Transcribe Persian audio to text

Persian recordings often come from journalists, researchers and families spread across several countries, so the recordings arrive from many kinds of devices: voice memos on phones, video calls, and archived radio. The language is written right to left in a script that joins letters, and the transcript follows that direction, with numbers and Latin words kept in their natural order inside each line.

Two details matter when you proofread. The first is the half space, a zero-width non-joiner that Persian writers use inside many compound words and verb forms. Word processors differ in how they display it, so look at your export in the application where the text will finally live. The second is that spoken Persian differs from formal written Persian: contractions and informal verb forms are common in conversation, and a transcript reflects what was said rather than what a textbook would write.

Tehran Persian, Dari and Tajik vary in vocabulary and accent, and accuracy varies with audio quality, accents and background noise. Use speaker labels to keep a multi-person conversation readable, and export SRT or VTT if you want right-to-left subtitles.