Czech audio and video to text

Common Czech and regional speech

Czech has a literary standard and a spoken form called common Czech (obecná čeština), where people say vokno or mladej instead of okno and mladý. A transcript may follow either, so choose a style and edit toward it. Moravian and Silesian speech are different again. Slovak is a separate language and is not covered by this page.

Diacritics and word endings

Czech letters like č, š, ž, ř, ě and ů are kept, and a wrong mark can produce a different word, such as být and byt (to be and flat). Nouns, adjectives and verbs change their endings by case, so look for a wrong ending when reading. Quotation marks in Czech normally open low, like „this“, though a transcript may use plain quotes.

What Czech speakers use it for

Lectures at Czech universities, work meetings, interviews, podcasts, and subtitles for videos that go on YouTube or social media. Short-form clips are also common, where fast captions are needed and a quick correction pass is enough. For a longer interview, check the names of people and firms first, because those are the words the recogniser has seen least.

Which model works best for Czech

Czech 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.

The small model is dependable for clear Czech speech but struggles with fast common Czech, names and loanwords. Because wrong diacritics and endings look plausible, read slowly while listening, or self-host with large-v3 for important material.

Subtitles and captions in Czech

Czech subtitles usually use two lines of about 40 to 42 characters, with a break at a natural phrase boundary. Decide whether to write spoken forms or standard forms, and keep it consistent. The SRT and VTT files are UTF-8, so every diacritic displays correctly.

The uploader is set to Czech

The upload box on this page transcribes as Czech, 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.