Last updated · Written by the RecordingToText team · How we work
The five formats at a glance
| Format | Extension | Timestamps | Speakers | Best for | Opens in |
|---|---|---|---|---|---|
| Plain text | .txt | Optional, inline ([1:05]) | Optional (Name: ) | Reading, quoting, notes | Anything |
| SubRip | .srt | Start and end per cue | Only as text (Name: ) | Video captions, widest support | YouTube, Premiere, DaVinci Resolve, CapCut, VLC |
| WebVTT | .vtt | Start and end per cue | Yes, <v Name> voice tags | Captions on websites (HTML5 <track>) | Browsers, Vimeo, YouTube, most web players |
| CSV | .csv | Start and end columns | Speaker column | Research, spreadsheets, coding interviews | Excel, Google Sheets, R, Python |
| JSON | .json | Any precision, down to words | Yes | Developers, search, custom players | Any programming language |
SRT (SubRip) explained
1 00:00:00,500 --> 00:00:03,400 Host: Thanks for joining us today.
- Cues are numbered from 1 and separated by a blank line.
- Timestamps are
hours:minutes:seconds,milliseconds, with a comma before the milliseconds. - There's no standard speaker field, so names are usually written into the text.
WebVTT explained
WEBVTT 00:00:00.500 --> 00:00:03.400 <v Host>Thanks for joining us today.
- The file must begin with
WEBVTT, be UTF-8 encoded and served astext/vtt. - Timestamps use a full stop before the milliseconds, and the hours are optional.
- The
<v Name>voice tag identifies the speaker. Cue settings can position captions, andNOTEblocks hold comments.
The single most common mistake is mixing the two separators. 00:00:03,400 is SRT and 00:00:03.400 is VTT; many players silently ignore cues with the wrong one.
Caption guidelines (with sources)
The Netflix English (USA) Timed Text Style Guide is one of the most widely referenced public caption standards:
| Rule | Value |
|---|---|
| Characters per line | 42 maximum |
| Lines per subtitle | 2 maximum |
| Reading speed, adult programs | Up to 20 characters per second |
| Reading speed, children's programs | Up to 17 characters per second |
| Minimum duration | 20 frames (about 0.83 s at 24 fps) |
| Gap between subtitles | At least 2 frames |
Sources: Netflix Partner Help Center, "English (USA) Timed Text Style Guide" and "Timed Text Style Guide: Subtitle Timing Guidelines"; W3C, "WebVTT: The Web Video Text Tracks Format". Checked October 2026.
Which format should you use?
- Article, report or meeting notes: TXT, or DOCX if you need formatting.
- Captions for a video file or an upload form: SRT.
- Captions on your own website: WebVTT, through the HTML
<track>element. - Research interviews: CSV, one row per segment with speaker and times.
- Software and search: JSON, ideally with word-level timestamps.
Don't have a transcript yet? Transcribe an audio recording to text for free: you get TXT and SRT, and Premium adds VTT, CSV, JSON with word timings, DOCX and PDF.
Open source
The converter on this page uses the open-source transcript-formats library (MIT license). It parses and writes all five formats and checks captions from the command line, and its sample transcripts are free to reuse under CC BY 4.0.
Cite this page
RecordingToText, "Subtitle and transcript formats, compared", https://recordingtotext.com/subtitle-formats
Frequently asked questions
What is the difference between SRT and VTT?
Both store numbered or timed caption cues. SRT uses a comma before milliseconds (00:00:03,400) and has no standard speaker field. WebVTT must start with the line WEBVTT, uses a full stop (00:00:03.400), allows positioning and styling, and can mark speakers with <v Name> voice tags. Use VTT for web players and SRT for editing software and most upload forms.
How do I convert SRT to VTT?
Paste the SRT or open the file in the converter on this page, choose WebVTT and press Convert. The converter adds the WEBVTT header, switches the millisecond separator and turns Name: prefixes into <v Name> voice tags. It runs in your browser, so the file isn't uploaded.
How many characters per line should subtitles have?
A common standard is the Netflix English (USA) Timed Text Style Guide: at most 42 characters per line and two lines per subtitle, with a reading speed up to 20 characters per second for adult programs (17 for children's programs).
Which transcript format should I use for research interviews?
CSV, with one row per segment and columns for start time, end time, speaker and text. It opens in Excel or Google Sheets and imports into R, Python and most qualitative analysis tools. Keep a TXT or DOCX copy for reading.
Can I get word-level timestamps?
Not in SRT, VTT or CSV, which store one time range per cue or segment. JSON can hold word-level timing. RecordingToText Premium exports JSON with a start and end time for every word.
Is the converter free and private?
Yes. It's free with no signup and runs entirely in your browser; your file never leaves your device. The code is open source under the MIT license.