Paste garbled text, click one button. The tool systematically tries source byte reconstruction × target encodings, shows real‑time progress, and ranks results by readability. All in your browser – no data leaves your device.
Mojibake (文字化け) is garbled text caused by decoding bytes with a wrong character encoding. For example, the UTF‑8 bytes for “é” (0xC3 0xA9) when interpreted as Latin‑1 become “é”. This tool reverses the process by re‑encoding the garbled string back to bytes using the wrong encoding, then decoding those bytes with the correct encoding.
Auto Detect tries multiple common encoding pairs:
For each candidate, it computes a readability score based on the proportion of printable ASCII characters and the absence of the Unicode replacement character (�). The candidate with the highest score is highlighted. You can click any candidate to copy it or apply it as the manual result.
garbled string → treat as encoding‑A bytes → decode with encoding‑B → readable text
Example: é (bytes 0xC3 0xA9) decoded with Latin‑1 gives "é", but re‑decoding those bytes with UTF‑8 gives "é".
Verified The default example "Café, über, £100" now correctly repairs to "Café, über, £100" using UTF‑8 → ISO‑8859‑1.
| Correct encoding | Wrong encoding | Example (garbled → fixed) | Tool verified |
|---|---|---|---|
| UTF‑8 | ISO‑8859‑1 | Café → Café | ✅ Works |
| UTF‑8 | Windows‑1252 | € → € | ✅ Works |
| Windows‑1252 | UTF‑8 | € (but different) → “€” | ✅ Works |
| ISO‑8859‑1 | UTF‑8 | über → über | ✅ Works |
Character encodings map numbers to glyphs. When a file lacks encoding metadata (no BOM, wrong HTTP header, misconfigured database), the application guesses. Common causes:
Problem: A data analyst received a CSV file containing customer names with diacritics (e.g., “Müller”). Opening in Excel with default settings showed “Müller”.
Solution: Using this tool’s Auto Detect, the text “Müller” was correctly identified as UTF‑8 bytes misread as Latin‑1. The tool repaired it to “Müller”. The analyst then used the fix to clean the entire file with a script based on the same logic.
Outcome: Thousands of records restored without manual re‑typing.
Sometimes text is misinterpreted twice, e.g., UTF‑8 → Latin‑1, then that result is saved and later misinterpreted again as UTF‑8. Our tool can be applied iteratively. For example, “Müller” (double) can be fixed once (UTF‑8→Latin‑1) to “Müller”, then again to “Müller”. The Auto Detect may not directly resolve double mojibake, but you can run it twice.