Smart quote converter

Turn “smart” punctuation back into the characters on your keyboard: curly quotes to straight, the ellipsis character to three dots, fancy spaces to ordinary ones. Each rule has its own switch, and the reverse button puts the typography back.

The characters autocorrect adds without asking

Type a straight quote in Word, Google Docs or most note apps and it is silently replaced with a curly one — opening or closing depending on where it landed. Type three dots and you get a single ellipsis character. Type two hyphens and you get an en or em dash. None of this is announced, and the document that results contains characters you never pressed a key for.

Chatbot output arrives the same way for a different reason: it was trained on text that had been through exactly those tools, so it writes typographic punctuation by default. PDFs add their own contribution — copy text out of one and the fi and fl pairs often come back as single ligature characters, which is why a pasted word can look right and fail every search for it.

All of it is correct typography and all of it is a problem the moment the text leaves the document it was written in.

What breaks when a curly quote reaches code

A curly apostrophe in a JavaScript or Python string does not close the string, so the file will not parse. In JSON it is worse, because the file often parses and produces a value nobody expected. In a CSV, a quote that is not the plain double quote is not a field delimiter, so the row splits in the wrong place and the import fails several thousand rows later.

Then there is mojibake. If you have ever seen ’ where an apostrophe should be, that is a curly apostrophe encoded as UTF-8 and read back as Windows-1252 — three bytes shown as three characters. Straightening the quotes before the text crosses a system boundary removes the whole class of problem, because the ASCII apostrophe is the same byte in every encoding anyone still uses.

The quieter failure is search. A support agent looking for a phrase they can see on screen finds nothing, because they typed a straight apostrophe and the record holds a curly one. Databases and search indexes treat them as different characters, which they are.

Going the other way, and why it is a guess

Published prose should have curly quotes — they are what typesetting has used for centuries, and straight quotes in a printed article read as an error. The Smarten quotes button converts in that direction, along with ... to an ellipsis and -- to a dash.

It cannot be right every time, and it is worth knowing where it fails. A straight apostrophe is used for three different things: a contraction, a possessive, and a foot mark. A straight double quote is a quotation mark or an inch mark. Nothing in the character says which, so the conversion reads the character in front of it and guesses: after a letter or a digit it produces a closing curl, otherwise an opening one.

That gets ordinary prose right and reliably fails on 'tis, '90s, and 6'2", all of which want a closing curl in a position where an opening one is the sensible guess. It is also why you should never smarten a code sample: every quote in it is syntax, and curling it breaks the code silently.

Each rule, on its own

Every row here was produced with that one rule switched on and the other four off, so the right-hand column is exactly what unticking everything else would give you.

The spaces row looks unchanged, and that is the point: the before column holds a no-break space and a narrow no-break space, and neither of them looks any different from the ordinary space that replaces it. The difference only shows up when something tries to match on it.

Common questions

Other text tools

• Case converterAll tools
• Sentence case converterCase
• Title case converterCase
• Uppercase converterCase
• Lowercase converterCase
• camelCase converterCode
• Snake case converterCode
• Kebab case converterCode
• Remove line breaksLines
• Remove extra spacesLines
• Sort lines alphabeticallyLines
• Remove duplicate linesLines
• Reverse textLines
• Word counterCount
• Clean up AI textAI text
• Em dash removerAI text
• Invisible character removerAI text
• Markdown formatting removerAI text