跳到主要内容
PPDF123

Extract Text From a Scanned PDF

OCR PDF to Text recognizes the text on scanned pages in the languages you choose and returns a plain .txt file. Pages that already contain text are copied as they are, and you can add a heading before each page.

把文件拖到这里，或点击选择

支持 PDF · 单个文件不超过 500.0 MB · 一次最多 20 个文件

也可以按 Ctrl/Cmd+V 或从剪贴板菜单粘贴文件。

文档语言
最多选择 4 种。语言越少,识别越快越准。

OCR PDF to Text reads the pages of a PDF and returns a plain .txt file. Pages that have no usable text layer are rendered and recognized with Tesseract in the languages you pick; pages that already contain text are copied from the file, so a mixed document is not recognized twice.

Pick up to four languages that appear on the pages, choose whether each page gets a heading line such as --- Page 3 ---, and optionally restrict the run to a page range. Force OCR recognizes every selected page instead of trusting the existing text layer.

Limits are stated up front: up to 50 pages that need OCR per request on this site, counted after skipping pages that already have text, and files up to 500 MB. Layout is not preserved: columns and tables come out as running text, so use the table tool when the structure matters.

功能

  • Plain .txt output; text layers are reused and scans are recognized
  • Up to four recognition languages at once
  • Optional heading before each page
  • Optional page range
  • No account or sign-in required

适用场景

  • Pulling the words out of a scanned letter, report or ID document
  • Feeding scanned pages to a search index or an AI workflow as text
  • Copying a few paragraphs from a photographed page without retyping

How do I extract text from a scanned PDF?

  1. Upload the scanned PDF
  2. Choose the document languages
  3. Decide whether to add a heading before each page, and enter a page range if needed
  4. Click Process and download the .txt file

限制与边界

  • At most 50 pages that need OCR per request on this site; self-hosted servers can change it
  • Columns, tables and reading order are not reconstructed
  • Accuracy depends on scan quality and on picking the right languages
  • Upload limit on this website: 500 MB per file, sent in chunks above about 95 MB. A single direct API request body is capped at 100 MB.

常见问题

Why does my text look jumbled?
OCR reads pages line by line. Multi-column layouts, tables and sidebars can interleave. Use a page range for single-column pages, or the table tool for tabular data.
Is text that is already in the PDF copied or recognized again?
Copied. In Normal mode a page with a usable text layer is taken as it is; only image-only pages are recognized.
Can I get Markdown instead?
Yes. OCR PDF with the Markdown result returns a .md file that keeps headings and lists where it can detect them.
What happens with a very long scan?
A request that needs more than the page limit is rejected with a clear message. Enter a page range and run the document in parts.

最后更新:

用代码调用

本站每个工具都是普通 REST 接口。匿名使用不需要账号或 API Key，给浏览器、开发者和智能体同样可用。

curl -X POST "https://pdf123.xyz/api/v1/misc/ocr-text" \
  -F "[email protected]" \
  -F "languages=eng" \
  -F "ocrType=Normal" \
  -F "pageMarkers=false" \
  -F "pageNumbers=all" \
  -o output.pdf

也提供 MCP 工具，供说 Model Context Protocol 的客户端调用（JSON-RPC 2.0，POST /mcp）。 完整 API 文档

在 AI 代理中使用

Skill

装上这个 Skill，Claude Code、Codex、Cursor 等 AI 代理就能替你完成「OCR PDF 转文本」： /skills/pdf123-ocr-text.md

全部 Agent Skill

文件只用于本次处理，完成后自动删除。