On behalf of the organizing committee, we welcome the participants of the 2015 edition of the IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU). This edition of ASRU followings numerous past instances, occurring once every two years, most recently the 2013 edition in Olomouc, Czech Republic, and in the Waikoloa Village, Hawaii, USA in 2011.
This year we meet at the Firesky Resort & Spa in Scottsdale, Arizona, USA, from December 13 to December 17. Scottsdale is located in the beautiful Sonoran Desert providing a favorable climate during December, spectacular natural scenery, and a vibrant downtown with easy access to the Phoenix metropolitan area. We hope you find the location and facilities of this 4-star resort conducive to meeting with your colleagues during and after the organized workshop sessions.
The ASRU 2015 organizing committee made a great effort to grow the technical program with high quality talks, a great variety of scientific contributions, and social activities that will benefit the research community. This year ASRU received a record number of 224 full paper submissions. The technical chairs assigned each paper to at least three reviewers carefully selected based on their background from a pool of 156 reviewers. We are very thankful to reviewers for providing timely feedback, with each reviewer providing reviews for more than 4 papers, on average. Based on these reviews, the technical committee accepted 107 papers, with a resulting 47.8% acceptance rate. A big thank you to the technical co-chairs for managing this complex process rigorously and promptly: Michiel Bacchiani, Lin-Shan Lee, Karen Livescu, Satoshi Nakamura, Maurizio Omologo, and Michael Picheny. Additionally, the demonstration chairs reviewed and accepted 9 high quality demonstration papers with contributions from both academia and industry. We are grateful to the demo chairs Thomas Schaaf, Patrick Nguyen, and Marsal Gavaldà for managing this important part of the program.
In addition to papers, this ASRU edition hosts challenge tasks as a new component of the workshop program. The goal is to support and promote shared research problems of interest to the community by providing a public, peer-reviewed venue for the reporting of results. Lead by Michiel Bacchini, an open call for challenge tasks was issued in October 2014, and closed at the end of December 2014. Four proposals for challenge tasks were received, of which three were accepted: “3rd CHiME Speech Separation and Recognition Challenge” organized by Jon Barker, Ricard Marxer, Emmanuel Vincent, and Shinji Watanabe; “Automatic Speech Recognition In Reverberant Environments” (ASpIRE) organized by IARPA (point of contact: Mary Harper); and “The MGB Challenge—Recognition of Multi-Genre Broadcast Data” organized by Steve Renals, Phil Woodland, Mark Gales, Pierre Lanchantin, Thomas Hain, Oscar Saz, Peter Bell, and Andrew McParland. Papers related to the challenges - from participants and organizers - have been reviewed and evaluated in the same way as all ASRU papers, with a similar acceptance rate to the overall workshop.
Each day of the workshop opens with a keynote talk. The first keynote speaker is Rich Caruana from Microsoft Research, who will speak about “Do Deep Nets Really Need To Be Deep?”. On Tuesday, we host Prof. Hermann Ney from RWTH Aachen University with a talk on “Bayes decision rule and word error rate in ASR: Some principal and mathematical issues”. On Wednesday, Prof. Jason Eisner from Johns Hopkins University speaks about “Graphical Models Over String-Valued Random Variables”. We conclude on Thursday with Jerry Chen from NVIDIA diving into the new GPUs architectures with “Accelerating HPC – From Simulating to Learning the World”.
Additionally, we have six invited speakers bringing the latest in the state of the art for a variety of topics. This year our speakers are Oriol Vinyals, Heiga Zen, Tomohiro Nakatani, Dan Bohus, Kai Yu, and Steve Renals. Many thanks to the invited speaker chairs, especially Geoff Zweig for spearheading the effort.
A total of 11 poster sessions during the workshop cover topics ranging from fundamental speech recognitions topics to spoken language understanding and dialog. The demonstration session will take place on Wednesday morning. Paper contributions related to the challenge tasks will be presented during the regular poster sessions on Tuesday and concluded by a panel session - “Lessons learned from organizing and running the challenges presented at ASRU” - moderated by Geoffrey Zweig and hosting the challenge organizers and some of the challenge participants.
The social program includes a welcome reception on Sunday evening, and the banquet dinner on Wednesday evening at Desert Foothills. Desert Foothills is located in the pristine desert of North Scottsdale, outdoors in a “rustic” western setting. Many thanks to our local chair Mike Seltzer for organizing the banquet. Additionally, participants may attend networking & hosted wine hour at FireSky every evening starting at 17:00.
We made a number of efforts to make ASRU as accessible as possible to students. First, we secured $5,000 in student travel funds from the IEEE Signal Processing Society, which was used to make 8 student travel grants. We further offered 8 positions as student volunteers, and waived the registration fee as financial support. Overall we were able to offer either a grant or volunteer position to any student-author who requested financial support. In addition, we organized discounted rooms at the workshop venue for 20 students. Finally, we continued the ASRU tradition of offering student registration at a discount of 40% vs. the normal registration fees. This year, students will make up 19% of attendees, which is similar to the past 2 instances of ASRU -- 17% in 2011 and 21% in 2013.
We had an extensive sponsorship campaign and received the generous support of the following companies: 1) Platinum Level: Microsoft; 2) Gold Level: Amazon.com, Google; 3) Silver Level: Interactions, Mitsubishi Electric, VocalIQ; 4) Bronze Level: Appen, IBM Research, Raytheon – BBN Technologies. Thank you to all our sponsors for their substantial contributions, which has enabled much of the support offered to students. We are grateful to our sponsor chairs (Shinji Watanabe, Chao Wang, Bowen Zhou, and Gunnar Evermann) for helping on this task.
We thank our dedicated financial chair, Florian Metze in navigating the intricacy of the workshop budget and IEEE procedures. We also thank our regional publicity chairs (Panayiotis Georgiou, Alexandros Potamianos, and Tomoki Toda), publication chair (Svetlana Stoyanchev), and advisory board chairs (Honza Cernocky,Giuseppe Riccardi, and Rohit Prasad).
We also thank our absolutely fantastic student volunteers: Chuang Ding, Zhou Yu, Shivesh Ranjan, Matt Mirsamadi, Shabnam Ghaffarzadegan, Adriana Camacho, Paola Gallardo, and Chelsey Jurado. Thank you!
In addition we are grateful for the extensive contributions of Lance Cotton, Michael Simon, and Billene Cannon from Conference Management Services (CMS), as well as Kartik Patel and Nicole Allen from IEEE Meetings, Conferences, and Events for their extensive help and support.
As this program goes to press, registrations have reached capacity. When we planned the workshop and proposed the venue more than a year ago, we allowed sufficient event space for an increase of 50% over the previous ASRU’s attendance of 169. We now have an unprecedented 250 registered attendees. We are delighted to see that interest in the field has skyrocketed in the past 2 years, to levels far beyond our expectations! At the same time, we regret that some researchers will be unable to attend due to space, and we will inform organizers of future instances of ASRU of the possibility of large, unexpected growth.
We look forward to seeing you in Scottsdale, and share with you this exciting program.
Sincerely,
Pino Di Fabbrizio and Jason D. Williams
General Chairs