Korean audio and video to text

Seoul speech and regional dialects

The standard of Seoul is the easiest input. Dialects such as Gyeongsang, Jeolla and especially Jeju are harder, because vocabulary and sentence endings differ, and the model tends to write them in a more standard way or misses words. North Korean vocabulary and pronunciation can also cause errors. Korean speakers often mix in English loanwords, which are written inconsistently.

Hangul, spacing and speech levels

Korean is written in Hangul with spaces between word groups, but the spacing rules (띄어쓰기) are subtle, so the transcript may join or split words differently from a style guide. The text records whatever speech level was spoken: 합쇼체, 해요체 or casual 반말. It does not convert between them.

What speakers of Korean use it for

Recorded university lectures for revision, meeting notes (회의록), interviews, church sermons, podcasts, and captions for YouTube videos or drama clips you own. Students often transcribe a lecture to search for a term before an exam, which is far faster than scrubbing through the audio.

Which model works best for Korean

Korean 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).

Korean is a well-supported language, and a single speaker with a good microphone gives a solid first draft on the small model. Expect corrections on names, English loanwords and word spacing. Fast panel discussions or noisy rooms benefit from self-hosting with large-v3.

Subtitles and captions in Korean

Korean subtitles are usually kept to around 16 to 20 characters per line, broken at word-group boundaries so particles stay with the word before them. Check the sentence endings if the tone matters for your audience. SRT and VTT files are UTF-8, so Hangul displays properly in VLC, YouTube and common editors. Spoken Korean is quick, so shorten lines where needed.

The uploader is set to Korean

The upload box on this page transcribes as Korean, 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.