Transcripcions

Transcripcions are the written records of spoken language created by listening to audio or video material and converting it into text. The term is used in Catalan-speaking contexts as the plural of transcripció. Transcriptions serve to capture what was said in a durable, searchable form and are distinct from translations, which render content into another language, whereas transcriptions render the original speech.

Types include verbatim transcripcions, which aim to reproduce all utterances, hesitations, fillers, and discourse markers; and cleaned or edited transcripcions, which remove excessive hesitations and nonessential noise. In linguistics, there are phonetic transcripcions that use specialized symbols to represent sounds, and orthographic transcripcions that reflect standard spelling. Subtypes exist for media production, such as subtitle-ready transcripts that include time codes and speaker tags.

Process and conventions: transcripcions are produced by listening to recordings, marking speaker changes, and inserting time codes. Common conventions include [hh:mm:ss] stamps, labels for each speaker, and indications of overlapped speech or inaudible segments. Transcripts can be stored in various formats, including plain text, SRT or WebVTT for subtitles, and specialized corpus formats for research.

Applications and challenges: transcripcions support accessibility, research, journalism, legal and medical documentation, and content archiving. Challenges include achieving high accuracy, handling accents and noisy recordings, dealing with multiple speakers, and respecting privacy and copyright requirements when working with sensitive material.