Turn a headline, product name, or file title into a clean URL slug: lowercase, hyphen-separated, accents stripped, punctuation gone. Accented characters are decomposed and flattened rather than deleted, so 'Crème Brûlée' becomes 'creme-brulee' instead of 'crme-brle', and characters that carry meaning are spelled out — an ampersand becomes 'and' rather than silently vanishing. Kebab, snake, dot, camel and Pascal case are all available, along with an optional length limit that trims on a word boundary rather than mid-word. Paste a whole list and it converts one slug per line, which is the fast way to migrate a set of titles. Everything runs in your browser, so unpublished titles stay private.
| Input | Output | Why |
|---|---|---|
| Héllo, World! 2026 | hello-world-2026 | Accent flattened, punctuation dropped |
| Rock & Roll | rock-and-roll | Ampersand spelled out, not deleted |
| Crème Brûlée | creme-brulee | NFD decomposition strips the accents |
| 50% off | 50-percent-off | Percent sign becomes a word |
| Straße | strasse | Sharp s expands to ss |
| 日本語 | (empty) | No ASCII equivalent to fall back to |
A readable slug tells both a person and a crawler what a page is about before they open it, and it is what gets shown in search results and shared in links. Short, descriptive, hyphen-separated slugs are the long-standing recommendation; hyphens are read as word separators while underscores historically were not.
They are decomposed into a base letter plus a combining accent, and the accent is dropped — é becomes e, ñ becomes n, ż becomes z. That keeps the word readable instead of losing the character entirely, which is what happens when a tool simply strips anything non-ASCII.
Scripts such as Chinese, Japanese, Arabic and Devanagari have no ASCII equivalent to fall back to, so they produce an empty slug and you will need to supply a Latin title yourself. Modern browsers and search engines do handle non-ASCII URLs, but they are awkward to share and often percent-encoded into unreadability.
Hyphens for URLs — search engines treat them as word separators. Underscores are the convention for code identifiers, which is what snake_case mode is for. Dot separators are occasionally used for file naming.
Long enough to be descriptive, short enough to read at a glance — commonly around 50 to 60 characters. Set a length limit and the slug is cut at the last word boundary before it, never mid-word.
No. This is local string processing. Titles you paste, including unpublished ones, are never sent to a server, logged, or stored.
Text is lowercased, then a small substitution table replaces characters that should read as words rather than disappear — an ampersand becomes 'and', a percent sign becomes 'percent', the German sharp s becomes 'ss' — because dropping them silently turns 'Rock & Roll' into 'rock-roll'. The result is normalised to Unicode NFD, which decomposes an accented letter into its base character plus a separate combining mark, and the combining-marks range U+0300 to U+036F is then stripped: that single step handles most of Latin-script Europe without a per-language transliteration table, turning 'Crème Brûlée' into 'creme-brulee'. Everything outside a-z, 0-9 and whitespace becomes a word break, runs of whitespace collapse, and the remaining words are joined with the chosen separator or camel-cased. A length limit cuts on the last separator before the limit rather than mid-word, since a slug ending in half a word looks broken. Bulk mode runs the same pipeline per line. All of it is local string processing.