A domestic scientist developed a speech recognition system capable of converting voice messages in Kazakh language into text. Moreover, Orken Mamyrbayev’s device is able to recognize two different voices of people talking simultaneously and automatically separate them into two different channels. The intelligent system is already working in social networks as a chatbot. The Kazakh invention attracted the attention of a well-known Dutch company, which proposed further promotion of the product. The foreign company plans to build the system into one of their products already this year. The author of the project is currently improving the device - the system converts audio messages into text only by 60 percent. But by the end of the year, the program’s vocabulary will be replenished with 5,000 hours of spoken words and it will recognize almost all speech. According to Orken Mamyrbayev, his research is aimed at the development of the modern Kazakh language.
“Thanks to the system, you can, first, examine the dialects. Second, recognize the accent. Third, you can determine the emotions of the person speaking. The device also makes it possible to determine what language the person speaks - Kazakh or Russian. Whether it is a man or a woman, of what age? In the future, we may also add rare words. For example, upload economics dictionaries, mathematical terminological dictionaries that people will be able to work with in the future,” said Orken Mamyrbayev, a project owner.
Translation and editing by Galiya Khassenkhanova
