Home Finance Health Math Converters Text Tools AI & Media Everyday Developer Tools Business Tools Blog About Contact

Mojibake Fixer

Paste garbled text — Café, broken CSV exports, mangled emails — and get readable text back. Auto-detects the wrong encoding.

100% private — repaired in your browser, works offline

Fix garbled text

Paste garbled text to begin
Seeing é where é should be? That's mojibake — and it's almost always reversible.

How it works

1

Paste the garbled text

Drop in text like "Café" from a broken CSV export, an old database, or a mangled email. Everything is processed in your browser — nothing is uploaded.

2

Auto-detect the mistake

The tool tries dozens of encoding pairs — re-encoding your text as the suspected wrong encoding, then decoding as the right one — and scores each candidate for readability.

3

Pick and export

The best repair is applied automatically. If it looks off, browse the ranked "try all" list or hit a one-click preset, then copy or download the fixed text.

Why text turns into gibberish

Computers store text as numbers (bytes), and an encoding is the agreed map between bytes and letters. UTF-8 is today's standard: the letter é is stored as two bytes, C3 A9. But if a program reads those same bytes using the older Latin-1 map — where every byte is one letter — C3 becomes à and A9 becomes ©, and "Café" turns into "Café". The bytes were never damaged; they were just read with the wrong map. That misreading is called mojibake (Japanese for "character transformation").

It strikes most often at boundaries between systems: CSV files exported from one tool and opened in another, data migrated from legacy databases, emails passing through misconfigured servers, and web scrapes saved with the wrong charset. Because the original bytes are usually intact, the fix is simply to reverse the misreading — encode the garbled text back into bytes using the wrong map, then decode those bytes with the right one.

Sometimes the damage is layered: text misread once, saved, then misread again ("Café"). The auto-detect here includes two-pass repairs for exactly that case, and if the automatic pick misses, the ranked list lets your eyes — the ultimate encoding detector — choose.

Recognising mojibake: the patterns and what they should be

Almost all mojibake follows one story: text was saved as UTF-8, then opened by a program that assumed Windows-1252. Because accented characters take two bytes in UTF-8, each one splits into two wrong characters. Once you know the common pairs, you can spot the problem at a glance:

  • é should be é — the classic "Café" on menus and in CSV exports
  • ’ should be a right single quote (’), and “ / †are the matching double quotes
  • ñ should be ñ, ü should be ü, and à should be à
  • … should be an ellipsis (…) and — an en dash

To stop it happening in the first place: in Excel, save spreadsheets as "CSV UTF-8" rather than plain CSV; when importing, use Data > From Text/CSV and set the encoding to UTF-8 explicitly. In databases and web pages, declare UTF-8 at every step — connection, storage and page <meta charset> — because a single step assuming a legacy encoding garbles everything downstream.

Pasted text breaks in two ways: garbled characters, fixed here — and mangled formatting, spacing and line breaks. For the second, PasteFix cleans up text you paste from PDFs, emails and web pages.

Broken characters are one subtitle disease; broken timings are the other. Once the text reads correctly, the Subtitle Editor repairs the cues themselves — shifting, splitting and re-timing SRT and VTT files entirely in your browser.

Frequently asked questions

Mojibake is garbled text caused by reading bytes with the wrong character encoding — the classic example is "Café" instead of "Café". It happens when UTF-8 text is opened as Latin-1 or Windows-1252, which is common with CSV exports, old databases, and email gone wrong.
The tool tries dozens of encoding pairs: it re-encodes your garbled text as the suspected wrong encoding, then decodes those bytes as the suspected correct one. Each candidate is scored for readability (real letters and punctuation score well, control characters badly), and the best-scoring repair wins.
Use the "try all" list — it shows every candidate repair with a preview, ranked by score, so you can eyeball the right one. There are also one-click presets for the most common cases, like UTF-8 read as Latin-1 or double-encoded UTF-8.
Often, yes. Double-encoded text (UTF-8 misread as Latin-1, then saved as UTF-8 again) looks like "Café". The auto-detect includes two-pass repairs, and there's a dedicated double-encoding preset button for exactly this case.
No. All encoding repair happens in your browser with the built-in TextDecoder API — your text never leaves your device, and the tool works fully offline.
Because é in UTF-8 is the two bytes C3 A9, and Latin-1 reads each byte as its own character: C3 becomes à and A9 becomes ©. So “Café” turns into “Café” every single time — it is the fingerprint of UTF-8 text opened as Latin-1 or Windows-1252. If your garbled text is full of à sequences, the “UTF-8 read as Latin-1” preset will fix it in one click.
When Excel opens a CSV by double-clicking, it assumes your local legacy encoding (usually Windows-1252) instead of UTF-8, so accented characters split into mojibake. Import instead via Data > From Text/CSV and choose UTF-8, or paste the garbled text into the fixer above to recover it.
Often, yes. If text was mis-decoded and then saved wrong a second time, run it through the fixer, and if the output still shows Ã-style sequences, run it once more. Characters that were replaced by a ? or a replacement character in the first bad save are lost for good, but clean double-encoded text usually recovers fully.

Related Tools

Email Button Generator

Build bulletproof email CTA buttons free — VML fallback for Outlook plus styled anchor, live preview, custom colors…

WhatsApp Chat to PDF — Free Chat Book Maker

Turn a WhatsApp chat export (.txt) into a beautiful printable chat book. Filter by date, align your own messages, then…

Voice Dictation — Free Speech-to-Text Online

Dictate live into text with free browser speech recognition. 8 languages, live results, copy or download notes. Works…

Hashtag Generator — Instagram & TikTok Hashtags

Hashtag Generator — curated hashtag sets for Instagram, TikTok, X, YouTube and LinkedIn. Click chips to copy, shuffle…

Text to Slug

Turn headlines into clean, SEO-friendly URL slugs.

Character Counter

Count letters and characters, with or without spaces.