities are now strongly interrelated, and routinely borrow ideas from each other. The paper proposes a comprehensive overview of the common foundations and the differences between these approaches. A large number (almost 400 citations) of established and recent techniques are explored according to four transverse axes: 1) the acoustic impulse response model, 2) the spatial filter design criterion, 3) the parameter estimation algorithm, and 4) optional postfiltering. A list of software and data resources is also provided. The article is co-authored by Sharon Gannot, Emmanuel Vincent, Shmulik Markovich-Golan and Alexey Ozerov, three of them are currently AASP TC members.

Conference and Workshops

WASPAA 2017: IEEE Workshop on Applications of Signal Processing to Audio and Acoustics, The next WASPAA will be held from October 15th to October 18th 2017, as usual, at the Mohonk Mountain House in New Paltz, New York, USA. The general chairs are Patrick A. Naylor from Imperial College, London UK and Meinard Müller from International Audio Labs, Erlangen, Germany.

ICASSP 2017: AASP TC members and voluntary reviewers have been busy reviewing papers submitted to ICASSP 2017 held in New-Orleans, LA, USA. The number of submissions to the AASP area was 323, exhibiting the constant growth of the field due to the steady increase of available computing power at relatively low cost allowing increasingly sophisticated processing to tackle longstanding challenges in acoustic signal processing. Paris Smaragdis, the Review Subcommittee Chair, handled the review task (with the assistance of Patrick Naylor and Sharon Gannot). Many area chairs and secondary area chairs (all members or associate members of the TC) assisted Paris in assigning reviewers and writing meta-reviews. A special session on “Speaker localization in dynamic real-life environments”, was jointly organized by the SAM and AASP TCs. The special session, co-chaired by Peter Willett and Sharon Gannot, attracted a large audience from both communities.

HSCMA 2017: Many members of the AASP TC are also involved in the Hands-free Microphone Arrays and Speech Communications (HSCMA), held in March 2017 in San Francisco, CA, USA, overlooking the bay bridge. The workshop is held once every three years. It brings together members of the AASP and Speech (mainly recognition) communities. The workshop was co-chaired by Malcolm Slaney (Google), Jerome Bellegarda (Apple), Ivan Tashev (Microsoft Research), two of them associate members of our TC. The workshop is held once every three years. It brings together members of the AASP and Speech (mainly recognition) communities. Over 100 participants enjoyed 43 presentation and 3 distinguished keynote speakers.

IWAENC 2016: Many members of the AASP TC are deeply involved in the International Workshop on Acoustic Signal Enhancement (IWAENC). IWAENC is held bi-annually in even years in September. The IWAENC 2016 edition took place in September 2016 in Xi’an, China and was co-chaired by Jingdong Chen and Andy Khong. The 164 attendees enjoyed the luxurious Tang Dynasty West Market Hotel, close to the ancient town of Xi’an at the entry point to the famous Silk Road, in addition to excellent presentations of 82 papers and six keynote and tutorial talks by leading researchers in the field.

Associate Editors:

Members and associated members of the AASP TC are very active in editorial tasks.

The following members (or strongly affiliated colleagues) are currently acting as Associate Editors (AEs) for IEEE/ACM Transactions on Audio, Speech and Language Processing: Simon Doclo, Augusto Sarti, Hirokazu Kameoka, Richard C. Hendriks, Tuomas Virtanen, Roland Badeau, Alexey Ozerov, Sven Nordholm. The following members and associate members are acting as Senior Area Chairs: Vesa Välimäki, Sharon Gannot (retiring at the end of August 2017) and Patrick Naylor.

Jacob Benesty and Augusto Sarti are serving as Senior Area Chairs for IEEE Signal Processing Letters and Boaz Rafaely is serving as an AE for the same journal.

Emanuël Habets is serving as an Editor in Chief for EURASIP Journal on Audio, Speech, and Music Processing.

AASP Challenges

One of the most important mechanisms for fostering research in emerging disciplines is the “TC Challenges” endeavor. Several years ago, the AASP TC has established the Challenges Subcommittee, currently chaired by Tuomas Virtanen (past-chair Emmanuel Vincent), in order to encourage research and development with comparable and reproducible results, and to stimulate new ground-breaking approaches to specific problems in its technical scope.

Recently, a very successful TC Challenge dissemination workshop, summarizing the Detection and Classification of Acoustic Scenes and Events (D-CASE) challenge, took place as a satellite event to EUSIPCO 2016 with Tuomas Virtanen and Mark D. Plumbley as co-chairs. The workshop attracted 82 submissions and 68 participants. A second DCASE edition will be held in 16 - 17 November 2017, Munich, Germany. Tuomas Virtanen and Annamaria Mesaros will co-chair this event.

The forthcoming “LOCATA” challenging will address acoustic localization and tracking and will be organized by Heinrich Löllmann, Christine Evers, Hendrik Barfuß, Patrick Naylor and Walter Kellermann.

Bird Audio Detection (BAD), organized by Dan Stowell, Hervé Glotin, Yannis Stylianou and Mike Wood
will be presented in a special session at EUSIPCO in September 2017, Kos Island, Greece. Other challenges are planned in the near future.

The AASP TC also organized the IEEE SP cup 2017 on real-time beat tracking, in conjunction to ICASSP 2017. The organizers of this event were Craig Jin and Matthew Davies.

The AASP TC Challenges Subcommittee will be glad to assist prospective organizers in running a successful challenge by providing scientific and organizational feedback and by sharing industrial sponsorship contacts.

Emerging Topics
“Traditional” disciplines, e.g. microphone array processing, source localization and speaker separation, are still constantly growing, seeking for new and more involved applications such as hearing devices and dynamic scenarios, e.g. robot audition, taking advantage of the steady increase of available computing power at relatively low cost. Music signal processing is exhibiting continuous growth due to factors including monetization opportunities for managing and marketing large music libraries. In parallel, we are witnessing in recent years the emergence of new disciplines into the AASP field, most notably data-driven (as opposed to model-based) methods, especially deep neural networks. DNNs find nowadays diverse applications and are not limited anymore to single microphone speech enhancement and to “classical” classification problems. Many papers are applying DNNs to source separation, beamforming and localisation and to diverse music applications. This comes hand-in-hand with a dramatic increase in the number of submissions (and accepted papers) in the field of “acoustic scene classification” and “event detection”.
These changes are reflected in the recently updated EDICS list prepared by the EDICS subcommittee, chaired by Simon Doclo (past chair Emanuël Habets) and submitted to the relevant journals.

Publicity and Communication 
The Publicity Subcommittee, chaired by Dorothea Kolossa, is actively seeking for ways to outreach a larger community by using new media, e.g. Twitter and Facebook. Fabio Antonacci, our webmaster, is responsible for regularly updating the information posted on the AASP TC website. Specifically, we are collecting information regarding tutorials, databases and software resources.