From a dictionary
to a voice

Crimean Tatar is on the endangered-language lists, and it has almost no technology of its own: no usable machine translation, no speech recognition, no synthesis to simply listen to a text. Ana-Yurt™ AI is an attempt to walk the whole road — from a plain dictionary to a voice assistant you can actually talk to in the language. Speech came first for two reasons: the people whose living speech can still be recorded are not getting younger, and recognition and synthesis pull each other up — old recordings become text, text with audio becomes training material for a voice, and the voice makes the audiobooks that barely exist. None of this runs in a lab or on a server farm; it runs on one home machine with one good GPU, in the evenings. We publish results, audio samples and an honest account of what broke along the way — not the data or the weights. This site is the shop window of that work: listen to what already works and follow how it goes, while the long-form articles of the series are published on ana-yurt.com.