In our continuing effort to better understand different speech synthesis techniques on the same data, we have devised a challenge that will help us better compare research techniques in building corpus-based speech synthesizers.

The Blizzard Challenge 2006 is to take the provided ATR 5000 utterance database and build a synthetic voice. Unknown sentences from an independent source will be generated and each participant will synthesize them with their system. The speech will then be put on the web for evaluation. The results were presented at a satellite workshop on Interspeech 2006 -- ICSLP in Pittsburgh, PA.

This year we had 14 entries from all round the world, 2 from North America, 7 from Asia and 5 from Europe. This year the database made available was substantially larger though we also offered two databases sizes: the full 5000 utterance databases, and a subset consisting of 1200 (ARCTIC) utterances.

Blizzard Challenge 2006 Workshop

Held in the Allegheny I, Westin Conference Hotel, Pittsburgh PA, Saturday September 16th 2006.

• 10:00-10:30 Coffee
• 10:30-10:45 "Blizzard Challenge Welcome and Introduction" Alan W Black and Keiichi Tokuda.
• 10:45-11:00 "The Cerevoice Blizzard Entry 2006: A prototype Database Unit Selection Engine", Matthew P. Aylett, Christopher J. Pidcock, Mark E. Fraser, Cereproc, Edinburgh. PDF
• 11:00-11:15 "Building Probabilistic Corpus-based Speech Synthesis Systems from the Blizzard Challenge 2006 Speech Databases", Sinsuke Sakai, Academic Center for Computing and Media Studies, Kyoto University. PDF
• 11:15-11:30 "The Nitech-NAIST HMM-based speech synthesis system for the Blizzard Challenge 2006", Heiga Zen, Tomoki Toda , Keiichi Tokuda, Nagoya Institute of Technology and Nara Institute of Science and Technology. PDF
• 11:30-11:45 "The Blizzard Challenge 2006 CMU Entry introducing hybrid trajectory-selection synthesis", John Kominek, Alan W Black, Language Technologies Institute, Carnegie Mellon University. PDF
• 11:45-12:00 "The Jess Blizzard Challenge 2006 Entry", Peter Cahill and Julie Carson-Berndsen, University College Dublin, Ireland. PDF
• 12:00-12:15 "USTC System for Blizzard Challenge 2006 an Improved HMM-based Speech Synthesis Method", Zhen-Hua Ling, Yi-Jian Wu, Yu-Ping Wang, Long Qin, Ren-Hua Wang, University of Science and Technology of China. PDF
• 12:15-12:30 "Developing a Test Bed of English Text-to-Speech System XIMERA for the Blizzard Challenge 2006", Tomoki Toda, Hisashi Kawai, Toshio Hirai, Jinfu Ni, Nobuyuki Nishizawa, Junichi Yamagishi, Minoru Tsuzaki, Keiichi Tokuda, and Satoshi Nakamura, ATR, Japan. PDF
• 12:30-14:00 Lunch, Orchard Resturant, Westin Hotel
• 14:00-14:15 "Multisyn Voice for the Blizzard Challenge 2006", Robert Clark, Korin Richmond, Volker Strom and Simon King, CSTR, University of Edinburgh. PDF
• 14:15-14:30 "OpenMary ­ Open Source Unit Selection as the Basis for Research on Expressive Synthesis", Marc Schroeder, Anna Hunecke, Sacha Krstulovic PDF
• 14:30-14:45 "A Study on How Human Annotations Benefit the TTS Voice", Min Chu, Yining Chen, Yong Zhao, Yusheng Li and Frank Soong, Microsoft Research Asia, Beijing, China. PDF
• 14:45-15:00 "IVO Blizzard 2006 Entry", Lukasz Osowski and Michal Kaszczuk, IVO Software, Poland. PDF
• 15:00-15:15 "Multi-tier Non-uniform Unit Selection for Corpus-based Speech Synthesis", Jin-Hui Yang, Zhi-Wei Zhao, Yuan Jiang, Guo-Ping Hu, Xiao-Ru Wu, iFLYTEK Research, Hefei, China. (presented by Zhen-Hua Ling USTC) PDF
• 15:15-15:35 "The IBM Submission to the 2006 Blizzard Text-to-Speech Challenge" Ellen Eide, Raul Fernandez, Ron Hoory, Wael Hamza, Zvi Kons, Michael Picheny, Ariel Sagi, Slava Shechtman, Zhi Wei Shuang. IBM T.J. Watson Research Center, IBM Haifa Research Center, and IBM China Research Lab (Two Systems IBM-Watson and IBM-Haifa). PDF
• 15:45-16:00 Coffee
• 16:00-16:30 "Blizzard Challenge 2006: Results" Christina L. Bennett and Alan W Black. PDF
• 16:30-17:00 "Discussion and Future" open discussion of challenge and next years challanges
• 19:00- Dinner (for those interested)