SRT to TXT without losing speaker names or source times

A subtitle file can be a useful starting point for an AI video edit. It already contains the speaker’s words and references to moments in the recording. But converting that file to plain text can accidentally remove information you still need.

We shipped a free SRT and VTT to text converter at Valmera. It runs in the browser, requires no account, and exports a TXT file. Here are the decisions behind it—and how to use the result to plan an edit.

Disclosure: this article is published by Valmera, which makes the converter and an AI video editor.

1. Keep the words, including numbers and repetition

An SRT file uses numbers to identify subtitle cues. The dialogue can contain numbers too:

1
00:00:01,000 --> 00:00:03,000
Our 2026 launch starts with three tutorials.

2
00:00:04,500 --> 00:00:07,000
Three tutorials, each with a complete example.

The plain-text result should be:

Our 2026 launch starts with three tutorials. Three tutorials, each with a complete example.

Removing every numeric line or repeated phrase is a different editorial operation from extracting a transcript. A speaker might deliberately repeat an important point. Our converter keeps repeated speech instead of guessing what to delete.

Rolling captions can therefore produce repeated text in the result. Review those repetitions before using the transcript as a script or editing brief.

2. Preserve speaker labels when the file contains them

WebVTT can carry a speaker name in a voice annotation:

WEBVTT

00:01.000 --> 00:03.000
<v Maya>Hello &amp; welcome.</v>

Stripping every tag immediately would lose “Maya.” The converter first turns the voice annotation into a label, removes common caption styling, and decodes the text:

Maya: Hello & welcome.

That label is useful when selecting an interview answer or deciding which speaker needs to remain visible. The converter does not infer missing names; it preserves labels present in the file. The W3C WebVTT specification describes cue text, voice annotations, comments and styling blocks.

3. Make source timestamps optional

A continuous paragraph is convenient for reading. An editing brief often needs a way to locate a sentence in the original recording.

Turn on Keep start timestamps for editing references to produce text like this:

[00:00:01.000] Our 2026 launch starts with three tutorials.

[00:00:04.500] Three tutorials, each with a complete example.

These are the original cue start times. They are not newly calculated word timings, and they do not automatically follow a later video edit.

Before using a timestamp to request a cut, check that the subtitle file belongs to the same recording you uploaded. A subtitle export from an earlier edit may use a different clock. Quoting the exact phrase alongside the time makes your instruction easier to locate.

4. Report broken blocks instead of returning half a transcript

A converter should not quietly skip a malformed section and present the remaining text as complete. Our converter reports blocks with missing timing lines, invalid intervals, or missing separators between cues.

The implementation has 15 focused tests covering complete SRT conversion, WebVTT metadata, speaker labels, Unicode and entities, source timestamps, paragraph layouts, repeated speech, line endings, and malformed input. We also checked file loading and the contents of a downloaded TXT file in the browser.

This utility reads UTF-8 subtitle files up to 5 MB. It extracts existing subtitle text; it does not listen to audio or generate a missing transcript. The file is read and converted locally rather than uploaded for conversion.

Use the transcript to specify a reviewable edit

Open the converter, paste or load your subtitles, and choose a layout. Copy the passages you want to discuss, or download the transcript.

Then give your editor an outcome and the material that must survive. For example, replace the placeholders with real phrases from your recording:

Use the uploaded interview. Open with the complete answer beginning “[exact phrase]” at [original source time]. Keep the explanation beginning “[exact phrase]” and its qualification. Finish with the practical next step. Aim for a concise vertical clip with readable captions, then show me a preview so I can review the pacing.

The editing brief builder can help add the intended audience, destination, target length, protected material and caption preference. You can plan that brief with a conversational assistant, then use Valmera Studio or an authorized MCP connection to edit the video.

Review the rendered preview with sound. A readable transcript helps you choose the story; the video reveals whether the cuts, framing and pacing work.

Valmera—the world’s best AI video editor. Join thousands of creators. Come try Valmera. The subtitle converter is free; see current plans for video editing.