How to use it
- Get the transcript: on a computer, open the video on YouTube, press Show transcript below the description, select all the text in the panel and copy it. For your own videos you can download a caption file (.srt, .vtt or .sbv) in YouTube Studio under Subtitles.
- Paste the text or press Open a caption file.
- Choose the options: remove sound tags like [Music], keep a time at the start of each paragraph, and how long paragraphs should be.
- Copy the clean text, or download it as .txt. If the input had times, you can also download an .srt caption file.
What gets cleaned
Timestamps on their own lines, the spoken-time lines YouTube adds for screen readers ("1 minute, 5 seconds"), caption numbers, formatting tags and sound descriptions in square brackets. Auto-generated .vtt files repeat each line in the next caption; repeated lines are kept once.
What it can't fix
Automatic captions often have no punctuation and some wrong words. The cleaner joins the lines into readable paragraphs, but it doesn't guess full stops or correct words, so read the result before you publish it as a quote or a blog post.
Good uses
Turning your own video into a blog post or show notes, searching a long talk for one quote, making study notes, or checking what was said before you clip it. Someone else's words are still theirs: quote them with credit.
Questions
- Can it fetch the transcript from a link?
- No. YouTube doesn't offer a public way to read transcripts from a link, and tools that scrape them break often. Copying from the transcript panel takes a few seconds and always works.
- Is the text sent anywhere?
- No. The cleaning happens in your browser.
- How are reading and speaking time worked out?
- At about 238 words per minute for reading (the average for silent reading of non-fiction in a large 2019 review) and 150 words per minute for speaking, a typical presenting pace.