Hindi, Tamil or Bengali text shows ठcharacters? Fix it here

What the broken text looks like

Every Indian script has its own fingerprint. Each letter becomes three characters, and the first two are always the same pair for a script – so you can tell the language even from the broken text:

• Hindi, Marathi, Nepali (Devanagari): à¤ and à¥, e.g. नाम for नाम.
• Bengali and Assamese: à¦ and à§, e.g. বাংলা for বাংলা.
• Tamil: à® and à¯, Telugu à° and à±, Kannada à² and à³, Malayalam à´ and àµ.
• Gujarati: àª and à«, Punjabi (Gurmukhi) à¨ and à©.
• Rupee sign: ₹ instead of ₹. The danda । becomes ।.

नाम,शहर,भाषा
सीता,कोलकाता,বাংলা
राम,भारत,हिंदी
கோவை,മലയാളം,₹1499

नाम,शहर,भाषा
सीता,कोलकाता,বাংলা
राम,भारत,हिंदी
கோவை,മലയാളം,₹1499

Why it happens

In UTF-8, a letter of an Indian script is stored as three bytes – ह is E0 A4 B9. A program that expects Windows-1252, the old Western character table, shows one character per byte: à, ¤ and ¹. The bytes are still all there, they are only read the wrong way. That is why Mojibuster can turn them back into the right letters.

The most common source is Excel: a CSV exported from Google Sheets, a web shop, a CRM or a government portal is UTF-8 without a BOM, the invisible marker at the start of a file. Excel then guesses Windows-1252. Older websites, WordPress databases with latin1 tables and e-mails with a wrong charset cause the same error.

How to fix a CSV file for Excel

1. Drop the CSV file into the box above. Mojibuster repairs the broken letters and shows a preview.
2. Keep Save with BOM switched on – it’s preselected for CSV files – and click Download.
3. Double-click the downloaded file – Excel now shows Hindi, Tamil or Bengali correctly.

If the file itself is fine and only Excel shows it wrongly, you can also import it: Data → From Text/CSV, then choose 65001: Unicode (UTF-8) under File Origin. More on this: fix special characters in Excel and fix CSV encoding.

How to save CSV files with Indian scripts correctly

• In Excel, choose Save as → CSV UTF-8 (Comma delimited). The plain “CSV (Comma delimited)” type uses the old Windows character table and replaces Hindi letters with ?.
• In Google Sheets, File → Download → CSV gives UTF-8 without a BOM. Run the file through Mojibuster with the BOM option before you open it in Excel.
• In databases, use utf8mb4 for tables and connections – see MySQL and phpMyAdmin encoding.

When the text can’t be repaired

If Excel or another program saved the file in the old Western character table, every Hindi letter was replaced by ?. The information is gone and can’t be restored – Mojibuster counts these spots so you know where to look. Go back to the original source and export it again as UTF-8. More: question marks instead of characters.

Some bytes – for example the virama ् and the vowel sign ु – turn into invisible control characters when read as Windows-1252. If a program removed them, Mojibuster works out the missing letter from its neighbours. Where both readings are possible, it leaves à¥ in place instead of guessing – repair the original file if you can: it still contains every byte.

Kruti Dev, Chanakya and other legacy fonts

Text typed in Kruti Dev, Chanakya, DevLys or Bamini looks like fgUnh when the font is missing – that is not an encoding error but a font that maps Latin letters to Hindi shapes. Mojibuster doesn’t convert these fonts yet. It repairs text that was saved as Unicode (UTF-8) and read the wrong way.

Background: what is mojibake? – and if the text shows Ã along with à¤, it was converted twice: fix double-encoded UTF-8.

Common broken characters

Frequently asked questions

More guides

• Mojibake fixer for all kinds of text Repair garbled characters in text and files.
• Fix CSV encoding Repair special characters in CSV exports and save as UTF-8.
• Excel characters When Excel shows é instead of é after opening a CSV.
• Fix UTF-8 Remove mojibake from databases, emails and websites.
• Fix subtitles Repair SRT and VTT subtitles with broken characters.
• What is mojibake? Meaning, origin and causes of garbled text.
• Double-encoded When é first became é and then é.
• é instead of é Table of é, ñ, ü and more – and what’s behind them.
• Apostrophes & quotes ’, “, – and … instead of apostrophes, quotes and dashes.
• Question marks instead of characters ? or � instead of letters – what can be rescued and what can’t.
• WordPress & database Repair SQL dumps and WordPress content showing é instead of é.
• MySQL & phpMyAdmin Set the charset of table, connection and import correctly.
• Outlook Fix broken characters in e-mails and contact exports.
• PowerShell -Encoding, console and script files in PowerShell 5.1 and 7.
• Python & pandas encoding in read_csv, to_csv, open() and requests.
• Notepad++ Choose or convert the encoding – the crucial difference.
• Customer data & CRM Repair customer lists from CRM and ERP without uploading them.
• Big5, GBK, Shift_JIS → UTF-8 Convert old Chinese, Japanese and Korean files to UTF-8.
• For businesses & data protection officers How Mojibuster protects data – and how to verify it yourself.