Ukrainian audio and video to text

Dialects, Surzhyk and Russian mixing

Standard literary Ukrainian is the easiest case. Speech in the west (Galician, Hutsul, Transcarpathian), the centre and the east differs in vocabulary and sound, and strong dialect gives more errors. Surzhyk, the Russian-Ukrainian blend, and speakers who alternate between the two languages are hardest, because one sentence can hold words from both. Pick Ukrainian when it dominates and check the Russian stretches by ear. Diaspora Ukrainian with local loanwords behaves the same way.

Letters, apostrophes and quotation marks

Check the letters that set Ukrainian apart: і, ї, є and ґ. Watch the apostrophe in words such as п'ять and м'ясо, which can appear as a straight mark or a typographic one, and searching breaks if you mix them, so pick one when you edit. Names may follow older Russian-based spellings (Київ, not Киев), so check places and people. Ukrainian quotes are normally «…» with a dash for dialogue.

What speakers of Ukrainian use it for

Students and teachers record lectures (лекції) at universities and in online classes. Journalists, researchers and oral-history projects transcribe interviews. Volunteers and NGOs record meetings and briefings. Creators add Ukrainian captions to videos, and many people turn voice messages into text. Where testimony or interviews matter, a careful proof-read is essential, because names and places must be right.

Which model works best for Ukrainian

Ukrainian is supported with a good amount of training material, but less than the largest languages, so the small model makes more slips than it does for English or Spanish, especially on rare words, names and fast or noisy speech. Expect a draft that needs a proper read-through.

On tryscriptorium.com the no-account preview and regular accounts run the small model (the larger medium and large-v3 models run when you self-host Scriptorium). In an account, pick Small in the model menu rather than Auto: Auto prefers the fastest, least accurate model for long files or a busy queue.

Select Ukrainian, not auto-detect, whenever the recording is Ukrainian. The small model manages clear audio, but mixed Russian-Ukrainian speech, place names and phone-call quality are likely to need correction. For long or important recordings, proof-read the whole text.

Subtitles and captions in Ukrainian

Ukrainian subtitles read best at about 42 characters per line and two lines at most. Keep the apostrophe consistent, and confirm і, ї, є and ґ show in your player's font. SRT and VTT exports are UTF-8, so YouTube and VLC display Ukrainian correctly. Subtitles stay Ukrainian, so check any Russian-language stretches separately.

The uploader is set to Ukrainian

The upload box on this page transcribes as Ukrainian, so the recogniser does not have to guess the language from the first seconds. If the recording is in another language or switches languages, use Audio to text, which detects the language automatically. Automatic transcription is a draft and no accuracy figure is promised: check names, numbers and anything you will rely on.