A team of Belarusian specialists has launched the Sonora project, which pursues an ambitious goal: to save the Belarusian language from digital disappearance and teach AI to speak Belarusian naturally and fluently. To build the necessary foundation for this effort, the team has launched a fundraising campaign on the Gronka platform.
Of the €13,000 required, the project has already raised more than €3,000, and the campaign is continuing. However, the organisers stress that this is not about purchasing a finished product, but about investing in a foundation without which the Belarusian language risks being sidelined by technological progress and pushed out of the digital space of the future.
Project Philosophy: A Language in a World Built Around Voice
The modern world is rapidly moving towards voice control, AI assistants and interactive applications. If a language is not represented in these technologies, it gradually disappears from the everyday lives of future generations. The Sonora initiative, a text-to-speech model, emerged as a response to this challenge: to ensure that Belarusian is heard in modern technologies, including applications, audiobooks, education and artificial intelligence.
Without efforts from the community itself, global corporations are unlikely to invest in Belarusian AI development because the market is commercially small. That is why the creators of Sonora propose investing in the first and most important step — the creation of a high-quality studio audio dataset, a collection of professional recordings that will be used to train artificial intelligence.
Robot or Human: Why Quality Is Being Decided Now
The idea behind the project is not merely to teach a machine to read Belarusian letters. The goal is to preserve living speech, its melody, character and emotions. It is at the dataset creation stage that the future sound of Belarusian AI is determined: whether it will be a dry, mechanical robotic voice or natural human speech.
To achieve this, a multidisciplinary team has come together, including AI engineers responsible for the technical implementation and training of the model; linguists and language experts, among them linguist-engineer Uladzislau, who oversee linguistic accuracy; as well as audio specialists and project managers led by Hanna Maklakova.
This combination of science and technology will create a foundation that can later be used by any developer building Belarusian-language applications, navigation systems or voice assistants.
How the Funds Will Be Used
To implement the project over the next three to five months, the team requires financial support to cover specialist fees and equipment rental. The funds will be directed towards:
- Preparing a balanced text corpus (linguistic database);
- Professional studio voice recording, including the work of voice actors and sound engineers;
- Technical processing and annotation of completed audio data for AI training.
“By supporting Sonora, you are helping not just a project — you are helping the Belarusian language find its voice in a world that is already being built around speech and artificial intelligence,” the creators say.
The project’s detailed concept and vision can be explored on the Sonora website at sonora.you/#idea, while support for the digital future of the language can be provided through the fundraising page.
