Braille Translator
Paste a short piece of English, or braille you already have in Unicode or ASCII, choose Grade 1 or Grade 2, and get the other side back — with the dot patterns, Unicode and ASCII views side by side. Copy the result, or download TXT, BRF (for embossers), or printable PNG. Made for learners and the people helping them check a passage line by line. Up to 5,000 characters per run.
Convert
Ready. The engine loads locally in your browser on first Convert.
Start with one of five short scenarios, or paste your own text.
For braille → English, choose Unicode or ASCII. ASCII input uses lowercase NABCC (text_nabcc.dis); BRF downloads use uppercase ASCII for embossers, so convert BRF to lowercase or paste Unicode if re-importing. Maximum: 5,000 Unicode code points.
Unicode braille
—
Dot patterns / numbers 1–6
Unicode cells show the pattern; numbers name raised dots 1–6. 0 = blank cell. For a slate, mirror each cell and write right to left; turn the paper over to read the raised dots.
—
ASCII braille
—
English result
—
Engine: liblouis 3.39.0 · en-ueb-g1.ctb · text_nabcc.dis. LGPL 2.1 or later · Source, build recipe & credits. Unmodified tables. Grade 2 depends on context. Round-trip match ≠ correctness. Learning and self-check only.
Export result
UEB Grade 1 alphabet quick reference
These twenty-six cells are Grade 1 mappings from this site’s liblouis 3.39.0 tables—the same engine the converter uses. Dot numbers name raised dots 1–6. Use this chart to check a single letter; paste a whole passage above when you need Grade 2 contractions or reverse translation.
| Letter | Unicode cell | Dots |
|---|---|---|
| A | ⠁ | 1 |
| B | ⠃ | 12 |
| C | ⠉ | 14 |
| D | ⠙ | 145 |
| E | ⠑ | 15 |
| F | ⠋ | 124 |
| G | ⠛ | 1245 |
| H | ⠓ | 125 |
| I | ⠊ | 24 |
| J | ⠚ | 245 |
| K | ⠅ | 13 |
| L | ⠇ | 123 |
| M | ⠍ | 134 |
| N | ⠝ | 1345 |
| O | ⠕ | 135 |
| P | ⠏ | 1234 |
| Q | ⠟ | 12345 |
| R | ⠗ | 1235 |
| S | ⠎ | 234 |
| T | ⠞ | 2345 |
| U | ⠥ | 136 |
| V | ⠧ | 1236 |
| W | ⠺ | 2456 |
| X | ⠭ | 1346 |
| Y | ⠽ | 13456 |
| Z | ⠵ | 1356 |
UEB Grade 1 and Grade 2
Which braille do I need?
Current standard: UEB in the US since 2016; the UK and Australia have also unified on UEB.
- Reading pre-2016 materials → may be EBAE / older British code
- Printing or writing now → UEB
- Unsure → use UEB, and mark contractions in the result one by one
The same short sentence below was converted with this page’s liblouis 3.39.0 tables—not hand-copied. Grade 2 shortens words that Grade 1 spells cell by cell:English: “You can go now.”
Letters and basic symbols without contractions.
⠠⠽⠕⠥⠀⠉⠁⠝⠀⠛⠕⠀⠝⠕⠺⠲
ASCII (NABCC),you can go now4
Includes contractions; context matters, so a round-trip match alone does not prove the contraction choice was correct.
⠠⠽⠀⠉⠀⠛⠀⠝⠪⠲
ASCII (NABCC),y c g n{4
In Grade 2, you, can, and go become single-letter contractions, and ow in now uses a groupsign—those shortenings do not appear in the Grade 1 line.
Braille cell basics
A standard literary cell is six dots in two columns of three, numbered 1–2–3 down the left and 4–5–6 down the right. Raised dots encode letters, punctuation, and—in Grade 2—contractions. A blank cell is a space between words.
When you are learning the alphabet, use Grade 1 and the dot patterns view so each print letter maps to a fixed cell. Grade 2 reuses some of those cells as contractions, which is why the same English letters can look shorter after Convert. This page is a translator for short passages; it is not a full braille-alphabet course.
How to use the three result views
After Convert, the same UEB cells appear in three forms. Choose by what you need to do next—none of them is “more correct” than the others.
- For sending (Unicode ⠁) — copy Unicode braille into a message, document, or compatible display that understands the Braille Patterns block (for example ⠁).
- For checking (Dots 1–6) — use Dot patterns / numbers 1–6 to compare one cell at a time. Numbers name raised dots;
0is a blank cell. - For tactile reading (ASCII / BRF / PNG) — you need an embosser, a slate and stylus, or a refreshable braille display; ordinary printing is not enough. Copy lowercase NABCC ASCII for compatible software, download a 34 × 25 BRF file for braille embossers, or export a printable PNG dot sheet (with an optional mirrored layout for slate and stylus).
How to use
- Choose direction and UEB Grade 1 or UEB Grade 2. For braille → English, also choose Unicode or ASCII.
- Paste into the active input. Your other direction’s draft is kept, but only the active input is counted and converted.
- Convert, review the results, then choose a copy view, copy, or download TXT, BRF (for embossers), or printable PNG in the export area below the results.
Built for learners and helpers checking a short passage line by line. Nothing here replaces a certified braille proofreader.
FAQ
Do I need Grade 1 or Grade 2?
Use UEB Grade 1 for uncontracted letter-by-letter practice. Use UEB Grade 2 when you are checking contracted UEB. This site only offers those two UEB options (no separate EBAE / older British locale list). Pick explicitly — the site will not silently guess.
Which engine and tables does the converter use?
A locally built WebAssembly version of liblouis 3.39.0. Grade 1 selects the unmodified en-ueb-g1.ctb table; Grade 2 selects en-ueb-g2.ctb. The selected literary table and its includes run in both directions—there is no separate back-translation table. The active table appears near the results, alongside the LGPL license, corresponding source, and table credits.
What is the difference between Unicode and ASCII braille?
Unicode braille uses the Braille Patterns block (e.g. ⠁). ASCII braille encodes cells with printable ASCII characters using the six-dot subset of NABCC (text_nabcc.dis). Use lowercase ASCII letters: uppercase NABCC uses dot 7 and is rejected. Digits and punctuation encode cells, not their print meanings. This is not uppercase BRF encoding. In source data, a cell such as ⠀ may be written as the escape \u2820; paste the actual Unicode braille character here, not the six-character escape.
What if my text is over 5,000 characters?
Split or shorten it to 5,000 Unicode code points or fewer, then Convert again. Combining marks count separately. Your input is kept and never silently truncated; previous results clear until a new successful conversion.
Does my passage leave this browser?
Conversion runs in your browser. The engine and tables are downloaded from this site on first use, but the converter does not send your passage to a translation server. Copy and TXT export use browser features; if engine loading fails, your input remains so you can retry Convert.
Why can't I upload an image or PDF?
This version converts typed or pasted literary English and UEB braille only. It has no OCR, image, photo, or PDF input, so it cannot recognize text or braille from a scan.
Is a perfect round-trip proof that my braille is correct?
No. A match only shows that the forward and reverse conversions agree. Braille may omit accent or capitalization details and reuse cells ambiguously, while Grade 2 contractions depend on context. Plain-text input also supplies no bold, italic, or underline typeform data. Check the cells and the engine limits; do not treat a round trip as certification.
Where do I convert braille back to English?
Use the same converter on this page (choose Braille → English), or open the dedicatedbraille to English page for reverse-direction help and examples. For letter lookup and a printable sheet, use the UEB braille alphabet chart. A full Grade 2 lesson track is still out of scope for this first version.
Engine, scope & known limits
Uses unmodified en-ueb-g1.ctb / en-ueb-g2.ctb tables in liblouis 3.39.0. Back-translation is context-dependent and not a unique reconstruction of print.
Agreement between forward and reverse conversions does not certify UEB accuracy. Plain-text input also passes no bold, italic, or underline typeform emphasis data.
- Publication, examination, or certification workflows
- Legal, medical/pharma, or safety-critical wording
- Commercial braille publishing or multi-volume press pagination (BRF export provides a 34 × 25 draft layout only)
- Any use requiring certified 100% accuracy or OCR/image input
Technical audit notes, measured liblouis 3.39.0 behaviors & LGPL source credits
Back-translation is imperfect. Braille can omit which accent a letter carries, mark only some capitals, or use one cell for punctuation, a math sign, or a contraction. A reader fills in those gaps from context; software must make choices too. The liblouis manual’s back-translation notes therefore caution that the process may not be perfect.
The same UEB tables run in both directions. Upstream explains that tables may bend braille rules to improve reversibility, or use separate tables for forward and back-translation. This site adds no rule-bending or separate reverse tables: it uses the same unmodified en-ueb-g1.ctb or en-ueb-g2.ctb literary table, with its includes, for the selected grade in both directions. Reverse output is not a unique reconstruction of the original print.
Round-trip match ≠ correctness. A matching round trip shows that two conversions agree; it does not prove correct UEB, uniquely recovered English, or preserved emphasis. Grade 2 contractions are context-sensitive, and the information gaps above remain even when the text matches.
Measured on this build. Selected local checks—not an accuracy rate or certification—found:
- Grade 2 “His child was in there.” round-tripped exactly through the same table. A
noforrule excludes that rule line from forward translation; it does not by itself prove that a word cannot round-trip through other rules. - “I can’t go.” returned as “I can't go.” in both grades. Curly double quotes likewise returned as straight quotes, so exact typography was not preserved.
- In Grade 2, dots 56-2 decoded to a comma, but translating that comma back produced dots 2. Some accepted cell sequences therefore lose an indicator on a reverse-then-forward round trip.
- The single cell ⠃ decoded to “but” in Grade 2 and “b” in Grade 1. The selected grade changes the reading.
- “café” triggered a non-ASCII warning but round-tripped unchanged in both grades. A warning calls for inspection; it does not necessarily mean text was lost.
Evidence: docs/evidence/nofor-bidirectional-2026-10-30.md (selected examples and a bounded source-operand sweep, not a natural-language corpus).
Plain text carries no emphasis styling. A textarea supplies characters, not bold, italic, or underline spans. Convert therefore passes no typeform emphasis data to liblouis, so those emphasis indicators are not generated. That is information lost at the plain-text UI boundary, not an engine fault.
Undefined input can produce visible escapes. By default, liblouis emits \xhhhh for an undefined print character during forward translation and \ddd/ for an undefined braille pattern during back-translation. Treat these as signals to inspect the input and result, not as readable translation; also review any non-ASCII warning shown near the results.
Three views of the same cells—not an “advanced” mode. Unicode braille, six-dot NABCC ASCII, and dot numbers are standard representations of the engine-produced cells. They are provided for different learning and copying needs; none is a more authoritative translation.
TXT vs. BRF export. The UTF-8/LF TXT file contains English and the three braille views for reference. For braille embossers, use Download BRF to export standard uppercase ASCII with CRLF line endings and optional 34 cells × 25 lines pagination.
The official liblouis online translator describes itself as a testing page rather than a complete transcription tool. This page adds explicit controls, side-by-side views, and visible limits for short learning checks—not a certification of the output.
- Up to 5,000 Unicode code points per run. Combining marks count separately; longer input is refused without truncation.
- Literary English ⇄ UEB only; no OCR, image, or photo input. Only six-dot braille is supported.
- Not for publishing, exams, legal materials, pharmaceuticals, wayfinding/signage, or commercial press pagination.
Rules: ICEB Unified English Braille · engine: liblouis 3.39.0, licensed LGPL 2.1 or later. The source notes link the corresponding source and preserve table credits, including ViewPlus Technologies, American Printing House for the Blind, and RNIB.