Full report JSON
How it works
Mojibake happens when bytes are decoded with the wrong character encoding. For example, UTF-8 café can appear as café under Windows-1252. Repair reconstructs those bytes and decodes them correctly; it cannot recreate information already replaced by �.
What you can do
Choose UTF-8, Windows-1252, literal Latin-1, Shift_JIS, GB18030, Big5, EUC-KR or Mac Roman. Inspect every pass and its bytes. Reverse simulation helps reproduce an encoding mistake.
How to use it
- Paste the garbled text. Choose the encoding that displayed it incorrectly and the original encoding; the sample uses Windows-1252 and UTF-8.
- Run one pass first. Increase to two or three only for repeated encoding mistakes. Review the byte report before copying or downloading.
Errors and limits
Only exact round trips are accepted. Correctly encoded text may fail a repair path; this is not an encoding detector. The imported file is read as UTF-8 text, not as original legacy bytes.
Up to 10,000 UTF-16 input units, 250,000 output units, a 1 MiB UTF-8 file and 20 seconds per run. Reports use machine-readable JSON field names.
Frequently asked questions
Why can’t this recover a replacement character?
U+FFFD often replaces an unknown byte sequence. Many original sequences produce the same replacement, so the original file or bytes are needed. A reversible conversion is still not proof that the chosen language is correct.
Stays in your browser?
Files and text are processed in this tab.