Image to Morse Code Converter
An image to Morse code converter reads the characters inside a picture and turns them into International Morse code — or works in reverse, reading printed dot-and-dash patterns out of an image and returning plain text. Upload a PNG, JPG, JPEG, GIF, BMP, or WEBP file up to 5 MB and the result appears in a few seconds.
Free, no signup, works on mobile and desktop. Nothing is uploaded to a server — the recognition runs inside your browser.
- Use clear, high-contrast images with good lighting
- Keep text straight and avoid rotation or skew
- For Morse code images, select "Morse Code in Image" in settings
- Try different image processing modes if results are poor
Which direction do you need?
This is the setting most people get wrong, and it is the single most common reason a conversion returns nothing useful. Pick the mode that matches what is actually printed in your image, not what you want to end up with.
| What’s in your image | Set “Text Type” to | What you get back |
|---|---|---|
| Ordinary words — a sign, a screenshot, a page of a book | English Text | Those words rendered as Morse, e.g. HELLO → .... . .-.. .-.. --- |
| Printed dots and dashes — a puzzle, a cipher, a telegraph chart | Morse Code in Image | The decoded message in plain English |
| Both, side by side (a Morse reference chart) | English Text | The letter labels; the dot-dash column usually needs a second pass |
How this works
The tool runs optical character recognition on your image to pull out the characters, then maps each character against the International Telecommunication Union’s Morse table (ITU-R M.1677-1) — the same specification used in licensed amateur radio and aviation beacon identifiers, not a simplified chart.
In decode mode the order reverses: OCR extracts the literal dot, dash, space and slash characters, the tool splits that string at letter and word boundaries, then looks each sequence up in the same ITU table. A sequence that has no valid ITU equivalent is marked with # rather than guessed at, so you can see exactly which part failed instead of receiving a confidently wrong answer.
The four image-processing modes, and when to use each
- Auto — leave this on for clean screenshots and typed text. It is right most of the time.
- Enhance Text — for photographs of paper. Sharpens character edges softened by camera focus or JPEG compression.
- Grayscale — for colored text or colored backgrounds. Strips hue so the engine judges contrast alone.
- Black & White — the aggressive option. Forces every pixel to pure black or white. This is the mode to try when a faded scan or a low-contrast photo returns nothing; it will also destroy anti-aliased screen text, so don’t use it as a default.
Why OCR struggles with Morse code specifically
This is worth understanding, because it explains almost every failed decode and it is not a limitation you will find documented on most converter pages.
OCR engines are trained overwhelmingly on running prose. Their language models expect letters forming words, and they use that expectation to resolve ambiguous glyphs. Morse code gives them none of that. A line like -.-. --.- is, to the engine, a meaningless run of punctuation, so every character has to be resolved on shape alone — with no word-level context to fall back on. Four specific failures follow from this:
- Dots read as commas. At small sizes a period and a comma differ by a few pixels of tail. Prose OCR resolves this from context; Morse has no context.
- Dashes read as underscores, en dashes, or hyphens interchangeably. The tool accepts
-,_and–as dashes for exactly this reason. - Repeated characters get collapsed. Some engines normalize a run of identical punctuation, turning
...into..or a single ellipsis glyph. This silently changes S into I. - Spacing gets misjudged. Morse encodes meaning in gap width — one space between letters, a slash or wide gap between words. OCR normalizes whitespace by default, which is precisely the information you cannot afford to lose.
The practical consequence: decoding Morse from an image is inherently less reliable than converting ordinary text to Morse, and it always will be. Check the output against the length of the original message before trusting it.
What decodes reliably, and what doesn’t
Rather than quote a single accuracy percentage — which would be meaningless across image types this different — here is what to expect, ranked from most to least reliable. This ordering follows directly from how optical character recognition works, explained in the section above.
| Image type | Expect | Why |
|---|---|---|
| Screenshot of typed text | Near-perfect | Pixel-sharp glyphs, uniform background, no lens distortion — the ideal case for OCR |
| Photo of a printed page, good light | Reliable | Real characters, but focus softness and JPEG artifacts cost some precision |
Printed Morse as . and - characters |
Usable, verify it | No word context to resolve ambiguous punctuation; check the decoded length against the original |
| Low-contrast or faded scan | Try Black & White mode | Often recoverable, but the default pipeline will under-read it |
| Handwritten Morse | Frequently fails | Inconsistent stroke length and spacing — the two things Morse encodes meaning in |
| Morse drawn as bars, blocks, or shapes | Will not work | There are no text characters present for OCR to recognize at all |
If your image sits in the bottom half of that table, the fixes in the next section will move it up more than any tool setting will.
How to photograph Morse code so it actually decodes
Image quality determines the result far more than the tool does. Five things move the number:
- Crop tight. Remove everything that is not the text. Surrounding page edges, hands, and background objects all give the engine extra shapes to misread.
- Get the line height up. Aim for a text line at least 30 px tall in the final image; below roughly 20 px, dot-versus-comma confusion becomes routine.
- Straighten it. More than about five degrees of rotation measurably degrades recognition. Most phone galleries have a straighten tool.
- Maximize contrast, not brightness. Dark text on a light background beats a brightly lit photo of gray text. Avoid on-camera flash on glossy paper — the hotspot erases characters entirely.
- Prefer PNG over JPG for screenshots. JPEG compression puts artifacts around small punctuation, which is exactly the detail Morse depends on.
Characters the engine commonly mangles
If your decode is close but wrong, check these substitutions first — a single one of them changes the entire message.
| Intended | Often read as | Effect on the decode |
|---|---|---|
. (dot) |
, |
Character is dropped from the sequence entirely |
- (dash) |
_ or – |
Handled — both are accepted as dashes |
... |
… (ellipsis glyph) |
S becomes unreadable |
/ (word gap) |
| or 1 |
Words run together |
| Wide letter gap | Single space | Two letters merge into one invalid sequence |
What this tool cannot do
Stated plainly, so you don’t waste an upload:
- Handwritten Morse is unreliable. Neatly ruled, well-spaced handwriting sometimes works. Casual handwriting generally does not, and no OCR-based tool will fix that.
- Morse drawn as graphics won’t read. If the dots and dashes are filled shapes, bars, or blocks rather than actual
.and-text characters, there is nothing for OCR to recognize. - Video and flashing lights are a different problem. For a blinking-light signal, use the Morse code light translator; for audio tones, use audio to Morse code.
- Non-Latin scripts are not supported. ITU Morse covers A–Z, 0–9 and standard punctuation only.
Privacy
Recognition runs client-side in your browser. Your image is not transmitted to us, not stored, and not shared. Closing the tab discards it.
Frequently Asked Questions
Upload the image, leave Text Type set to “English Text,” and press Read Text from Image. The tool runs OCR to extract the words, then maps each character to its ITU Morse equivalent. Output appears as dots and dashes and can be copied, downloaded, or played as sound.
Yes. Switch Text Type to “Morse Code in Image” before uploading. The tool then reads the literal dot and dash characters instead of treating them as text, splits them at letter and word boundaries, and returns the decoded English message.
Accuracy depends almost entirely on image quality, not on the tool. Clean screenshots of typed text decode close to perfectly. Photographs decode well with good contrast and tight cropping. Handwritten Morse is the weakest case and often fails outright.
The three usual causes are the wrong Text Type setting, text that is too small in the frame, and low contrast. Try cropping tighter around the text, switching image processing to Black & White, and confirming the direction setting matches what is printed.
It works, but less reliably than on ordinary words. OCR engines use surrounding words to resolve ambiguous characters, and Morse offers none, so dots and commas and dashes and underscores are easily confused. Always sanity-check the decoded length against the original.
PNG, JPG, JPEG, GIF, BMP, and WEBP, up to 5 MB. PNG is the better choice for screenshots because JPEG compression adds artifacts around small punctuation, which is the exact detail Morse decoding depends on.
Sometimes. Neat, evenly spaced handwriting with clear separation between dots and dashes can decode successfully. Casual handwriting usually will not, and this is a limit of optical character recognition itself rather than a fixable flaw in the tool.
No. Recognition happens in your browser. The image is never sent to our servers, never stored, and never shared. It is discarded as soon as you close the tab.
Yes, completely free with no account, no upload limit, and no watermark. There is no paid tier.
A standard Morse code translator converts text you type. This one adds an OCR step so you can start from a photo, screenshot, or scan instead, which matters when the source is printed material you cannot easily retype.
Explore More Morse Code Tools
About this tool
MorseCode.live is built and maintained by a small independent team. We built this converter and the surrounding reference material ourselves — no white-label toolkit. Character mappings are checked against ITU-R M.1677-1.
Found an image that decodes incorrectly? Send it to us — failed cases are how we find the gaps worth fixing, and they feed directly into the next release.
Version history
- v2.0 — August 2026. Rebuilt the upload and results interface: clearer separation between the two conversion directions, a visible processing indicator, and results that stay on screen while you adjust settings. Fixed several decoding bugs, including mishandling of word-boundary slashes and inconsistent behavior when switching processing modes without re-uploading. Documentation on this page rewritten to match what the tool actually supports.
- v1.0 — April 2026. Initial public release of the image converter, with OCR extraction, both conversion directions, and sound, light, and vibration playback of the result.
