Scientists create first Kazakh synthetic voices

Scientists at Nazarbayev University created the first Kazakh sythetic male and female voices. They were named Iseke and Raya, which are the derivatives of real names of the two people, whose voices formed the basis of the program. The invention converts text into audio and promotes the use of Kazakh language in the digital sector.

“We collected more than 90 hours of voice by the two voiceover artists and developed two systems that can help convert text into speech. Firstly, we collected this amount of data for the first time in Kazakhstan. Secondly, it is all publicly available. Anyone can download this data and use it for non-commercial and commercial purposes. It can be used with other components in the future in virtual assistants such as Siri and 2GIS navigation app,” said Saida Mussakhojayeva, Data Scientist at the Institute of Smart Systems and Artificial Intelligence at Nazarbayev University.

The project took almost a year. Over this time, the artists have narrated many words, sounds, letters and numbers, so that the digital Iseke and Raya could easily pronounce common, unknown or even non-existent words. Several Kazakh companies have already showed their interest in the program.

The scientists shared not only the entire database, but also the program code and pre-trained models so that other inventors could use them for further development of Kazakh artificial intelligence technologies.