Transcribe Norwegian Audio to Text — Free, No Upload

Norwegian is a smaller language for speech models than English or Spanish, and that shows up in a practical way: the lighter models make noticeably more spelling and word errors on it, while the largest one is close to usable without heavy editing. For that reason this page preselects the Best model instead of the default, and it fixes the language to Norwegian so that detection does not drift. The cost is time and memory: the Best model is the slowest, and on a phone it may be too heavy, in which case drop to Accurate.

Typical material is meeting recordings for minutes, lectures and seminar sessions, research interviews, podcast episodes and language-learning practice. The transcript comes out in Bokmål with æ, ø and å in place and ordinary punctuation, so it can be pasted into a document as a working draft. English words, which are common in Norwegian workplace speech, are usually kept as English. Review names, place names and figures, which are the usual trouble spots; and if your speakers are from the west or the north, expect to correct more words than for Oslo speech, since dialect remains the main source of errors. There is no per-minute limit (files up to 8 GB), and in the default Private mode the audio is handled in your browser without being uploaded.