Transcript File Formats Explained: TXT vs SRT vs VTT vs CSV
What TXT, SRT, VTT, Markdown, JSON and CSV are each actually for, so you export the one your next step needs instead of guessing.
The Scriptorium team ·
Pick the format by what happens to the file next, not by which one sounds most complete. A video editor wants SRT or VTT. A word processor wants plain text or Markdown. A spreadsheet wants CSV. Sending the wrong one is the most common reason an export looks broken when it opened perfectly well — it just was not built to open in that program. video to text and audio to text both offer every format below on the same transcript, so switching later costs nothing.
Plain text (.txt)
Just the words, no timing. The right choice for pasting into an email, a document, or anywhere the reader only needs to read what was said. Two variants exist: plain and timestamped — the timestamped version adds a time marker before each line, useful for a transcript you will refer back to by ear without needing full subtitle formatting.
SRT and VTT — for video players and editors
Both are subtitle formats: numbered blocks of text, each with a start and end time. SRT (SubRip) is the older, more universally supported format and the safe default for a video editor or an upload form that only names one format. VTT (WebVTT) is the web standard — what an HTML5 <video> element expects — and supports a little more styling. If a platform accepts either, it makes essentially no difference which you pick; if it only names one, use that one.
Markdown (.md) — a readable report
A formatted document: headings, paragraphs, speaker labels where available, and — when the transcript has them — the chapters and keywords generated from it. This is the export worth choosing when the transcript itself is the deliverable, not an input to something else — sharing it with a colleague, publishing it, or keeping it as a readable record.
JSON (.json) — for a program to read
The complete structured data: every segment, its exact timing, and word-level timestamps where they were generated. This is for feeding the transcript into another piece of software — not something to open and read directly, and the only format here aimed at a developer rather than a reader.
CSV (.csv) — for a spreadsheet
One row per segment, with its timing and text in separate columns. Useful for sorting, filtering, or counting things about a transcript in a spreadsheet — how long each speaker talked, how many segments fall in a given time range — none of which a spreadsheet can do with a plain text file.
If you are not sure
SRT is the closest thing to a universal default among the timed formats: most video tools accept it, and if a platform's upload form does not say which subtitle format it wants, SRT is the reasonable first attempt. For anything that is not going into a video player, plain text is the same kind of safe default.