Some characters still look broken after one pass. — The text may have been mis-decoded twice. Choose the double-decode interpretation, which applies the repair a second time.
A few characters show as a black diamond question mark (�). — Those bytes were lost or replaced before you pasted them and cannot be recovered. Re-export the text from the original source as UTF-8.
Curly quotes and dashes are fixed but one odd symbol remains. — Try the Latin-1 interpretation instead of Windows-1252; the two differ only for a handful of punctuation and symbol bytes.
Frequently asked questions
What causes mojibake?
Mojibake happens when text saved in one character encoding is read in another. The classic case is UTF-8 bytes being decoded as Windows-1252 or Latin-1, so multi-byte characters like accents, smart quotes, and emoji split into several wrong single-byte glyphs.
A curly apostrophe (’, U+2019) is three UTF-8 bytes. Decoded as Windows-1252 they become â, €, and ™, which is why "don't" appears as "don’t". Re-decoding those three characters as UTF-8 rebuilds the original ’.
What is the difference between the Windows-1252 and Latin-1 fixes?
Windows-1252 and Latin-1 (ISO-8859-1) are almost identical but differ in the 0x80–0x9F range, where Windows-1252 holds smart quotes, dashes, and the euro sign. Try the Windows-1252 fix first; switch to Latin-1 if a punctuation or symbol character is still wrong.
Can mojibake always be repaired?
Often, but not always. If the original bytes were replaced with a generic placeholder (the � replacement character) before you copied the text, the information is gone and no fixer can recover it. The repair only works while the garbled-but-reversible bytes are intact.
Is the text I paste uploaded anywhere?
No. The mojibake repair runs entirely in your browser, so the text you paste is processed locally on your device and is not sent to a server.