IT and Society: Russian speech recognition service to be designed

Following the appearance of popular international speech recognition services ran by Apple and Google in the United States, similar service may be designed in Russia to recognize Russian speech. The Ministry of Education and Science has decided to allocate some 30 million rubles to conduct relevant studies. 

Open contest, announced by the Ministry of Education and Science last week, will determine the software developer. The Ministry hopes that the future technology will leave all popular products of the main competitors behind judging by its scientific, technological and engineering potential.

The new system will be able to recognize not less than 10 various dialects of the Russian language, including pronunciation of inhabitants of the Caucasus, Volga Region and Asian area of the country. The prototype will also understand mixed phrases, therefore the system will be taught English.

User’s query may be either interrogative or declarative sentence, or key words. The total length of the query must not exceed 512 symbols, the recognition quality of dictated speech will not be lower than 80% of the whole text. The Russian technology will adapt to new words from any subject area due to the constant update of its morphological dictionary.

Developers believe that unlike Google Voice, the new service will be available to run both with Internet connection or offline. However in the second case the speech recognition will be less accurate, what is explained by a limited vocabulary, stored directly on the computer.