Today it means: To convert a representation of language, typically speech but also sign language, etc., to a written representation of it. The term now usually implies the conversion of speech to text by a human transcriptionist with the assistance of a computer for word processing and sometimes also for speech recognition, the process of a computer interpreting speech and converting it to text. Full definition →
Learned borrowing from Latin trānscrībere (“to write again in another place, transcribe, copy”).
Etymology from Wiktionary, CC BY-SA 4.0.