Skip to content
Translate PDF

Translate PDF Online Free

Automatic language detection  ·  133 target languages  ·  Urdu, Hindi & Arabic supported  ·  Word (.docx) download

Drop your PDF here

or pick a file from your device

Supported: .pdf  |  Max 100 MB

-
-
Detecting document language…
Popular translations — tap to set the target

Built-in translator · Urdu, Hindi, Arabic & 133 languages · download as Word (.docx) with layout preserved

Translation Complete

Your translated Word document is ready.

Translating…
Translating your document…
Large documents may take a minute or two.
How It Works
  1. Upload any text-based PDF
  2. We detect the source language automatically
  3. Choose a target — Urdu, Hindi and Arabic are one tap away
  4. We translate the text and give you a Word (.docx) that keeps the layout

Languages include
Urdu · Hindi · Arabic · Spanish · French · German · Chinese (Simplified & Traditional) · Japanese · Korean · Turkish · Bengali · Punjabi · Tamil · Persian · Pashto · Sindhi · Portuguese · Russian · and 115 more…

Best results with text-based PDFs. Scanned image PDFs require OCR first.

Runs on this app’s own translator (no AI sidecar). Text is sent only to the translation engine for the request.

Guide

Free PDF Translator — 133 Languages, Urdu & Arabic Included

Translate a PDF into any of 133 languages and download an editable Word document (DOCX) with the layout preserved where the text layer allows. PdfPeaks detects the source language automatically — including Urdu, Arabic, Hindi, Chinese, Japanese, Korean, Russian, Persian, Pashto, Sindhi and other non-Latin scripts — then converts the text so you can read, edit and share it. Free, no sign-up.

How It Works

Upload

Choose a PDF with extractable text when possible. The language is detected as soon as the file is selected.

Pick a target

Use the searchable list, a popular-pair shortcut (English → Urdu, Urdu → English, English → Hindi, English → Arabic), or accept the suggested target.

Download

Save the translated Word (.docx) file, then review formatting, names and numbers in the output.

Security & Privacy

Summaries are built on our servers from sentences in the extracted text. Translation runs on our servers when an offline language pair is installed; otherwise extracted text may go to a free translation service to finish the job. Text is not used to train models and is deleted after the job TTL — see our privacy details.

Common Use Cases

  • Immigration and visa paperwork: Understand foreign-language forms, letters and supporting documents.
  • Global business: Read contracts, reports and supplier emails from overseas offices.
  • Education: Share readings with multilingual students and parents.
  • Product support: Localize manuals and FAQs for new markets.
  • Personal: Read foreign-language letters, receipts and official notices.

Tips for Best Results

Machine translation can misread idioms, names, and legal terms — use a certified translator for official submissions. Output is DOCX (not always a PDF). After editing, convert back with Word to PDF. Guides: how to translate a PDF · Urdu ↔ English · English ↔ Hindi. Input: PDF · Output: DOCX.

Supported languages

Every target below uses Google's translation model. Detection covers the same set plus regional script variants that older detectors miss.

  • South Asian: Urdu, Hindi, Bengali, Punjabi, Gujarati, Marathi, Tamil, Telugu, Kannada, Malayalam, Nepali, Sinhala, Assamese, Odia, Sindhi, Sanskrit, Bhojpuri, Maithili, Dogri, Konkani, Pashto, Meiteilon (Manipuri), Mizo.
  • Middle East and Central Asia: Arabic, Persian (Farsi), Hebrew, Turkish, Azerbaijani, Kurdish (Kurmanji and Sorani), Uyghur, Kazakh, Uzbek, Turkmen, Tajik, Kyrgyz, Armenian, Georgian.
  • East and Southeast Asia: Chinese (Simplified and Traditional), Japanese, Korean, Vietnamese, Thai, Lao, Khmer, Burmese, Malay, Indonesian, Filipino, Javanese, Sundanese, Cebuano.
  • Europe: Spanish, French, German, Italian, Portuguese, Dutch, Polish, Romanian, Czech, Slovak, Hungarian, Greek, Swedish, Norwegian, Danish, Finnish, Ukrainian, Russian, Bulgarian, Serbian, Croatian, Bosnian, Slovenian, Macedonian, Belarusian, Lithuanian, Latvian, Estonian, Albanian.
  • Africa and others: Swahili, Amharic, Tigrinya, Hausa, Yoruba, Igbo, Zulu, Xhosa, Somali, Afrikaans, Malagasy, Shona, Sesotho, Krio, Lingala, Luganda, Oromo, Akan, Ewe, Bambara, Sepedi, Tsonga, Hawaiian, Maori, Samoan, Latin and more.

How language detection works

The detector reads the first pages of extracted text and identifies the writing system before guessing the language. Arabic script is separated into Urdu, Persian, Pashto, Sindhi, Uyghur, Sorani Kurdish and Arabic by letter shapes; Cyrillic is split into Russian, Ukrainian, Bulgarian, Serbian, Macedonian, Belarusian, Kazakh, Kyrgyz, Mongolian and Tajik; Devanagari is split into Hindi, Marathi and Nepali; Han characters are separated from Japanese kana and Korean Hangul, and Traditional Chinese is distinguished from Simplified. Latin-script languages then use statistical detection plus language-specific words for close pairs such as Malay/Indonesian, Turkish/Azerbaijani, Zulu/Swahili and Hausa/Swahili. The badge next to the detected language shows how the result was reached.

When detection can fail

Automatic detection is a convenience, not a guarantee. It struggles when:

  • The PDF is a scan: there is no text layer to read, so run Make PDF Searchable (OCR) first.
  • The text layer is broken: fonts without Unicode maps extract as symbols or question marks — common in old or print-to-PDF files.
  • Pages are very short: a title page with two words gives the detector little to work with; a multi-page document is more reliable.
  • The document mixes languages: quotes, references and tables in another language can shift the guess.
  • The script is shared: Urdu and Persian, or Malay and Indonesian, can look similar in short samples.

Translation accuracy and formatting

The Word output keeps paragraphs, headings and page flow where the PDF has a clean text layer. Tables, columns, stamps and text inside images are the usual problem areas: translate the text, then compare the result against the source before sharing it. Machine translation is suitable for understanding, drafts and internal use. For contracts, court filings, medical records or immigration submissions, have a qualified human translator review the result — names, dates, amounts and legal terms deserve a second pair of eyes.

Frequently Asked Questions

133 targets, including Urdu, Hindi, Arabic, Bengali, Spanish, French, Chinese, Japanese, Persian, Pashto and Sindhi. The target list is searchable — type a language name or code to jump to it.

English to Urdu, Urdu to English, English to Hindi, English to Arabic and English to Spanish are the most requested. Shortcut buttons for these pairs sit above the language list and set the target in one tap.

We read the first pages of extracted text and identify the writing system before guessing the language. Arabic-script variants (Urdu, Persian, Pashto, Sindhi, Uyghur, Sorani Kurdish), Cyrillic variants, Devanagari languages, Chinese, Japanese and Korean are separated by their letters; Latin-script languages use statistical detection plus language-specific words. The badge next to the result shows how it was reached.

Detection needs real text. Scans with no text layer, files whose fonts extract as symbols, very short documents and pages that mix languages can confuse it. You can still translate — choose the target language yourself; the source only affects the suggestion. For scans, run Make PDF Searchable (OCR) first.

That would change nothing, so the tool stops and asks for a different target. If a document is mostly English with a few Urdu quotations, choose English → Urdu to translate the whole file.

An editable Word document (.docx) with layout preserved where the text layer allows. It is not a translated PDF download — after edits, use Word to PDF if you need a PDF again.

We preserve layout in the DOCX where the text layer allows it. Tight tables and overlapping stamps often need a visual check. Official filings should be reviewed by a human.

Not until there is a text layer. Run Make PDF Searchable (OCR) first, then translate. Camera photos of pages fail more often than 300 DPI flat scans.

Use it to understand a draft. Certified or court work still needs a human translator. Treat names, numbers, and legal terms as things you verify in the source.

Urdu–English and other installed pairs are translated on our servers. If that offline pair is missing, extracted text (not the PDF) may go to a free translation service, then it is deleted. Details are in the Privacy Policy.

Yes for standard use.

Rate Translate PDF

Loading ratings…
Loading reviews…