Extension

Bilibili Subtitles to Markdown
Saving a video transcript as a local .md file

If a Bilibili video has AI subtitles, the NoteLoom extension pulls that text into a reading view, splits it into blocks (by the video's own chapters when it has them), and keeps a clickable timestamp on every segment. The clip button in the reading view header turns what you see into Markdown, and you save it as a .md file in a folder you choose. No selecting text, no copy-paste. Two limits up front: the subtitles have to exist on Bilibili's side, and what you get is speech written out word for word.

Who is actually reading this

Searching for this in English puts you in a small group. Mostly it is people learning Chinese who want the transcript next to the video, students abroad watching course recordings on Bilibili, and developers who read Chinese and treat the site as a technical video library. This page is written for those people rather than a general audience, because there is no general audience for Bilibili transcripts in English.

If you are on YouTube instead, the extension does the same thing there, with 30 second or chapter based segmentation and its own language bar.

Whose AI wrote those subtitles

This part gets confused often, so plainly: the AI subtitles belong to Bilibili. The platform runs speech recognition on an uploaded video and serves the result as a subtitle track, the same track you can turn on while watching. NoteLoom reads that track and formats it. Mishearings and mangled technical terms come from Bilibili's recognition, and NoteLoom passes them through unchanged.

So the quality of your saved notes is the quality of Bilibili's subtitles on that particular video. A clear studio talk transcribes well. A noisy vlog with three people talking over each other does not.

What saves well and what does not

The video What you get
A lecture or talk with Bilibili AI subtitles Works. Timestamped blocks, chapter headings when the video has chapters, and the full text
A video with no AI subtitle track Nothing to pull. The reading view has no transcript to show
A 分P series One part at a time, one file per part
A video with several subtitle languages Pick the track in the language bar before you clip
You want the key points, not the words Not what this does. The text is verbatim speech

The last row is the one that trips people up. A transcript is raw material. It is long and it repeats, and a 40 minute talk turns into several thousand characters of spoken sentences. That is useful when you want to search the words, quote a line, or read faster than the video plays. It is not a set of notes yet.

The steps

  • Install the extension in Chrome, Edge, or Arc, then open the Bilibili video you want.
  • Open the reading view, either from the toolbar popup or with Alt+R. On a video page the extension fetches the subtitle track instead of reading the page text, so what appears is the transcript, already broken into blocks with a timestamp on each one, following the video's own chapters when it has them.
  • Click a timestamp to jump the player to that second. The current segment highlights as the video plays, and the panel scrolls along with it. Ctrl+F inside the panel searches the whole transcript rather than the browser's find bar.
  • If the video has more than one subtitle language, pick the one you want in the language bar first. Clipping saves the track that is showing.
  • Click the clip button in the reading view header. The transcript converts to Markdown and opens in NoteLoom for a look.
  • Pick a local folder and save. You now have a plain .md file that any Markdown editor can open.

Switching to another video in the same tab reloads the transcript on its own, since Bilibili is a single page app and the extension follows the navigation.

The honest limits

  • No AI subtitle track on the video means no transcript. This is the most common reason it does not work, and nothing in the extension can create subtitles that Bilibili never made.
  • The text is a word for word transcription of speech. Um, so, you know, and every restarted sentence are in there. NoteLoom does not summarize, condense, or tag it, currently.
  • Timestamps are anchors to the video, so a saved file is most useful while you still have the video to jump back into.
  • Chromium browsers on the desktop only. Saving to a local folder needs the File System Access API, which Firefox, Safari, and mobile browsers do not fully support.
  • A 分P series gives you one file per part. Stitching them into one document is manual.

What to do with the file afterwards

The transcript lands as normal Markdown, so the cleanup happens in whatever editor you like. In NoteLoom that means the same folder you saved to: open the file, delete the filler, promote the lines that matter into headings, and write your own summary at the top while the video is still fresh. Naming the file with the date and the video title makes it findable later, which a folder of transcripts badly needs.

For language learners the raw text has a second use. Search it for a phrase you heard, read the sentence in context, then click the timestamp and hear it again.

Related reading

The NoteLoom browser extension Reading view, subtitle extraction, and clipping to Markdown. How to open a .md file What the saved transcript file is and what can read it. Taking notes with Markdown Turning a raw transcript into something you will reread. Online Markdown editor Open the folder of saved transcripts in the browser and edit them.

FAQ

Does every Bilibili video have subtitles I can save?
No. The transcript comes from the AI subtitles Bilibili itself generates for a video, and plenty of videos do not have them. If the video has no AI subtitle track, there is nothing to pull and the reading view has no transcript to show. Uploader-provided subtitle tracks and multiple languages show up in the language bar when they exist.
Do I need to be logged in to Bilibili?
Usually not. The AI subtitles come from a public Bilibili endpoint, and the extension fetches them from the video page itself using whatever cookies your browser already has. Some videos are restricted at the platform level, and those follow the same rules as watching them.
Will NoteLoom summarize the video for me?
Not currently. What you save is the subtitle text as spoken, cleaned into Markdown with timestamps, and chapter headings when the video has chapters. It reads like speech, filler words included. Cutting it down, adding a title, or writing your own takeaway is your job in the editor afterwards.
The subtitles say "AI subtitles". Whose AI is that?
Bilibili's. The platform runs speech recognition on the video and serves the result as a subtitle track. NoteLoom reads that track and formats it as Markdown. The transcription quality, the punctuation, and any mishearings all come from Bilibili's side.
What about a multi-part (分P) video?
Each part has its own subtitle track, so the extension handles them one part at a time. Switch to part 2 and the transcript reloads for part 2. That also means one clip gives you one .md file per part rather than the whole series in a single file.
Does the subtitle text get uploaded anywhere?
No. The fetch happens in the page from your browser, and the Markdown goes to a folder you pick on your own disk. There is no NoteLoom server in the path and no account to create.

Save a Bilibili transcript to your own folder

Add the extension to Chrome, Edge, or Arc, open a video that has AI subtitles, and clip the transcript into a .md file on your disk. Nothing is uploaded and there is no account.

Add NoteLoom to Chrome Or open the editor first