Strip accents and transliterate text to plain ASCII

About this tool

accent-stripper converts Unicode text into ASCII-friendly text. It is for the everyday cleanup jobs where accents and non-ASCII letters break matching, filenames, URLs, search keys or legacy systems: café becomes cafe, Straße becomes Strasse, Ærøskøbing becomes AEroskobing, and Москва becomes Moskva.

The default Transliterate to ASCII mode uses a transliteration table, not just Unicode decomposition. That matters for letters such as ß, ø, Æ, Ł, Đ, Greek, Cyrillic and CJK characters, because many of them do not carry a removable combining accent. For a conservative normalization pass, switch to Marks only: it removes combining marks (é → e) but leaves letters like ß or Ж alone unless your unmapped policy removes or replaces them.

Worked example

Input:

Crème Brûlée à la Française — Straße, Ærøskøbing, Москва

Default output:

Creme Brulee a la Francaise -- Strasse, AEroskobing, Moskva

For slug preparation, enable lowercase and collapse whitespace:

  Déjà Vu: Señor Piñata!  

becomes:

deja vu: senor pinata!

You can then pass that through a slugifier or punctuation filter if you need dashes instead of spaces.

Options and limits

• Conversion mode controls the core behavior. transliterate aims for readable ASCII equivalents; marks-only removes combining marks and is deliberately narrower.
• Characters still non-ASCII chooses what happens after conversion: keep them, remove them, or replace each one with the replacement text.
• Replacement text must be ASCII and at most 8 characters. It only applies when the unmapped policy is replace.
• Keep these characters protects literal non-ASCII characters from conversion. Use this when a language-specific letter such as ñ must remain distinct.
• Lowercase result runs after conversion, so transliterated uppercase letters become lowercase too.
• Collapse whitespace trims each line and squeezes spaces/tabs to one plain space while preserving line breaks.
• Return JSON audit report returns the converted text plus counts of input characters, output characters, converted/kept/unmapped characters, and whether the final output is pure ASCII.
• The input limit is 200,000 characters per run. Split larger documents.
• This tool does not translate words between languages. It approximates characters, so names and addresses remain recognizable but not linguistically perfect.

FAQ