The Foundation for the Preservation and Development of the Bashkir Language has launched a new project Bashkortsoft. It is the first of its kind to translate whole sentences from Bashkir into Russian, from Russian into Bashkir. The project was voiced by the famous Bashkir TV presenter Nail Yunusov. Bashkir-Russian and Russian-Bashkir translators are available on the web service: https://bashkortsoft.ru/.
Based on the parallel data collected in 2020, an ensemble of 8 neural networks was trained, demonstrating the best recognition quality. The accumulation and processing of materials available simultaneously in the Bashkir and Russian languages continue. Processing of news in two languages, published in 2021, is made. More than 14 thousand web pages were processed in total.
The recordings dubbed by the presenter were analyzed and processed by experts. The result of this work was a database ready for use in training a neural network synthesizer of Bashkir speech. A total of 19,892 sentences were processed with a total duration of 40 hours.
“A computer program capable of dubbing any text in the Bashkir language by the voice of Nail Yunusov was created due to the support of the Foundation for the Preservation and Development of the Bashkir Language. Anyone can test the synthesizer on the website of the Russian-Bashkir translator bashkortsoft.ru. The site users have the opportunity to listen to the translation of the text in the Bashkir language,” one of the site developers, Aidar Khusainov, commented.
The Foundation for the Preservation and Development of the Bashkir Language believes that the implementation of the project will contribute to the active development of information technologies in the Republic of Bashkortostan and contribute to the parity functioning of the state languages of the Republic of Bashkortostan.
Author: Leila Aralbaeva