Translation. Region: Russian Federal
Source: Novosibirsk State University – Novosibirsk State University –
NSU and NSTU students Kirill Voinov and Nikita Bychkov are developing a new application for instant simultaneous translation of spoken speech from Russian into foreign languages and vice versa. The product is intended for both tourists and businesses, and will allow communication with foreigners without a language barrier: the user will hear the translation in headphones, and his words will be voiced in another language through the smartphone speaker.
The neural network will know Russian, Chinese and English and specialize in translation accuracy and ease of use. To do this, the product will have a complex of neural networks that translate speech and recognize accents, filter noise and synthesize a voice that is 60% similar to the original, and in a conversation of more than two people, the program will voice their words in different voices so that the user understands who is speaking.
Such technologies are becoming popular against the backdrop of globalization, growth of tourism and international business contacts. Such a solution is becoming necessary in conditions when personal meetings are already half replaced by online negotiations and the dependence on correct translation increases. The development of artificial intelligence and neural networks has made it possible to bring closer the future with one language for the whole world – now it is not science fiction, but a tool that works.
—This is a step higher than Google Translate voice. Our application will perform full simultaneous translation of speech, with a minimum number of delays and high accuracy. We have already achieved 80% accuracy and continue to improve the program, — emphasized the project manager, a bachelor’s student Faculty of Economics, NSU Kirill Voinov.
The second developer on the team and project partner, Nikita Bychkov, represents the Faculty of Automation and Computer Engineering at NSTU.
The developers will launch two versions of the product:
— for travelers and everyday communication — a translator app for Android;
— for companies — a complex program with a focus on business translation, taking into account terms and context.
Kirill notes that the application uses new technologies for Russia – a morpheme tokenizer – a special translator from human to neural network language, which will allow to understand Russian better than Western neural networks. For a month now, the team has been developing the main neural network translator, then will train neural network assistants, and then launch the application by the end of September – beginning of October 2025. A commercial product for business will appear in early 2026. Before the official release, the project will undergo a series of free pilot tests with corporate partners.
The developers also emphasize that in Russia, only Yandex is developing such technologies. In other countries, such functions are only being implemented and so far work in the subtitle format or are limited to individual devices, as in the future, the Apple application.
Kirill spoke about the development prospects of the project:
— We want to determine where our product is needed: either in the app market, then we will focus on expanding the number of languages and adding colloquial expressions and slang, or our translator will be of interest to exporters, then we will expand the dictionary through specialized terminology packages. Most likely, we will start with the oil and gas industry.
Please note: This information is raw content directly from the source of the information. It is exactly what the source states and does not reflect the position of MIL-OSI or its clients.