Which export format should I use for my transcript?

Every serious transcription tool exports multiple formats — the question is which one you need. A short decision guide:

TXT (plain text) Unformatted prose, optionally with speaker labels and block timestamps. Use it for: reading, searching, quoting, pasting into other tools, quick scanning. Maximum compatibility — opens anywhere. No styling, no precise per-word timing.

SRT (SubRip) The classic subtitle format. Looks like: ``` 1 00:00:01,500 --> 00:00:05,000 Hello and welcome. ``` Use it for: YouTube (direct upload as captions), Vimeo, LinkedIn, Facebook, Final Cut Pro, Premiere, VLC. Universally accepted. No styling, no positioning. Comma as the millisecond separator.

VTT (WebVTT) The modern format for HTML5 video. Always starts with the line `WEBVTT`. Supports CSS styling, positioning, color changes. Use it for: <video> tags with <track kind="captions">, libraries like Video.js, Plyr, Mux Player. Period (not comma) as the millisecond separator.

JSON Structured data with everything: words, timestamps, confidence scores, speaker IDs, metadata. Example: ``` { "text": "Hello world", "words": [ { "word": "Hello", "start": 0.12, "end": 0.45, "confidence": 0.98, "speaker": 0 }, { "word": "world", "start": 0.45, "end": 0.78, "confidence": 0.96, "speaker": 0 } ] } ``` Use it for: your own apps, search indexes (Elasticsearch), custom players, Python/JS processing pipelines. Maximum flexibility, but you render it yourself.

DOCX Word document with formatting, speaker headings, timestamp columns. Use it for: academic research, journalism, handoff to non-technical clients. Not always emitted directly — often via Pandoc or by importing TXT/JSON into Word.

PDF For finalized, archival versions — not an editing format. Export after editing.

Rule of thumb If you only need one: JSON. Every other format can be generated from JSON. If you need a delivery format: TXT for humans, SRT for YouTube, VTT for the web.

DeepScript exports TXT, SRT, VTT, and JSON from a single transcription via one endpoint (`/v1/transcriptions/{id}/export?format=…`).