Which system this is, precisely
The output here is uncontracted English braille, usually called Grade 1. Every print letter becomes one braille letter, in order, with indicator cells added where the rules require them. The letter assignments and the indicators follow Unified English Braille, the code adopted across the US, UK, Canada, Australia, New Zealand and several other countries between 2004 and 2016.
It is not Grade 2. Contracted braille replaces common letter groups and whole words with single cells — and, the, ing, ation and roughly 180 others — under rules that depend on where the group falls in a word, whether it crosses a syllable boundary, and what the word means. Nearly every book, menu and sign produced for adult readers is Grade 2, because it is around a quarter shorter and much faster to read. Nothing on this page implements any of it, and running uncontracted output through a Grade 2 reader will simply read as a slow, verbose, but correct transcription.
The indicators, which is where most of the rules live
Braille has 26 letters and 64 cells, so anything beyond a lowercase letter is signalled by a prefix cell rather than by a new pattern.
| Indicator | Dots | Cell | What it does |
|---|---|---|---|
| Capital | 6 | ⠠ | Capitalises the next letter |
| Numeric | 3-4-5-6 | ⠼ | Turns a-j into 1-9 and 0 until something ends numeric mode |
| Grade 1 | 5-6 | ⠰ | Forces the next cell back to a letter |
| Numeric decimal | 4-6 | ⠨ | Decimal point inside a number |
| Numeric comma | 3 | ⠤ | Thousands separator inside a number |
The numeric indicator is the one that produces surprises. Digits are not separate patterns — 1 through 9 and 0 reuse the cells for a through i and j. After the numeric indicator, ⠁ means 1 rather than a, and it keeps meaning 1 until a space or a non-numeric character ends the mode. That creates a genuine ambiguity when a letter follows a digit directly, as in 3a or 5th, and UEB resolves it with the grade 1 indicator: the a-j cell after a number gets a 5-6 prefix so it reads as a letter again. This tool emits it. Letters from k onwards need no indicator because they were never digits.
The capital indicator here is emitted once per capital letter. UEB also has a capital word indicator — the dot 6 cell doubled — for a run of capitals, and a passage indicator tripled for longer stretches. A word in all caps therefore comes out longer here than a transcriber would write it, though it is not wrong: reading it back gives the same letters.
What is on screen is not braille
The characters in the box above are Unicode braille patterns, U+2800 through U+28FF. They are a visual notation for the dot arrangement and nothing more. Three things follow from that, and all three matter.
First, an actual braille reader cannot read them. Braille is tactile. Producing something readable means an embosser and paper of the right weight, or a refreshable display, and getting the dot height, dot spacing and cell spacing within the tolerances that make the difference between comfortable and painful to read over a page.
Second, a screen reader will probably not announce these as braille. Speech synthesisers generally read U+2800 characters as nothing at all, as a string of dot numbers, or as unknown characters, and behaviour differs between NVDA, JAWS and VoiceOver. A blind visitor using speech gets nothing useful from the output box. If a braille display is connected, what reaches it depends on the screen reader's own translation table and its contraction setting, which may well re-translate the text rather than pass these patterns through.
Third, the fonts. Most current operating systems render braille patterns, but on older systems they appear as boxes or question marks. That is the reason the dot-number view exists — dot numbers survive any font.
When you need a transcriber instead
Use this for learning the alphabet, for checking a short label, for understanding what an indicator is doing, or for reading dot numbers off something you have found. Do not use it for anything a person will rely on. Signage has statutory dot geometry and, in the US, ADA signage requires Grade 2 with specific character heights and mounting positions. Menus, packaging, exam papers and books all go to a qualified transcriber, who will apply contraction rules, layout conventions for headings, tables and page numbers, and a proofreading pass that no automatic converter substitutes for. That is a trade with a certification behind it, and the reason is that the failure mode of bad braille is a reader who cannot tell whether the error is theirs.
Questions people ask
Why is my number coming out with an extra cell in front of it?
That is the numeric indicator, dots 3-4-5-6, and it is required. Braille has no separate digit patterns — the digits reuse the letter cells for a through j — so without that prefix, the cell you would read as 1 is just the letter a. The indicator stays in force through consecutive digits and through a decimal point or a thousands comma, so 1,234.50 takes one indicator, not one per digit. It ends at a space or at anything that is not part of a number.
Can it convert braille back into text?
Not in this direction, deliberately. Going forwards is deterministic: each character has one representation. Coming back is not, because the same cell means different things depending on what came before it — the cell for a is also the digit 1 after a numeric indicator, and in contracted braille a standalone letter can be a whole word. A reverse converter has to model that state and guess at contraction, and a guess that looks plausible but is wrong is worse than no answer.
The output looks much longer than braille I have seen.
It will be, by roughly a quarter to a third, because this is uncontracted. Real braille books, signs and menus are almost always Grade 2, where common letter groups and whole words collapse into single cells — the word "and" is one cell, "the" is one cell, "ation" is one cell. Uncontracted output is correct and readable, just verbose. It is what beginners learn first and what is sometimes used for short labels, technical strings and passwords where contraction would be ambiguous.
Is this good enough for a sign on my workshop door?
For your own door, if you want to label it and nobody is depending on it, it will do. For anything covered by accessibility regulation it is not. ADA signage in the US specifies Grade 2 contracted braille, a dot base diameter and height range, cell and line spacing, rounded dots rather than square, and a mounting height — none of which come out of a screen. Signage is a manufacturing job, not a font job, and the shops that do it will translate the text as part of the order.
Which characters does it not handle?
Anything outside the Latin letters, digits, common punctuation and the symbol set listed in the breakdown table is dropped, and you are told which characters were skipped. Accented letters, currency symbols beyond the dollar, mathematical notation and any non-Latin script are all out of scope. Maths in particular is a separate problem: UEB has its own technical notation rules, and Nemeth code is still used for it in parts of the US.