In order to better understand and compare research techniques in building corpus-based speech synthesizers on the same data, the Blizzard Challenge has been devised. The basic challenge is to take the released speech database, build a synthetic voice from the data and synthesize a prescribed set of test sentences. The sentences from each synthesizer are then evaluated through listening tests.
Blizzard Challenge organisation describes how the challenge is organised, who runs it and tells you how you can get involved.
Blizzard Challenge Roadmap shows the full history and future plan for the challenge. The plan is updated annually and your input is very welcome.
There is a Google Group for discussion and announcements for the challenge:
Participants must join this group in order to receive information about the Challenge. You need to log in to a Google account, then request to join. If you do not have a Google account, you will need to create one.
The annual challenges
Blizzard Challenge 2022 - not run this year
Tools and data
The tools for running your own Blizzard-style listening test and the submissions and listening test results from several previous Challenges are available for your own research. Please remember to acknowledge Blizzard whenever you publish work that makes use of these tools or data. The corpora used in a few of the previous Challenges are also available.