Paste up to 500 characters to mask letter and number shapes in the displayed output. The result preserves separators and word length, so use placeholders rather than real sensitive material.
The masked character pattern will appear here...
The examples exercise words, digits, punctuation, email-like separators, non-Latin letters, and accented text. Each eligible code point becomes one block.
English
Hello World
Translation
█████ █████
Two five-letter words represented by two five-block groups
English
Plan A-19
Translation
████ █-██
Letters and digits masked while the space and hyphen remain
English
Email: a@b.com
Translation
█████: █@█.███
Email-shaped punctuation preserved around masked letters
English
中文 42
Translation
██ ██
CJK letters and Western digits each replaced by blocks
English
Room #7!
Translation
████ #█!
Hash and exclamation marks remain around masked characters
English
Résumé 2026
Translation
██████ ████
Accented letters and year digits represented by blocks
| Input class | Example | Result behavior |
|---|---|---|
| Unicode Letter | A, é, 中 | One █ per matched code point |
| Unicode Number | 7, ٢ | One █ per matched code point |
| Whitespace | space, line break | Remains in position |
| Punctuation | @, -, ! | Remains visible |
The matcher tests character categories rather than a small English alphabet. This table separates replaced characters from visible structure.
A reviewer should look beyond the dark appearance and list what the unchanged structure still communicates.
| Visible clue | What it can suggest | Safer mockup choice |
|---|---|---|
| Block count | Word or field length | Use a neutral invented length |
| Separators | Email, date, or identifier shape | Replace the whole sample pattern |
| Repeated groups | A familiar phrase or label | Vary placeholder wording |
| Existing symbols | Context around the hidden text | Remove unnecessary context |
This page produces new text. It does not alter a PDF, image, screenshot, word-processing layer, metadata field, revision record, or underlying source. For a real disclosure workflow, use purpose-built document tools and inspect the exported artifact independently.

Word length and punctuation can be inferred from the output. Spaces, @ signs, periods, hyphens, emoji, and repeated group sizes may reveal a format such as an address, date, identifier, or familiar phrase. A black-looking string is not automatically anonymous.

A font may draw █ with different width or vertical alignment, and a screen reader may announce repeated block characters without conveying a useful message. For accessibility, provide a normal-text explanation outside the effect and keep necessary instructions readable.

The source remains visible in the input panel until you edit it or press Reset. Copy transfers only the current output string. Reset clears the displayed fields, but this description is not a statement about other pages, browser extensions, or places where you paste text.

The transform has one character-class rule and no recovery database, hidden layer, or language interpretation.
Letters from Latin and other scripts are matched through the Unicode Letter category. Each matched code point contributes one █ to the result.
Digits and other Unicode Number characters are replaced too. The output does not distinguish whether a block came from a letter or a number.
The rule keeps spaces and punctuation because they are outside the matched categories. That preserves layout clues instead of concealing the overall pattern.
Many different inputs collapse to the same blocks. The function stores no per-character key in the output, so selecting the reverse direction runs the masker again.
Character classes, reverse mode, and responsible use
Check out these related translation and AI tools to enhance your experience
Use invented or already-public sample text. A block pattern can show shape in a mockup, but it is not a document-cleaning workflow.
The browser form enforces the length cap. Use placeholder wording when the real source should not be placed in a decorative tool.
The browser-local conversion returns before the component's provider fetch. It does not use a language model; neither sign-in nor a translation credit is involved in this specific masking action.
Reverse mode applies the same redaction and does not recover the source. The direction labels change, and the arrow swaps the text panels when output exists, but pressing the button replaces eligible characters with blocks again.
Enter a short placeholder, inspect the preserved outline, and choose a proper document workflow when source removal is the real requirement.