We invite you to participate
in the competition for the synthesis of lively Russian speech based on the technology of deep neural networks. The competition is designed for senior students, young professionals and all interested persons interested in machine learning and speech technology. The winner will receive 100,000 rubles!
The participants of the TTS challenge will have to create and train a synthesis system and voice with it several dozen sentences of the Russian text. The MDG provides the voice base for learning, the participants choose their own methods for achieving the result. We have recently talked about how to make neural network synthesis yourself.
To participate in the contest, you need to register
on the contest
page before July 12 , better early: we will start sending the first assignments from the middle of the month, there will be more opportunities to practice. After registration, the learning materials download page will be available, which will be open to all on 11 June. We give a little time for learning the system, after which once a week we will send test texts. On the synthesis of each control text are given day. The results of the participants for each task will be published on the site.
According to the results of the competition, we will hold a conference in the St. Petersburg MDG office, where we will invite the authors of the best solutions - we will pay the fare to everyone! At the meeting, we will invite the finalists to make a report on their method and synthesize another control text. While the participants will tell colleagues about the approach they used, our experts will choose the best synthesis. We will immediately summarize the results - and the winner will receive 100,000 rubles!