Morse Code Generator

Prepare the source, not just the output

The generator converts supported characters to International Morse; it does not translate a sentence into another language. Lowercase letters are converted through their uppercase form. Leading and trailing whitespace is removed, and runs of spaces, tabs or line breaks separate words. A line break in text input therefore creates a word boundary, not a paragraph feature in the encoded message. Check whether those boundaries are part of your intended phrase.

The mapping includes A-Z, É, digits 0-9, punctuation and four explicit prosign tokens. É also works when entered as E plus a combining acute accent because the source is normalized to composed Unicode. That does not mean all accented letters are accepted. Ä, Ñ, Ç, Cyrillic letters and emoji have no mapping here, and ß is not automatically converted into two S characters. The warning lists unsupported input; its remaining characters are still encoded.

For example, CAFÉ becomes -.-. .- ..-. ..-.. and decodes back to CAFÉ. By contrast, an unsupported character can disappear while the rest forms a perfectly valid signal. A valid-looking result is not proof that every source character survived. Keep the original and correct the spelling deliberately before copying anything that identifies a person, object or place.

Read character groups and literal punctuation correctly

The written output uses one space between encoded characters and / between source words. HELLO WORLD contains ten character groups and one word separator. If you collapse all those spaces for a graphic, the decoder cannot uniquely reconstruct the letters. If you replace a space between letters with a slash, you create a word break. Formatting is part of the message contract with the reader.

AB9 becomes .- -... ----.. A contains two marks, B four and 9 five; the three characters stay in one word because the source has no spaces. AB 9 instead contains a word gap before the digit. Use this contrast when designing a clue around an identifier: the tool preserves your divisions but cannot know which grouping your organization or puzzle expects.

A/B becomes .- -..-. -.... The middle group is the transmitted slash character. A B becomes .- / -..., whose slash is only the written word separator. The same care applies to email addresses: @ is .--.-., while a full stop inside an address is .-.-.-. There is no automatic validation of an address, date or mathematical expression.

Procedure signals are joined, ordinary letters are not

A prosign combines marks into a procedural signal without the normal inter-letter gap. Enter <AR>, <AS>, <SK> or <SOS> explicitly when you intend a joined signal. The angle brackets are input notation, not characters sent as part of the signal. Lowercase token names are accepted through uppercasing, but an unknown token such as <XYZ> is reported, not interpreted as its individual letters.

Compare AR, A R and <AR>. AR gives two letters in one word; A R gives two one-letter words; <AR> gives .-.-. as one joined group. In ITU-R M.1677-1, that pattern is also the cross/addition sign and is used to terminate a telegram. The decoder's prosign preference selects <AR> rather than + when context requires it. No setting can establish the sender's intent from the marks alone.

SOS and <SOS> offer another instructive comparison. Both have nine marks, but SOS inserts character gaps while the joined token keeps one-unit inner gaps throughout. At 20/20 WPM, ordinary SOS takes 1.62 seconds and <SOS> takes 1.38 seconds. Use this to explain a joined signal, not to simulate a certified distress transmission. The generator does not address a station or add a call sign, position, acknowledgement or operational procedure for you.

Choose playback settings for the recipient

Character speed sets the dot duration as 1.2 / WPM seconds. At 20 WPM, a dot lasts 0.06 seconds and a dash 0.18 seconds. At equal Farnsworth speed, letter gaps are 0.18 seconds and word gaps 0.42 seconds. Reducing Farnsworth to 10 WPM keeps those dots and dashes intact but increases the letter gap to approximately 0.654 seconds and the word gap to approximately 1.525 seconds. ARRL’s Farnsworth timing calculation explains the PARIS calibration behind these pauses.

For a beginner's name badge demonstration, 18/10 WPM lets you hear a reasonably compact character while allowing more time to identify it. For comparing two generated samples, keep both speed settings, pitch and volume unchanged and alter only the phrase. Otherwise you may be comparing rhythm settings rather than characters. Pitch from 300 to 1000 Hz changes tone colour, not spelling; the volume control does not override a muted device.

Repeat loops the same complete message. It is helpful for familiarization but can make a learner anticipate the sequence instead of recognizing each character. Turn it off for a single attempt, then use the chart or a different phrase to check the uncertain letter. Editing text or changing a setting stops playback; returning from a hidden tab also requires a deliberate new start. None of the generated examples starts automatically.

Pick the export that carries the information you need

Copy and text download carry the converted output. A text file is the easiest format for a worksheet author to inspect and revise, but keep a separate source phrase if you need the original spelling and punctuation. The image export creates a static PNG of written code. It does not contain speed, repeat, vibration or an embedded audio track. Check its readability at the final printed size, especially at line breaks.

WAV export carries one audio signal at your selected character speed, Farnsworth spacing, pitch and volume. It is mono, 44.1 kHz, 16-bit PCM. Repeat cycles are not included. Playback and WAV generation are limited to 90 seconds per signal, so a string of long digit patterns at a low Farnsworth speed can exceed the limit sooner than expected. Split a long lesson into short labelled samples rather than removing word gaps to squeeze it into a file.

A share link carries the input and chosen settings in the URL. It is useful when another person needs the same setup, but hidden-answer presentation is not secrecy and the URL is not encrypted message storage. Light and vibration are not armed by opening that link. Test the link yourself before distributing a puzzle and use non-sensitive text. For a durable classroom resource, a checked WAV plus an answer key may be clearer than a link whose recipient can change the settings.

Verify a message and understand what the check proves

A useful proofreading sequence is source review, warning review, decoding and listening. First confirm spelling and intended spaces. Next check unsupported characters. Then copy the written code into the decoder and compare the returned uppercase text with the supported source. Finally listen at the settings you will actually use. The round trip checks the map and separators; listening checks whether the chosen pace suits the activity. Neither establishes that a distant receiver will hear you.

Expect a few intentional differences on decoding. Letter case is not preserved. Multiplication × uses the X signal under ITU-R M.1677-1, so 2×3 comes back as 2X3. The signal cannot distinguish those source characters. Procedure patterns may come back as punctuation unless prosign preference is enabled. Record that choice if a transcript must retain procedural meaning rather than merely the equivalent marks.

Not every operational rule in the recommendation is automated here. Percentages, mixed fractions and minute/second signs have specific transmission conventions in the document; the generator does not expand % or infer those conventions from a number. For a puzzle, an agreed plain-language spelling may suffice. For studying a historical or operational telegram, consult the relevant rule and preserve the source notation instead of presenting the generated output as a full procedure formatter.

A generated phrase is not a historical recording

The 1844 message preserved in the Library of Congress's 1844 message tape connects a written phrase with a physical record of a transmission. It helps explain why a generator is useful: someone had to turn a known message into marks, and someone else had to interpret them. Re-entering that phrase here produces today's supported International Morse representation, not the original paper tape, apparatus sound or precise historical timing.

International compatibility developed through more than an alphabet. ITU’s account of the 1865 conference describes the 1865 agreement on service regulations, tariffs and Morse equipment. That history has a practical analogue in a modern handout: identify the chart being used, distinguish extension punctuation and preserve boundaries. A visually attractive clue is only usable when the other person can apply the same reading rules. For creative work, that shared convention is more valuable than adding dramatic historical claims to an unchecked message.

The older maritime examples are also useful when creating a known-text exercise. The 1909 Congressional Record on Jack Binns records Binns's CQD calls in 1909, and the Science Museum's account of Titanic wireless discusses Titanic's CQD and SOS traffic in 1912. Generate CQD as three ordinary letters, then compare SOS with the joined <SOS> token. Ask what an actual receiving operator still needed after recognizing a call: the sending station, position and message. Your generated sample can illustrate that distinction, but it is not an original recording or a complete reconstruction of the historical exchange.

Frequently asked questions