Work with transcriptsFree · No account needed

Transcript cleaner for SRT, VTT, and pasted text

Turn a transcript you already have into something you can read, quote, or study. Remove caption formatting, choose whether to keep timestamps, and preview the result before applying a change. Your text stays in this page.

Input Pasted transcript text · Local SRT, VTT, or TXT file · Formatting choicesOutput Cleaned TXT · Normalized SRT or VTT with original cue times · Preview and cleanup warnings

Your transcript

Choose a local subtitle or text file Or paste transcript text Input format

Choose the cleanup

Join caption line breaks into flowing text Remove recognized subtitle formatting tags Remove whole lines such as [Music] or ♪ Keep timestamps in TXT (including leading pasted markers) Include cue numbers in TXT

SRT and VTT downloads always retain original start and end times. Subtitle indexes are rebuilt; styles and positioning are omitted. TXT input never receives invented timings. With timestamps off, recognizable leading 0:05 or [0:05] markers are removed; review any time-like ratios at line starts.

Use a UTF-8 file up to 5 MiB, 2 million characters, and 10,000 cues or lines.

Cleaned TXT preview

Your cleaned transcript will appear here.

Preview shows up to 100,000 characters. Downloads include the full result. Your original remains in the input until you choose “Use TXT preview as input.”

How to use it

Paste your transcript or choose a local UTF-8 SRT, VTT, or TXT file. Automatic detection recognizes subtitle timing arrows and the WEBVTT header; choose TXT explicitly for literal timing-like prose.
Choose whether to join wrapped lines, remove recognized formatting tags, or remove whole non-speech lines such as [Music]. Keep timestamps if you need a reference back to the recording.
Preview the cleaned text. Download TXT for reading, or SRT/VTT when the input provides complete cue timings. Use the TXT preview as a new input only if you want to edit it further; Undo restores the previous source.

A small example

The example has three cues starting at 00:00:01.250, 00:00:05.000, and 00:00:09.125. Joining lines turns “Learning begins with” and “a good question.” into one sentence. Removing [Music] changes caption text without moving any surviving subtitle start or end time.

What to know

This tool does not retrieve captions from a video URL or create a transcript from audio. Supply text you are allowed to use.
Maximum: one 5 MiB UTF-8 file, 2 million characters, and 10,000 cues or text lines. The preview displays the first 100,000 characters; downloads include the full result.
SRT/VTT downloads require timed subtitle input. A pasted 0:05 marker is not an end time and is never used to invent a subtitle duration. Leading time-like markers are removed from TXT when Keep timestamps is off; enable it for ratios or times you mean to retain.
Normalized subtitle exports rebuild SRT indexes and omit cue identifiers, styles, regions, and positioning. Warnings identify omitted metadata or empty cues. Malformed timing blocks stop processing instead of silently disappearing.
Cleanup is structural, not editorial. It does not correct speech recognition errors, summarize, translate, or remove repeated rolling-caption text automatically. Only the latest applied cleanup can be undone.

References & format documentation

For assistants and developers

This is a browser tool with visible, labeled controls and local processing. Read the usage guide or machine-readable catalog for supported inputs, outputs, examples, and limits. No API key or server endpoint is available. A catalog listing is documentation, not automatic integration with every AI assistant.