Translate a Word or PDF document from English to Telugu

Upload an English Word file or a PDF with selectable text and get the same document back in Telugu: the headings, tables, numbering, header and footer where they were, and the dates, amounts and web addresses exactly as you wrote them.

English

Applications are invited under section 24(2) of the Housing Act, 2016. The last date is 31-03-2027 and the fee is Rs 250. For help write to help@example.gov.in or visit https://www.example.gov.in/apply.

Telugu, translated

గృహ నిర్మాణ చట్టం, 2016 లోని సెక్షన్ 24(2) క్రింద దరఖాస్తులు ఆహ్వానించబడ్డాయి. చివరి తేదీ 31-03-2027 మరియు రుసుము రూ. 250. సహాయం కోసం help@example.gov.in కు వ్రాయండి లేదా https://www.example.gov.in/apply ను సందర్శించండి.

A three-sentence notice written for our tests, as the English-to-Telugu translation returned it. The section reference, the date, the amount, the e-mail address and the web address all came back exactly as written.

Translate a Word file or a PDF with a text layer from English into Telugu. Machine translation, in beta; a free account with a confirmed e-mail address is needed.

Translate into Telugu

The document is edited in place, not rebuilt

A Word file is not turned into text and typeset again. Every paragraph style, indent, tab stop, table, list, header, footer and page setting stays the file’s own, and only the words change. That is why a notice with a heading, a centred bold date, a numbered list, a bordered table, tabs and a footer comes back looking like the notice you sent.

A PDF with a text layer is first converted to Word and then translated, so the result is a Word file you can edit. How closely it follows the PDF is the PDF-to-Word conversion’s doing, which is why a clean, digital PDF does best.

Numbers, dates and addresses are kept exactly

Before a sentence is sent for translation, its numbers, dates, section references, amounts, times of day, e-mail addresses and web addresses are replaced with markers, and put back afterwards. If a marker does not come back exactly once, the answer is not accepted. In the English-to-Telugu run of our test notice all 6 of 6 were back; across the layout check, 59 of 59 numbers, dates, references, addresses and e-mail addresses of the source were in the output.

A time such as 11:00AM is one item, not a number with letters left beside it, because an earlier version split it and the sentence came back half in English.

A sentence that cannot be translated stays in English, highlighted

If the engine’s answer fails a check, for instance because it dropped a marker or answered in the wrong script, the sentence is cut smaller and asked again, then asked again as written. A part that still cannot be translated stays in English and is highlighted yellow. It is never dropped, and never filled in with a guess, so a mixed paragraph is visible at once.

What to read carefully in the Telugu

Names of acts, schemes and offices are translated like any other words, and the engine may choose a different official name than the one your department uses. In our test, the Telugu for “Housing Act” read back as “Housing Development Act” when it was translated to English again. A Telugu reader found 7 of 7 passages of the first round acceptable, which is a small sample, not a promise.

The Telugu is real Unicode text, set in Noto Sans Telugu. If that font is not installed on your computer, choose any Telugu font in Word.

What it does not take

A scanned PDF with no text layer is not accepted yet: read it with Telugu OCR first. A document set in a legacy font is refused rather than translated from nonsense. Footnotes, endnotes and comments are not translated yet, and the result says so. Text in a picture is not translated.

What happens to each part of the file

Numbers, dates, sections, amounts, times, addresses
Kept exactly as written. A sentence where one comes back changed is not accepted.
Headings, lists, tables, header and footer
Kept: the file is edited in place. Header and footer text is translated too.
A sentence that could not be translated
Stays in English and is highlighted yellow. Counted in the report, never dropped.
A bold or coloured word inside a sentence
Takes the look of the sentence’s first run. A bold date or number keeps its own.
Footnotes, endnotes, comments
Not translated yet. The report says so.

Frequently asked questions

Can I get English and Telugu side by side?

Yes. Choose “Both languages” and each paragraph is followed by its translation, tables included. Headers and footers are translated only.

Will the layout stay the same if Telugu is longer than the English?

The page setup, styles and tables do not change, but text that needs more room than the English did will push later content down. In our test notice the page count was 1 before and 1 after. A long document may run to more pages, and the result reports it.

Can it translate a scanned PDF?

Not yet. A PDF has to have selectable text. For a scan, read it with Telugu OCR or English OCR first, correct it, and translate the Word file that comes out.

Is the translation a machine’s, and is it checked by a person?

It is machine translation, in beta. Before it was switched on, a Telugu reader and a Hindi reader read a first round of 16 passages: the Telugu reader found 7 of 7 English-to-Telugu passages acceptable, and the Hindi reader 5 of 6 English-to-Hindi, with one edit. That is a small sample, so read the result before you rely on it, and have it checked by a person for anything official or legal.

Do I need an account, and is there a limit?

Yes. Translation needs a free account with a confirmed e-mail address, and a limited number of pages a day and a month, counted in pages of 2,000 characters. The Limits page lists the numbers as they stand today.

What happens to my file?

It is stored encrypted while it is translated and removed by a cleanup that runs every 15 minutes, so nothing is kept longer than 4 hours after upload. The text is sent to an outside machine-translation service to be translated; the Privacy page says what is sent and what is not.

Related tools

More about translation