Punctuation normaliser
Remove Em Dashes & Smart Quotes
Em dashes, curly quotes, and ellipsis characters look good in prose and break code, JSON, and style guides. Normalise them to plain keyboard punctuation with one optional toggle — without touching hyphens, date ranges, or anything inside a code block.
Independent of Markdown stripping — use either or both
Cleaning options
What does each toggle do?
- Strip Markdown
- Removes *, #, _, backticks, link syntax, and list markers while keeping the words.
- Remove AI chatter
- Drops standalone openers and outros such as “Certainly! Here’s…” and “I hope this helps!”
- Remove <think> reasoning blocks
- Removes DeepSeek-style <think>…</think> reasoning, thinking tags, and unclosed reasoning blocks so only the answer remains.
- Remove emoji
- Removes pictographic emoji, flags, keycaps, and skin-tone modifiers. Copyright and trademark symbols are kept, and code blocks are protected.
- Preserve tables
- Keeps Markdown tables as aligned rows and real HTML tables in rich-text copy. Turn this off to flatten each row into “heading: value” text.
- Clean invisible Unicode
- Removes zero-width spaces, word joiners, stray control characters, and excess blank lines.
- Bullets from dashes
- Turns “- item” list markers into a typographic bullet (• item) in the plain-text output.
Uploads are read locally. .txt, .md, .csv, .json, .html and .xml up to 5 MB.
Copy for your app
Cleaned text appears here, live.
Cleaning report
Paste text above to see the cleaning report.
Review changes
Change review
removedadded
Paste text above to review what will change.
Download for your app
CtrlEntercopies rich textShiftfor plain text
100% client-side privacy:your text never touches a server. Uploaded files are read in the browser, all cleaning happens locally, and nothing is stored after you close the tab.
- No account required
- No text upload
- No text storage
- Works after page load, even offline
- No AI rewriting or text generation
Advanced options
The characters
Em dash, en dash, and the smart quote problem
An em dash (—) is a long horizontal rule used for an aside. An en dash (–) is shorter and usually marks a range. A hyphen (-) is the one on your keyboard. Three different characters, three different jobs, and only one of them is ASCII.
Curly quotes are the same story. Word processors replace straight quotes with “smart” ones automatically, and chat interfaces often emit them because the model was trained on typeset text. They look correct on a page and fail instantly in JSON, SQL, YAML, and most programming languages.
None of these characters are wrong. They are simply the wrong choice for certain destinations, which is why normalising them should be a deliberate toggle rather than a default.
The mapping
What each character becomes
The normaliser only rewrites punctuation characters. Hyphenated words and hyphen-written ranges are untouched.
| Before | After | Notes |
|---|---|---|
| em dash — | spaced hyphen - | U+2014 becomes “ - ” |
| en dash – | hyphen - | U+2013, and numeric ranges become “2024-2026” |
| curly “double” quotes | straight "double" quotes | U+201C / U+201D |
| curly ‘single’ quotes | straight 'single' quotes | U+2018 / U+2019 |
| ellipsis … | three dots ... | U+2026 |
| non-breaking hyphen ‑ | hyphen - | U+2011 |
Why bother
Four honest reasons to normalise
Code has to run
Curly quotes in a string literal, an em dash in a SQL comment, or an ellipsis in a config value are all syntax errors waiting to happen.
Brand rules are brand rules
Some style guides forbid em dashes outright, or require spaced hyphens. Normalising is faster than finding them by hand.
Translation and tooling
Legacy CMSs, translation memories, and diff tools handle ASCII punctuation more predictably than typographic punctuation.
Consistency across a team
If one writer types hyphens and another pastes em dashes, the document looks uneven. One pass fixes the whole block.
Before / after
The same sentence, normalised
Original
The launch — which slipped twice — is now set for “Friday”.Normalised
The launch - which slipped twice - is now set for "Friday".Hyphenated words like “client-side” and ranges like 2024-2026 stay exactly as they were.
The honest bit
Does this make text “less AI”?
Not on its own. Em dash frequency is a stylistic habit, not a watermark. Removing em dashes changes how a sentence looks; it does not change who wrote it, and no responsible tool should promise otherwise.
What normalisation does guarantee is mechanical: the characters are now the ones your destination expects. That is a real, verifiable improvement, and it is the only claim JustCleanText makes.
If you are editing for voice, the useful work is still reading the draft and rewriting the sentences that sound generic. JustCleanText will not do that for you.
Punctuation FAQ
Em dashes and smart quotes, answered
Which characters change, which stay, and how to normalise without stripping Markdown.
How do I remove em dashes from text?
Paste the text, turn on “Normalize punctuation”, and copy the result. Em dashes (—) and en dashes (–) become a plain hyphen with correct spacing, and curly quotes become straight quotes.
Why do people remove em dashes from AI text?
Em dashes are a well-known stylistic habit of language models, so some publishers and editors treat heavy em-dash use as a signal of AI authorship. Removing them is a style choice, not a factual fix — the writing still has to stand on its own.
What are smart quotes, and why replace them?
Smart quotes are typographic curly quotes (“ ” ‘ ’) that word processors insert automatically. They look better in prose but break code, JSON, SQL, and command-line examples, where a straight quote is required.
Will date ranges and hyphenated words be affected?
No. The normaliser only rewrites dash characters, not hyphens. Hyphenated words such as “client-side” and ranges written with hyphens such as “2024-2026” are left exactly as they are.
Can I normalize punctuation without stripping Markdown?
Yes. Every option is independent. Turn off Markdown stripping and keep only punctuation normalization if you want to preserve the original syntax.
Related cleaners
Other cleanup passes
Markdown Remover
Turn **bold**, # headings, and `code` back into clean sentences without losing links or code.
AI Text Cleaner
Strip the “Certainly! Here’s…” wrapper and the Markdown leftovers from any model output.
Hidden Character Remover
Delete zero-width spaces, word joiners, and directional marks that break search and code.
Start cleaning
Plain punctuation, on demand.
One toggle, applied outside code blocks, with a count of every character changed.
100% client-side · No account · Works offline once loaded