Blizzard Challenge 2019 Workshop

The Blizzard Challenge 2019 workshop is a satellite event of SSW10.

Call for participation

The Blizzard Challenge 2019 Workshop is the culmination of the Blizzard Challenge 2019 which is an open speech synthesis evaluation campaign using common data sets and a large listening test. The aims of the workshop are to present the results from the listening tests and for participants in the Challenge to describe their systems.

Who can attend the workshop ?

The workshop is open to all and we encourage participation from anyone interested in speech synthesis. However, please follow the registration procedure below.

Who can submit a paper to the workshop ?

All participants in the Challenge are required to submit a paper describing their entry. Papers will be refereed by the organisers and the Blizzard Challenge Committee

Organisers of the 2019 challenge

• Zhizheng Wu
• Zhihang Xie
• Simon King, University of Edinburgh, UK

Blizzard Challenge committee

• Simon King, University of Edinburgh, UK
• Alan W. Black, Carnegie Mellon University, USA
• Keiichi Tokuda, Nagoya Institute of Technology, Japan

Local arrangements

• Michael Pucher

Paper submission instructions

• Use the Interspeech 2019 authors' kit, but your paper should be SIX pages in length, including bibliography
• Remember that Blizzard is a scientific investigation - we are all trying to understand why some techniques work better than others.
• With this in mind, please write a detailed, technical paper aimed at a specialist audience. Focus on analysis and evaluation. Try to explain WHY your system performed the way it did, and what makes it different from other systems. Explain why your system is designed in a particular way. For example, report internal evaluations you have done to select certain methods.
• Submit your paper by email to blizzard@festvox.org by the deadline given in the official timeline and we will notify you of acceptance within a week. If you need an earlier notification (e.g., in order to obtain travel permission from your employer), then please contact us.

Location and date

Same venue as SSW10

Date: Monday 23rd September 2019

Workshop starts 9am

Programme

The workshop format is single-track with oral presentations from the participants in the Challenge, and the organizers.

Each system presentation slot is 15 minutes and presenters should speak for 10-12 minutes to allow time for questions and changeover. Presenters should bear in mind that the audience will mainly comprise speech synthesis experts; therefore presentations do not need to include extensive background material, but should focus on technical detail and go deep in to the specifics of each system. Please bring your own laptop to present, and a video adapter if you need one.

Presentations must include plenty of audio examples to illustrate the strengths and weaknesses of the system!

The venue this year is a conference facility very generously provided by the SSW10 organisers.

• 09.00 - 09.30 Welcome ; Overview talk ; Sponsors (Apple; Databaker)
• 09.30 - 10.30 System presentations — 4 x 15 minutes
  • University of Tokyo
  • CMU
  • National Taipei University of Technology
  • Nanyang Technological University
• 10.30 - 11.00 Break
• 11.00 - 12.00 System presentations — 4 x 15 minutes
  • Speech Technology Center
  • Tianjin University; Didi Chuxing; Huiyan Technology Co., Ltd. - NO SHOW
  • Institute of Acoustics, Chinese Academy of Sciences
  • Vivo AI Research Center
• 12.00 - 13.30 Lunch
• 13.30 - 14.30 System presentations — 4 x 15 minutes
  • Duke Kunshan University
  • Hithink RoyalFlush Information Network Co. Ltd
  • ShangHai Jiao Tong University
  • National Laboratory of Pattern Recognition (NLPR), Chinese Academy of Sciences - NO SHOW (someone else presented their slides)
• 14.30 - 15.00 Break
• 15.00 - 16.00 System presentations — 4 x 15 minutes
  • Sogou Inc; Northwestern Polytechnical University
  • Mobvoi Information & Technology Company; Northwestern Polytechnical University - NO SHOW (slides presented on the team’s behalf by Shan Yang)
  • Institute of Intelligent Machines (IIM) Chinese Academy of Sciences; University of Science and Technology of China - NO SHOW (slides presented by USTC)
  • Nanjing Horizon Robotics Integrated Circuit Co.,Ltd - NO SHOW
• 16.00 - 16.30 Break
• 16.30 - 17.30 System presentations — 4 x 15 minutes
  • Beijing Lingban Technology Co. Ltd
  • Tencent Technology Co., Ltd
  • Guangzhou Deepsound Technology Co. Ltd - NO SHOW
  • University of Science and Technology of China
• 17.30 Closing remarks
• 18.00 drinks (sponsored) & optional food (self-pay) at Bierheurigen zum Gangl which is very close to the venue (map)

Team IMU-415 from Inner Mongolia University did not present at the workshop, although they did submit a paper.

The following teams failed to submit a paper and therefore did not present at the workshop:

• Paopao — iQIYI, inc
• xmuspeech — Xiamen University
• PingAnTech — PingAn Technology

Practical information

Registration procedure and letters of invitation

• If you are registered for SSW10 then you may attend the Blizzard Challenge 2019 without further registration - simply bring your SSW10 badge with you.
• If you are attending the Blizzard Challenge 2019 without registering for SSW10, please email blizzard@festvox.org to register.

Events you are attending	How to register	Who can provide an invitation letter
Interspeech, SSW, Blizzard	No additional registration needed	Interspeech
SSW, Blizzard	No additional registration needed	SSW
Interspeech, Blizzard	blizzard@festvox.org	Interspeech
Blizzard only	blizzard@festvox.org	blizzard@festvox.org (*)

(*) when requesting a letter of invitation, please explain that you are only attending the Blizzard Workshop and therefore cannot get an invitation letter from Interspeech or SSW.

Cost

There is no cost to attend the Blizzard Workshop, thanks to our generous sponsors, Apple and DataBaker.

Accommodation and travel

No accommodation is arranged by the workshop: please make your own arrangements.

Published proceedings

The papers are published on festvox.org