AI Subtitle Studio
Free tool

Transcript cleaner & converter

Paste a Zoom, Teams or YouTube transcript, or upload it, and get clean text back: timestamps, speaker labels, um and uh and sound tags gone, the lines joined into paragraphs. Or turn a timestamped transcript into an SRT or WebVTT subtitle file.

  • Timestamps, fillers and sound tags removed
  • Download as TXT, SRT or WebVTT
  • Nothing stored, no account needed
or
Options — what to clean and how to lay it out
Clean up
Layout
seconds

Cleaned in memory and sent straight back — your transcript is never saved. SRT and WebVTT output needs timestamps.

How it works

Paste a transcript, get it back clean

Paste or upload

Paste the text, or drop a .txt, .vtt, .srt or .sbv file. The layout is recognised from the file itself, so there is nothing to select.

Choose what to clean

Everything useful is on to begin with. Open Options to keep timestamps, drop speaker names, or switch a clean-up off.

Copy or download

Read the result on the page, copy it, or download it as plain text, or as SRT or WebVTT when the transcript has timestamps.

What it cleans

Remove timestamps, speaker labels and filler words

Fixed rules, not a language model: the same transcript with the same options gives the same result every time, and your wording is never rewritten.

TimestampsThe time at the start of a line, YouTube's separate 0:05 lines, and the cue numbers and arrows of caption files. Kept in front of each paragraph if you want them.
Speaker names"Alice Smith:", "Speaker 1 0:05", "[Alice Smith] 10:22:01" and WebVTT voice tags. Kept as a label on each turn, or removed.
Filler soundsum, uh, erm and hmm, with the comma around them. The next word is capitalised again if the filler opened a sentence.
Sound and note tags[laughter], [Music], [crosstalk], (applause) and lines of music notes, plus HTML tags such as <i>.
StuttersRepeated words ("I I think") and false starts ("th-the"). Off to begin with, because a repeat is sometimes what was said.
Encoding and typingText that has turned into ’ or é, curly quotes and long dashes brought back to plain ones, double spaces, and >> speaker markers.
ALL CAPSAuto-captions that arrive in capitals can be put into sentence case. Off to begin with.
Line breaksLines that were only wrapped are joined into paragraphs, which start again when the speaker changes, when a speaker pauses between sentences, or when a paragraph has grown too long.
What it reads

Transcripts from Zoom, Teams, YouTube and more

Four layouts cover the common exports. A time or a speaker is only taken as one when the rest of the file agrees, so a line like "see you at 10:30" stays a sentence.

WebVTT, SRT or SBV caption file

Zoom, Microsoft Teams and YouTube all export one.

00:00:01.000 --> 00:00:03.000
Alice Smith: Hello everyone, um welcome.

A time at the start of each line

Notes and exports with the time in square brackets, or YouTube's transcript copied from the page.

[00:00:01] Hello everyone, welcome.
[00:00:04] Let's get started.

A speaker header above each turn

Transcripts with the name and time on a line of their own.

Alice Smith  0:01
Hello everyone, welcome.

Bob Jones  0:09
Thanks Alice.

Plain text

No times at all, with or without "Name:" in front of what each person said.

Alice: Hello everyone.
Bob: Thanks Alice.
Convert

Turn a transcript into SRT or WebVTT

A transcript that carries timestamps already has what a subtitle file needs. Choose SRT or WebVTT as the output and each timestamp becomes the start of a subtitle. It ends when the next one begins, or sooner if the words have had long enough to read, so a pause in the talk leaves the screen empty.

Speech is longer than a subtitle can be, so anything that will not fit on two lines of about 42 characters is cut at sentence breaks into several subtitles that share the time in proportion to their length. A caption file you upload keeps its own start and end times and is only cleaned.

Going the other way, from subtitles to a plain transcript, works the same: upload the SRT or WebVTT file and take the plain text. For a subtitle file that needs more than its words tidied, use the subtitle cleanup tool, which also fixes timing, or the subtitles to plain text converter if you only want the dialogue.

FAQ

Questions about the transcript cleaner

How do I remove timestamps from a transcript?
Paste the transcript (or upload the .txt, .vtt, .srt or .sbv file) and choose "Plain text" with "Keep timestamps" off, which is how it starts. The times at the start of each line, YouTube's separate 0:05 lines and the cue numbers and arrows of caption files are removed, and the words are joined into paragraphs.
Can it take the speaker names out too?
Yes. Untick "Keep speaker names" and the transcript comes back as running text with no "Alice Smith:" in front of anything. Leave it ticked and each paragraph starts with who said it, with consecutive lines by the same person joined into one turn.
Which transcripts does it understand?
Caption files (WebVTT, SRT and SBV — including the WebVTT that Zoom and Microsoft Teams export, with "Name: text" lines or voice tags), text with a time at the start of each line ([00:01:23] text, or YouTube's 0:05 above the words), text with a speaker header above each turn ("Alice Smith 0:05", or "[Alice Smith] 10:22:01"), and plain text with or without "Name:" labels. It works out which one you gave it from the file itself.
Does it remove "um" and "uh"?
Yes, with "Remove filler sounds" on (it is by default): um, umm, uh, uhm, erm and hmm, along with the comma around them, and the next word gets its capital letter back if the filler opened a sentence. The list is English only — "um" is a real word in Portuguese, so turn the option off for a transcript in that language. Words like "ah", "oh" and "uh-huh" are answers and are never removed.
Can I turn a transcript into an SRT or WebVTT subtitle file?
Yes, if the transcript has timestamps. Choose SRT or WebVTT as the output. Each timestamp starts a subtitle, which ends when the next one begins or when it has had long enough to read at a comfortable speed, whichever is first. Anything too long for two lines is split at sentence breaks into several subtitles that share the time. Caption files you upload keep their own start and end times.
Why does it say my transcript has no timestamps?
SRT and WebVTT need a time on every subtitle, and plain text with no times has nothing to build them from. Choose plain text as the output instead, or use the export that includes timestamps (most meeting tools have one). If only some lines have a time, the ones without are left out of the SRT or WebVTT and you are told how many.
My Zoom transcript has times like 10:22:01. Will the subtitles be hours late?
Those are usually the time of day rather than the time into the recording, and the converter uses them as written. If the subtitle file starts hours in, move it back with the Shift Timing tool, which takes a negative offset.
Is my transcript stored?
No. The text is cleaned in memory while the page waits and sent straight back; it is never written to disk. We log that the tool was used and which options were chosen, never what was pasted or uploaded.
Is this AI?
No. It is a fixed set of rules, so the same transcript with the same options always gives the same result, instantly. It does not rewrite your wording or guess at what was meant. To turn a recording into a transcript in the first place, the desktop app transcribes audio and video on your own computer.
Is there a size limit or an API?
Files up to 5 MB, and pasted text up to the same size. There is also an API endpoint at /api/v1/tools/transcript; see the API reference.
Related

More free tools and guides

Subtitle cleanup

Strip SDH, credits and ads from a subtitle file and fix its timing.

Subtitles to plain text

Just the dialogue from an SRT, WebVTT or ASS file.

Shift timing

Move a whole subtitle file earlier or later, such as one that starts hours in.

Subtitle editor

Fix the lines of an SRT or WebVTT file one by one, in your browser.

Transcribe Zoom, Teams and Webex recordings

No transcript to clean yet? Make one from the recording, on your own computer.

Transcribe a YouTube video

Get a transcript of any video, then clean it up here.
Prefer not to upload your file?

The AI Subtitle Studio desktop app has this same Tools section built in — plus AI transcription, translation and batch processing for a whole season at once, all running locally on your Mac or PC.

Get the desktop app