🔤 Developer Tools

Character Encoding Converter

Fix garbled Japanese text (mojibake) by auto-detecting a file's character encoding and reading it back as UTF-8. It also converts and saves text as Shift_JIS, EUC-JP, ISO-2022-JP or UTF-8. Detection and conversion run entirely in your browser, so files are never uploaded.

① Fix mojibake (file)

Drop a text/CSV file here, or click to choose

.txt / .csv and similar — the encoding is detected automatically

② Convert & save text

Browsers use UTF-8 by default. Use this to write out Shift_JIS etc. for legacy software.

How to use

To fix a garbled file, drop a .txt / .csv into section ①. The encoding is auto-detected, the text is re-read as UTF-8, and you can copy or save it. To write text out in a specific encoding, use section ②.

  • Fix by auto-detect: detects the file's encoding (Shift_JIS / EUC-JP / ISO-2022-JP / UTF-8) and reads it correctly.
  • Or set it manually: if detection is off, pick the source encoding and reload the file.
  • Save as UTF-8: export the fixed text as modern UTF-8.
  • Reverse too: write UTF-8 text out as Shift_JIS etc. for older software.

Handy for

  • Shift_JIS CSV exported by the Japanese edition of Excel that shows mojibake elsewhere
  • Reading EUC-JP or ISO-2022-JP text from older systems or email
  • Converting UTF-8 text to Shift_JIS for legacy software that requires it

Everything runs on your device; files and text are never sent to a server. See also Full-width / Half-width Converter and CSV ⇔ JSON.

FAQ

Are the files or text I enter sent to a server?
No. Detection and conversion all happen in your browser; your files and text are never uploaded or stored. It works offline.
Why are Shift_JIS and EUC-JP still needed?
The web is mostly UTF-8 now, but CSV files exported by Japanese editions of Excel are often Shift_JIS, and older Linux systems and email frequently use EUC-JP or ISO-2022-JP. Opening those in a UTF-8 tool shows mojibake (garbled text); reading them with the right encoding restores the original.
Is auto-detection always correct?
It is right most of the time, but very short or symbol-only text can be misdetected. In that case, set the source encoding manually (for example Shift_JIS) and reload the file.