This translator reads Morse code you can see. Upload a picture of a line of dots and dashes and it measures the marks; upload a video, or point your camera at a blinking light, and it reads the flashes instead. Everything is decoded in your browser.
Choose a picture of dots and dashes, or a video of a blinking light. The decoding happens in your browser and nothing is uploaded.
Visible Morse comes in two shapes, and the tool treats them differently. A picture is measured across, mark by mark. A video or a live camera is measured over time, frame by frame.
Below the tool are the things that make a picture readable, how to film a flashlight signal so it decodes cleanly, and what to change when the result is close but not quite right.
A recording rather than a picture goes to the audio translator, and code you can type goes to the translator.
A picture is read as shapes. Every mark is found as its own blob of ink, the ones sharing a line are gathered together, and the narrow marks become dots while the wide ones become dashes.
Finding marks as shapes rather than as busy columns is what lets a screenshot work. Morse typed in a font puts its dots on the baseline and its dashes at mid height, so the two never share a row, and it is spaced by the font rather than by Morse timing. Both readings are tried and the one that spells real letters wins.
A video is read as brightness over time. The brightest part of every frame is measured about forty times a second, and the light being on or off becomes the same dots, dashes and gaps.
This is not character recognition. It finds marks and measures them, without ever asking which letter a shape might be, which is why a clean line of dots and dashes decodes reliably and a paragraph of handwriting does not decode at all.
A torch flashing Morse across a room is the same message as a tone, just carried by light. Press Use the camera, aim it at the source, and press stop when the message finishes; the whole signal is decoded at once rather than letter by letter.
Two things decide whether it works. The light has to be the brightest thing in the frame, so a darker room helps a great deal, and it has to go properly dark between symbols rather than dimming.
Keep the tab in front while a video or a camera signal is being read. Browsers slow a background tab down, and the timings this decode depends on go with it.
Speed matters less than you would expect. Frames are timed as they arrive rather than assumed, so a signal sent slowly by hand reads as well as an even one, but anything faster than a few flashes a second outruns the camera.
Decoding a signal is a measurement, not a certainty. A cramped gap turns two letters into one, and a long dot becomes a dash.
The dots and dashes are shown above the text for exactly this reason. Paste them into the Morse code translator, fix the symbol that looks wrong, and convert again.
For a single letter, the Morse code alphabet chart is quicker than re-reading the whole line. For a signal you can hear rather than see, the audio translator does the same job with sound.
The image translator decodes one horizontal line of dots and dashes, with the marks clearly darker or lighter than the background. A printed line, a clean photo of handwriting, and a screenshot of Morse typed as text all work. A page of prose does not.
You can try it without a picture of your own. Press Try a sample picture and the tool decodes a built-in line of dots and dashes reading SOS. It uses the same decoder an uploaded file does, so the result is real.
It decodes a screenshot of Morse typed as text. Dots and dashes set in a font are spaced by the font, not by Morse timing, so the tool measures centre to centre as well as edge to edge and keeps whichever spells real letters.
It reads Morse code from a video when the message is a light switching on and off. The clip plays through silently while the brightest part of each frame is measured, so a torch, a lamp or a signal light all read the same way.
It works with a flashlight in real time. Press Use the camera, aim it at the light, then press stop when the message finishes. The light has to be the brightest thing in frame, so a dim room helps.
Your photo or video is never uploaded. The file is read into a canvas inside the page and analysed there, and camera frames never leave the device. That is also why it keeps working with the network disconnected.
A video takes as long as the clip itself to decode, because it is played through once while the brightness is measured. Keep the tab in front while it runs: browsers slow a background tab down, which stretches the timings the decode depends on.
A picture usually fails to decode because the contrast is too low or the marks are not in one row. A border or a caption rule is fine, since those are set aside, but a blurred photo or a shadow across the line is not.
Last updated:
