Garbled text
Repair candidates
Paste garbled text and it shows readable candidates by guessing the original encoding — no upload.
"café", "été", "譁・ｭ怜喧" — text copied from an email, a CSV, or a web page comes out garbled and unreadable. This tool takes that garbled string and turns it back into readable text. Most mojibake happens when text that was really UTF-8 gets opened as a different encoding (Shift_JIS, EUC-JP, Windows-1252, Latin-1, and so on). This tool takes your garbled text and works backwards — "if this were opened as encoding X, the original bytes must have been these" — then re-reads those bytes as UTF-8 to produce repair candidates. It tries the Shift_JIS misread, the EUC-JP misread, and the Western (Latin-1 / Windows-1252) misread at the same time, and ranks the candidates by how readable they look. The top one is marked "Best", and the count of CJK characters is shown, so you can compare and copy the one that reads correctly. Under the hood, guessing the encoding, reversing it, and re-decoding to UTF-8 all happen inside your browser in JavaScript — the reverse lookup tables are built at runtime from the browser's own decoders, so no extra software or dictionary files are needed. Your pasted text is never uploaded, stored, or sent to a server, which makes it safe for internal email excerpts, customer lists, or source code you'd rather not paste into a cloud converter (and it suits workplaces where uploading externally is against policy). Note: characters that have already collapsed into "?" or "□" (a break where the original bytes are lost) can't be recovered in principle — for those, no candidate appears or part of the text stays broken. To convert a whole file's encoding, try encoding-convert; to fix a garbled CSV with a table preview, try csv-encoding.
How to use
- Copy your garbled text and paste it into the "Garbled text" box (no file to open, no app to install).
- As you paste, repair candidates — reversing Shift_JIS, EUC-JP and Western misreads — appear ranked by how readable they look. The top one is marked "Best".
- Click "Copy" on the candidate that reads correctly. Your text is never uploaded anywhere.
FAQ
Why does mojibake happen, and how does this tool fix it?
Most mojibake happens when text that was really UTF-8 is opened as a different encoding — Shift_JIS, EUC-JP, or Windows-1252. This tool works backwards: "if it were opened as that encoding, the original bytes must have been these," then re-reads those bytes as UTF-8 to restore it. It tries several misread patterns at once and ranks the candidates by readability.
Is my pasted text sent to a server?
No. Guessing the encoding, reversing it, and re-decoding all run on your device in JavaScript. Nothing is uploaded, stored, or transmitted, so it's safe for email excerpts, name lists, or source code you'd rather not paste into a cloud converter, and for workplaces that forbid uploading externally.
Can characters that became "?" or "□" be restored?
No. Once a character has collapsed into "?", "□", or "�", the original byte information is gone, so it can't be recovered in principle. This tool fixes mojibake where the bytes still exist but were read as the wrong encoding.
Does it handle Western mojibake too, not just Japanese?
Yes. It also handles Western mojibake like "café" or "été" (text misread as Latin-1 / Windows-1252). It tries the main Japanese and Western misread patterns at the same time and puts the most readable candidate at the top.
What if I need to fix a whole file or a CSV?
To re-convert a whole text file's encoding, use encoding-convert (UTF-8 / Shift_JIS / EUC-JP). To fix a garbled CSV with a table preview, use csv-encoding. This tool is best for quickly turning a garbled string you've pasted on screen back into readable text.