Dutch audio and video to text

Netherlands, Flanders and regional speech

Dutch in the Netherlands and in Flanders shares one written standard but differs in sound, intonation and everyday vocabulary, for example Flemish goesting. Both are normally recognised as Dutch, though softer Flemish speech and fast Randstad speech each cause the odd miss. Heavy dialects such as Limburgs, West-Vlaams and Gronings sit far from the standard and are often wrong. Frisian speech in a recording will not be recognised as Frisian, so check those stretches by ear.

Compounds, the ij digraph and spelling

Dutch writes compounds as one word (belastingaangifte, vergaderruimte), so the commonest edit is a wrongly split or joined compound. Also check the ij digraph, capital IJ in names such as IJsselmeer, diaeresis words like financiën, and apostrophes in forms such as foto's and 's ochtends. Formal u and informal je or jij follow the speaker. English loanwords may be spelled either way, so choose one when you edit.

What speakers of Dutch use it for

Students at universities and hogescholen record colleges and hoorcolleges to build notes. Researchers transcribe interviews for qualitative studies. Municipal and company teams record vergaderingen to write notulen. Podcasters and YouTubers add Dutch ondertitels. Belgian users often have recordings that mix Dutch and French, which are best split into one language per file.

Which model works best for Dutch

Dutch is one of the languages Whisper-family models have seen the most speech in, so the small model can give a reasonable first draft from clear audio. That is not a promise of accuracy: it still slips on names, numbers, jargon and overlapping or noisy speech, so proof-read anything you will publish or rely on.

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).

The small model handles clear Netherlands Dutch well. Flemish accents, dialects and meetings where several people interrupt each other are where errors cluster. Recordings that switch between Dutch and English inside a sentence get weaker at the switch.

Subtitles and captions in Dutch

Keep lines to roughly 42 characters, two lines at most, and break at clause boundaries. Long compounds cannot be split, so rephrase in your editor if one pushes a line past the limit. Check that names with IJ and words with a diaeresis display properly in your player; exports are UTF-8. Subtitles stay Dutch.

The uploader is set to Dutch

The upload box on this page transcribes as Dutch, 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.