Ukrainian startup collecting recordings of native speakers of the Crimean Tatar language

The Ukrainian startup Respeecher is collecting  recordings of native speakers of the Crimean Tatar language as part of a project aimed at accelerating the appearance of the language in phone assistants, chatbots and automatic translators.

According to post from the Facebook page of the department of the president’s office focused on Crimea, the recordings will be used to train a “neural network” or a kind of artificial intelligence that in this case, will be used to teach computers how to process and understand speech and text in Crimean.   

The post states that Respeecher, which develops voice cloning for movie studios and game developers,  will not store data tied to any individuals without their permission and will not use the recordings. All data submitted will be used only for training the neural network  and for the analysis of the Crimean Tatar language in general.

Those willing to join the initiative can send audio recordings through this  Google form in Ukranian. The recordings should be done in environments with minimal background noise, ideally with the use of a microphone (though a good smartphone will work), and should be between 3o and 60 minutes in length.  Participants in Kyiv  can also make recording at the Respeecher studio.

The company is aiming to get as many participants as possible because the greater the diversity of accents and vocal tones the model can analyze, the more precise it will be in terms of language recognition.

RELATED: Crimean Tatar publishing house to be established in Ukraine

The Crimean Tatar language is one of the Turkic languages and is the mother tongue of Crimean Tatars. It is widespread in Crimea, mainland Ukraine, some parts of Turkey and the Republic of Uzbekistan and other countries where Crimean Tatars live. After the deportation of the Crimean Tatars from Crimea by Stalin in the 1940s, the language was ranked as severely endangered by UNESCO.

Image: Respeecher