Skip to video link
New: read subtitled videos without signing in · One-off sign-up bonus: 30 minutes + 10 AI tasks; carries over until used.Read free · Join for 30 min + 10 AI
Loading your workspace…
WELCOME TO VIDLEAF

Sign in to your workspace

or

YOUR VIDEO LIBRARY

My tasks and history

YOUR ACCOUNT

My account

Opening the reading link…
Knowledge in every video, within reach

Long videos, read the highlights.

Find useful ideas in long videos. Take the original words, translations and sources with you. Paste a link to a Bilibili or YouTube video and read it.

Free subtitle readingNo sign-in neededBilibili · YouTube
Current:Change link
Check subtitles first · Ask before transcribing

Guests have limited daily reads. After registration, existing subtitles are unlimited; transcription and AI use allowances.

Reading options
Upload local audio or video
Drop audio or video here, or click to choose a file
Processing details0 entries
No phone numberReading pages are private and not indexedNo ads
Compression is Intelligence, Part 1: Reinventing Entropy32:20 · Transcribed · 217 lines · EN / ZHOpen the full read →
00:00:44This question dates back at least to the 1940s, with Claude Shannon's seminal work…
00:22:11…Shannon analyzed a different model of the English language…: his wife Betty.
00:22:25As the story goes, Shannon pulled out a book and began asking Betty to guess each next letter.
00:30:21…he estimated the entropy of English to be about one bit per character.
Summary AI-generated

Start with the big picture

  • Compression and prediction are mathematically equivalent. 00:59
  • Robot instructions: 1.75 bits per instruction at best. 06:00
Claude ShannonInformation theoryCross-entropy
Chapters 16 chapters

↓

Source: Compression is Intelligence, Part 1: Reinventing Entropy · 3Blue1Brown
From videos already read on this site · abridged, summary translated

Follow each voice in an interview

Separate the voices, then filter the conversation by speaker.

How it works

Three steps: turn an hour-long video into a ten-minute read

No extension, no downloads. A video that already has subtitles is readable in seconds; for one without, you can choose transcription.

1

Paste a link

A Bilibili or YouTube video URL works. Vidleaf first detects the platform, the length and whether subtitles exist, then decides how to handle it.

https://www.youtube.com/watch?v=zxRgfBXn6Mg
Detected: YouTube · 22:00 · English subtitlesFree to read
Optional: translate · summarize1 AI call each
No subtitles? Transcribe 22 minOnly after you confirm
2

Read it through, or start with the summary

Subtitles are laid out in time order, side by side in two languages, with one-click Simplified/Traditional switching. In a hurry? Read the summary and chapters first and decide whether the detail is worth it.

Shao Allen × Justin Sun: In-depth Conversation4:22:30 · 68 chapters
06AI-assisted fitness and data management
07Starting at twenty without connections or resources
08AI advice for young people on limited budgets
3

Find it, tick it, export it

Search a word and jump straight to that line; click a timestamp to go back to the video. Tick the lines you need and export them with their timestamps.

04:25F-117 stealth aircraft cross the border, heading for Baghdad
05:04The stealth bombers go undetected, circling over Baghdad
05:25Tomahawk missiles launch against communication targets
2 of 625 lines selected↓ Selected SRT
Where it fits

A long video worth watchingis worth reading first

Lectures, interviews, launches and tutorials: read the transcript to find the part you need.

Podcasts and interviews

See who said what and cite their words with timestamps.

Reading language courses

Compare Chinese and English line by line; click timestamps to listen again.

Lectures and talks

Read the summary, then jump to the concept you need.

Tech talks and tutorials

Copy commands, flags and version numbers from the transcript.

Launches and keynotes

Read the key points, then decide which part to watch in depth.

Research and citation

Export selected lines with timestamps for your notes and references.

What you get

Read first. Process more when you need it.

Every feature is labelled with the quota it uses. Reading existing subtitles, searching, script switching and exporting are never metered.

Full subtitle reading

Free · no account

Read existing subtitles in order, navigate by chapter and switch between Simplified and Traditional Chinese.

Keyword search and timestamp jumps

Free

Search and highlight matching text; click timestamps to return to the video.

Summary

AI calls

Read an overview and chapter highlights with timestamps. One use per video, regardless of length.

Translation and bilingual view

AI calls

Compare Chinese and English subtitles line by line in one view.

Transcription for videos without subtitles

Transcription minutes

Transcribe audio into timestamped text when subtitles are unavailable. Review the minutes required before submitting.

Speaker labels and detection

Signed in · transcription minutes

Identify speakers automatically, review inferred labels and edit them manually.

Also: tick-and-export, history (once signed in), and live job progress that survives closing the page and coming back.

Capabilities and limits

Keep the source. Check what matters.

Subtitles

Platform subtitles

Human or automatic platform subtitles may omit lines. Sources and timestamps are preserved.

Transcription

Audio transcription

Audio-to-text may contain word or punctuation errors. Usage is counted in audio minutes.

Translation

Machine translation

Chinese–English translations need checking against the original. Usage counts as AI tasks.

Speakers

Speaker suggestions

Automatic voice separation needs human review. You can edit and merge speaker labels.

Free quota

Read captions free,with the quota spelled out

Transcription, speaker detection and AI are metered separately. Review each estimate before confirming purchased use.

No account
¥0

Open the page and start. Good for a first look.

  • Read subtitled videos, a few times a day
  • Search, script switching, tick-and-export
  • No transcription, summary or translation
  • No history kept
Start now
Plans & purchase

Pack S: 600 transcription minutes. Regular US$4.99 → US$2.49 for eligible first orders after registration. One-time purchase; no automatic renewal. See all allowances and terms before paying.

See plans & purchase

Base allowances reset Monday 00:00 Beijing time and do not accumulate. Sign-up bonuses carry over until used. When new processing is paused, saved content remains readable. Changes to allowances will be announced on the site in advance.

Questions

Before you start,you may want to know

Not answered here? There is a contact address in the footer, and the guide covers more.

Can I use it without an account?

Yes. Read, search and export subtitled videos without signing in, with a daily limit on new reads. Signed-in users can read subtitled videos without a limit, use transcription, summaries and translation within their allowances, and save history and personal labels. Email code or Google; no phone number needed.

Which platforms are supported?

Bilibili and YouTube for now. Paste the video page URL; BV-number links and youtu.be short links are both recognized.

Can I open a video in one step while watching it?

In the video's address, replace youtube.com or bilibili.com with vidleaf.app and press Enter to open it in Vidleaf.

What if the video has no subtitles?

When a completed subtitle check finds no subtitles, you may confirm transcription. Weekly allowance: 30 min / week; new transcription limit: 120 min / video. The confirmation shows duration and available minutes. If duration is unknown, actual audio duration is counted.

How much is free, and when does it reset?

Guests have limited daily reads. After registration, existing subtitles are unlimited; transcription and AI use allowances. Base allowances: 30 min / week and 10 / week, reset Monday 00:00 Beijing time. The one-off bonus (30 minutes + 10 AI tasks) carries over until consumed.

Will it cost money later? Will there be ads?

Reading existing subtitles is free. S packs and single passes are one-time purchases with no automatic renewal. View products in your account; prices, entitlements and refund terms are shown before purchase. We run no ads.

Can anyone else see what I read or summarized?

Reading pages are not indexed by search engines. Your jobs, history and personal labels are visible only to you. Subtitles, translations and summaries for public videos can be reused from cache; clearly marked homepage examples also show saved speaker detection. Uploaded content is not shared across accounts.

Separately, the operator publishes a small number of public transcript pages (/en/transcripts/) for YouTube videos their creators released under CC BY; they are unrelated to your tasks.

Where does my audio go during transcription?

Transcription, speaker detection and summaries are done by third-party AI services, so audio leaves this site. We ask for your consent separately before you first use them, and the outgoing requests carry no account information. The providers and retention periods are in the privacy policy.

Are the summaries accurate?

Summaries are AI-generated and can be skewed or incomplete. That is why every point carries its source time and the full subtitles stay next to it — for anything important, click back to the video and check.

Hand your next long videoto Vidleaf

Paste a video link to start reading. One-off sign-up bonus: 30 minutes + 10 AI tasks; carries over until used.

Video notes

Loading video…

Open original video ↗

See the video through its words

Speaker labels and settings

Automatic detection

Automatically label speakers when detection finishes. Your edits are kept.

View detection details

Speaker numbers are assigned separately within each part and do not represent the actual number of people. After applying, merge matching people below.

Edit speakers below, or select lines to correct them together.

Edit speakers

Add or edit people, then assign them to lines.

Add a speaker
Load subtitles to label speakers.
Merge labels for the same person

Choose the person to keep, then select other labels for that same person. This updates captions on every page and keeps the target person's name and role.

Select labels for this same person

Group labels for the same person

Compare voice samples across parts to suggest groups. Speaker labels are merged only after you confirm. A detection charge may apply.

Choosing a person next to a line saves automatically and changes only that line. Select checkboxes to update several lines together.

0 selected · Select sentences to copy or export0 lines selected
↓ Selected TXT↓ Selected SRT
Selected actions include only checked lines; All TXT/SRT exports the full transcript. Exports follow the translation view. Click a timestamp to open the video.
Enter keywords to find related content in the video.
TimestampKeywordsMatching content
Click a result to see its transcript context, or a timestamp to open the original video.
Vidleaf on PeerPush
Featured on There's An AI For That

Audio leaves your region

Transcription and speaker detection send your audio to the providers below, outside mainland China. This needs one separate confirmation, apart from the terms you accepted at sign-up.

Traffic statistics

Traffic statistics record browsing, actions and public IPs by default and link signed-in accounts automatically. You can disable IPs, actions or all statistics without affecting site features. Records are kept long-term and can be deleted; captions and passwords are not collected.

Notify me when the top-up pack opens

Leave an email and we will send one message when the top-up pack opens. Used for that only, nothing else.

Quick start

  1. Read a video:Paste a Bilibili or YouTube link and select Read video. Search for keywords or select a timestamp to return to the video.
  2. Summaries and translation: Generate key points under Summary. Choose English to Chinese or Chinese to English to read the original, translation, or both.
  3. Selection and people: Select lines to copy or download selected TXT or SRT. To export the full transcript, open Export and choose All TXT/SRT without selecting lines. Assign a person next to one line, or select several lines to label together.
  4. Continue later: Open My tasks to check progress, retry a failed task, or continue reading saved results.

Change the interface language at the top right. Change subtitle or summary language in the reading area. Saved results are reused when available.