TXT, SRT, VTT, JSON, CSV or Markdown?

The same three lines, six ways

The sample below is original text written for this guide, not a real video. Each block is produced by the same export code as the download buttons, so it is what a real file looks like in structure.

[00:00] First, write down the question you want the video to answer.
[00:04] Next, search the transcript for a distinctive phrase.
[00:08] Finally, check the matching moment in the source video.

First, write down the question you want the video to answer.
Next, search the transcript for a distinctive phrase.
Finally, check the matching moment in the source video.

1
00:00:00,000 --> 00:00:04,000
First, write down the question you want the video to answer.

2
00:00:04,000 --> 00:00:08,500
Next, search the transcript for a distinctive phrase.

3
00:00:08,500 --> 00:00:13,500
Finally, check the matching moment in the source video.

WEBVTT


00:00:00.000 --> 00:00:04.000
First, write down the question you want the video to answer.

00:00:04.000 --> 00:00:08.500
Next, search the transcript for a distinctive phrase.

00:00:08.500 --> 00:00:13.500
Finally, check the matching moment in the source video.

Start (Seconds),End (Seconds),Timestamp,Duration (s),Text
0.00,4.00,00:00,4.00,"First, write down the question you want the video to answer."
4.00,8.50,00:04,4.50,"Next, search the transcript for a distinctive phrase."
8.50,13.50,00:08,5.00,"Finally, check the matching moment in the source video."

**[00:00]** First, write down the question you want the video to answer.

**[00:04]** Next, search the transcript for a distinctive phrase.

**[00:08]** Finally, check the matching moment in the source video.

{
  "video": {
    "videoId": "VIDEO_ID",
    "title": "Example Video Title",
    "channelName": "Example Channel"
  },
  "language": "en",
  "total_segments": 3,
  "segments": [
    {
      "text": "First, write down the question you want the video to answer.",
      "start": 0,
      "duration": 4,
      "startFormatted": "00:00"
    },
    {
      "text": "Next, search the transcript for a distinctive phrase.",
      "start": 4,
      "duration": 4.5,
      "startFormatted": "00:04"
    },
    {
      "text": "Finally, check the matching moment in the source video.",
      "start": 8.5,
      "duration": 5,
      "startFormatted": "00:08"
    }
  ]
}

Notes on each format

TXT is the most portable. With timestamps each line starts with a time label; without them it is just the words, one caption line per row. For polished paragraphs, run it through the cleaner.

SRT numbers every cue and gives a start and end time. VTT carries the same timing under a WEBVTT header, with a period before the milliseconds, and is the format HTML5 video players understand. Both are meant for subtitle software, so inspect names, numbers and line lengths before publishing.

CSV gives every line its own row: start and end in seconds, a readable timestamp, duration and text. Text containing commas or quotes is escaped so spreadsheets read it correctly.

JSON is for programs. You get the language, a segment count and a list of segments with text, start and duration in seconds, and a formatted start time. A transcript request through the API returns structured data you can work with in the same way.

Markdown is plain text with light structure, convenient for note apps and for pasting into documents that render Markdown.

Choosing quickly

If you are unsure, start with TXT with timestamps: it is readable, searchable and keeps the time trail you may want later. Download another format only when the next tool needs it. To see all options in one place, use the transcript downloader, or go step by step with the download guide. Remember that the exports are only as accurate as the captions; see automatic versus creator captions.