Get a Bilibili transcript

Paste a Bilibili URL above and the spoken audio comes back as timestamped text. Bilibili is the platform where "this video has subtitles" is least likely to mean "this video has a subtitle file": a meaningful share of its content — fan-subbed and reposted material above all — has the text burned into the video image, and real closed captions are inconsistent, especially for non-Mandarin audio.

It's also worth being explicit that danmaku is not a caption track. The scrolling comment overlay Bilibili is known for is viewer commentary, not the video's spoken content, so scraping it gives you an audience reaction rather than a transcript. Transcribing the audio is usually the only way to get machine-readable text out of a Bilibili video, including ones that visibly appear to have subtitles.

  • Handles the platform's long-form skew — lectures, commentary and review content well past 10 minutes.
  • Works on videos whose subtitles are burned into the image, where there is no file to extract.
  • Transcribes the audio in its source language; it does not translate the video.
  • JSON export keeps segment timings if you're indexing a backlog rather than reading one video.

Captions on Bilibili, as they actually are

Real closed captions are inconsistent, especially for non-Mandarin audio. Bilibili's signature "danmaku" — the scrolling comment overlay — is not a caption track; it's viewer commentary and doesn't contain the video's actual spoken content. A meaningful share of the platform's video (especially fan-subbed or reposted content) has subtitles hardcoded into the image instead of stored as a separate file.

Typical length
Skews long-form relative to TikTok/Instagram — lectures, documentary commentary, and long-form review content are common, often 10+ minutes.
URL shape
bilibili.com/video/BV...

What you're doing with the transcript

The export format and the workflow both depend on what the text is for. These go into detail for Bilibili specifically:

FAQ

The free tier includes 15 minutes of video a month, transcripts in the language spoken, and TXT/SRT/VTT/Markdown/JSON export — enough to run a Bilibili video through and judge the output before committing to anything.