Everything in the app,
and how to use it

The highest-value thing you can learn. Nothing else comes close.

You need to recognize your call instantly when someone sends it, which is your cue that the contact is yours and it's your turn, and you need to send it cleanly and consistently. Both directions matter, and most people drill only the first.

How to drill it in the app:

• Set it once in Settings → Your callsign, so the app uses it throughout.
• Turn on Use my word list and put your callsign in it (with a few near-misses: calls that share your prefix or suffix). Then run Common Words: you're training to pick your call out of things that sound like it, which is the real skill.
• Rapid Fire → Call signs for volume, so your call stops being a special case and becomes one call among many.
• Sending Practice to key it back until the rhythm is automatic and identical every time.

The single character that rescues any exchange you've lost.

?, sent as ··−−··, means "repeat" or "I didn't get that". Send it and the other operator sends again. It costs one character to learn and it turns a failed contact into a successful one, which is why it belongs this early.

It's already in the base character set, with no setting to turn on, so it will come up in normal practice. Worth deliberately over-learning anyway: you want it available under pressure, when you're flustered and half a callsign short.

Two letters that complete most of the exchanges you'll hear.

An enormous share of US CW traffic ends in a two-letter state abbreviation, and the POTA exchange is essentially a signal report plus a location. Learn yours cold, and learn to copy the common ones.

Rapid Fire → State abbreviations drills exactly this, back to back.

Rounds out the vocabulary of a short contact.

• CQ: calling anyone.
• DE: "from", separating the called station from the caller.
• RST and 5NN / 599: the signal report. Nearly always 599 in practice; note the cut numbers, since 5NN is what you'll actually hear.
• TU: thank you, and the usual sign that the contact is done.
• 73: best regards, the standard sign-off.
• QRZ: "who is calling me?"

Abbreviations and Q-Codes cover these, and the Reference chart lets you tap any of them to hear it.

Multiple choice. The answer choices are drawn from characters you've already met, and they're deliberately sound-alikes: B against D, U against V, S against H, so the quiz mirrors the confusions that actually happen on the air. The number of buttons grows as you learn, up to the Answer choices limit in Settings.

No choices at all. Hear it, type it. Harder and more honest than multiple choice, because there's nothing to eliminate. This is the whole point of Type It, and a hardware keyboard works too if you have one paired: just type the answer and it's submitted.

Turn on Voice answers in Settings and say the answer instead of tapping it. It understands NATO phonetics ("Charlie"), letter names ("see"), and digit words ("niner"). When it isn't sure it asks "did you say…?" rather than marking you wrong, and it learns your corrections, so your accent and habits improve the matching over time. Uses the microphone; nothing leaves the device.

On iPhone and iPad, listening starts by itself the moment the tone ends, and your recognition time runs from when you start speaking. On Android you tap Speak answer for each prompt, because Android's recognizer plays a start chime every time it opens the microphone and would chime over every character if the app listened automatically; the clock runs from the tap.

Answer by sending, not selecting. Key the character on the on-screen touch key or a connected MIDI key and the app decodes what you sent. It's available in every choice quiz whose answer is a character or a word, with a Key answers toggle in the session bar so you can flip between keying and tapping mid-drill; drills whose answer is a meaning or a prosign show the choices instead. This is the fastest way to make sending and receiving reinforce each other. See Sending Practice for the dedicated mode.

The guided path. If you don't know what to practice, practice this.

Journey is a leveled route through the whole of CW, and it decides the curriculum for you. It runs through seven sections in order:

1. Letters & Numbers: two new characters per level
2. Punctuation: one per level
3. Prosigns: two per level
4. Q-Codes: two per level
5. Abbreviations: two per level
6. Words: four per level, from the 60 most common
7. Call Signs: three per level

How a level works

Each level introduces its new items and mixes them with what you already know. A progress bar fills on a hit and drains on a miss, and you clear the level by filling it, so a run of misses genuinely sets you back. (If you'd rather it didn't drain, that's toggleable in Settings.)

A level's new items get at least half of its prompts, so a prosign level actually drills the prosigns rather than quizzing you on letters you learned weeks ago, and a prosign's wrong answers are other prosigns once the pool has them, not three plain letters you could rule out from the labels.

The unlock map shows the whole route: what you've cleared, where you are, and what's ahead. Progress is saved, so you can put it down for a week and pick up where you left off. What you told the app at first run (and can change under Settings → Proficiency) unlocks the Journey that far too: someone who knows the letters starts at the first level with a digit in it, and "everything" starts at the first prosign level. It only ever unlocks forward, so cleared levels are never taken back.

The classic Koch ladder, with you holding the controls instead of Journey.

A-Z and 0-9, added one at a time as your recognition time comes down. Same engine as Journey, but you choose the settings and there's no level structure, just the ladder.

Meeting a new character

Before a character or prosign is drilled for the first time, the track stops and introduces it: the item on its own, its pattern in symbols and spoken out as "dah-di-dah", its sound when the screen appears and again on Replay, and a Start button when you're ready. Then the drill carries on with it in the mix. If you'd rather meet new characters cold, Settings → New characters turns the introduction off.

Track stage

The Characters track moves through stages as you improve: single characters → pairs → triples → words & call signs. Normally it advances on its own, but you can pin it if you want to stay on single characters for a while longer, or jump ahead to word-length copy.

The pin is on the setup screen and, since the stage is the thing you most often want to change once words start coming at you, it's one gear-tap away mid-session too: on iOS the Track stage picker sits in the session's Settings with a "Back to" button for the previous stage, and on Android the stage pills sit on the drill itself, in Characters and in Sending Practice.

Pinning is the right call when you want deliberate practice on one thing. Leave it unpinned when you just want the app to keep you at the edge of your ability.

Reach for this the moment two characters start blurring together.

The app keeps a confusion matrix. Every time you answer D when it sent B, it's recorded. Confusion Drill reads that matrix and builds a session out of your actual mix-ups, not a generic list of tricky pairs.

If you haven't recorded enough confusions yet, it falls back to your slowest active characters paired against their nearest sound-alikes, which is a good drill in its own right.

When to use it: when accuracy is fine but one or two pairs keep costing you. Ten minutes here is worth an hour of general practice for a pair you've been fumbling for weeks.

Arcade recognition: the Characters drill with a clock on it, and a game you can lose.

Aliens descend the play field in five columns, and you shoot each one by naming the character it carries before it reaches the ground. Two ways to play:

• Hear it, type it: the aliens come down unmarked. Each one is sent in Morse as it appears (tap it to hear it again); type the character you heard on the on-screen keyboard, and the lowest alien carrying it is hit. The keyboard is laid out QWERTY, with a digit row above it whenever the set has digits and any punctuation on a row below; keys outside your current set stay in place, dimmed and dead, so the layout never shifts as the set grows and the reflex you build is the one a real keyboard rewards. A hardware keyboard works too.
• See it, key it: each alien shows its character directly above it. Key it on the on-screen key or a hardware key, and the decoded character shoots the lowest one carrying it.

The aliens carry either your active Koch set, so the game drills exactly what you're learning, or the full alphabet and digits.

The speed ramp

In hear-it mode the game starts slow and speeds up only as you consolidate. A game opens 10 WPM below your character speed, never below 15, and steps up 1 WPM after six hits in a row at the current speed, until it reaches your character speed. A wrong shot resets that streak and leaves the speed alone; an alien that lands resets it and steps the speed back 2 WPM, never below where the game started. After any change of speed, up or down, the first three hits are settling hits that don't count, so a fumble while you adjust costs nothing. The upshot: at around 90% accuracy the speed climbs steadily, at 80% it crawls, and below that it all but stops; a speed that beat you takes eighteen clean hits to reach again. Farnsworth spacing doesn't apply, since a single character has no gaps to stretch. The setup card says where a game will open and where it climbs to, the score line shows the speed right now, and the end card shows the highest speed you reached. At a character speed of 15 there is nothing to ramp and every alien is sent at 15. In see-it mode nothing is sent, so there is no ramp: the decoder listens at your character speed.

Characters you miss come back

Within a game, a character you miss, whether an alien that lands or a wrong shot while it was the lowest alien on the field, is dealt more often until you've mastered it again: one miss makes it twice as likely to be the next alien, each hit on it brings that back down, and after two more hits than misses it's back to normal. The weighting is capped (a character is never more than four times as likely as the others), the game still prefers characters not already on the field, and it resets with every new game, so nothing you missed last week follows you. This works the same in both modes.

How it's scored

The end card shows your score, the wave you reached, accuracy, best combo and, in hear-it mode, the speed you reached. Your input, character set and difficulty are remembered for next time.

It counts. Every hit and miss feeds the same per-character stats and confusion matrix as the drills do: a wrong shot records the pair you confused, and an invader that lands is a miss on that character. A game is real practice, not a break from it.

The second arcade game: the same drill as Morse Invaders, but the enemies come at you instead of falling.

Enemies swoop onto the field one at a time along curved paths, alternating from the left and the right, and settle into a formation at the top. Once a few have settled, one at a time peels off and dives at your ship at the bottom; if it isn't shot it loops back to its slot and will dive again. You shoot an enemy by naming the character it carries. Two ways to play, the same two as Invaders:

• Hear it, type it: the enemies fly in unmarked. Each one is sent in Morse as it enters, and sent again as it starts a dive, so the threat announces itself (tap any enemy to hear it again). Type the character you heard on the on-screen QWERTY keyboard, the same one Invaders uses, and the most dangerous enemy carrying it is hit. A hardware keyboard works too.
• See it, key it: each enemy shows its character directly above it. Key it on the on-screen key or a hardware key, and the decoded character shoots the most dangerous one carrying it.

"Most dangerous" is decided for you: a diving enemy first (the lower it is, the sooner), then one climbing back from a dive, then the lowest row of the formation, then anything still flying in. The enemies carry either your active Koch set or the full alphabet and digits, and a character already on the field is not sent again while another is free, so a character usually names exactly one enemy.

The speed ramp

In hear-it mode the game starts slow and speeds up as you land hits, on a gentler ramp than Invaders'. A game opens 8 WPM below your character speed, never below 15, and steps up 1 WPM after every six hits (six in total, not in a row) until it reaches your character speed. A dive that gets through steps the speed back 1 WPM, never below where the game started; a wrong shot leaves it alone. Farnsworth spacing doesn't apply, since a single character has no gaps to stretch. The setup card says where a game will open and where it climbs to, the score line shows the speed right now, and the end card shows the highest speed you reached. In see-it mode nothing is sent, so there is no ramp: the decoder listens at your character speed.

How it's scored

The end card shows your score, the wave you reached, accuracy, best combo and, in hear-it mode, the speed you reached. Your input, character set and difficulty are remembered for next time, separately from Invaders'.

It counts. Every hit and miss feeds the same per-character stats and confusion matrix as the drills and Invaders do: a wrong shot records the pair you confused (what you pressed against the most dangerous enemy on the field), and a dive that gets through is a miss on that character. The session lands in your history as CW Galaga.

Arcade callsign copy: hear a callsign under pressure, find it on the screen, act on it before it's too late.

Cities and ships — the assets — line the bottom of the play field, each with its callsign written under it. Attackers come down from the top in five columns, and each one sends the callsign of the asset it is heading for in Morse as it launches (tap it to hear it again). Nothing on screen says where an attacker is going; only the callsign does. Copy it, find that asset, and route the defence to it before the attacker arrives. Two ways to route:

• Tap the target: tap the asset whose callsign you heard. The attacker aimed at it is destroyed. Tap an asset nothing is aimed at and the shot is wasted.
• Type the callsign: type what you heard on the on-screen keyboard — the same QWERTY layout as Morse Invaders, with a digit row when the callsigns can have digits and a backspace — and the defence routes itself the moment your copy matches an asset. A copy that runs past the longest callsign on the field without matching is a wasted shot. A hardware keyboard works too.

An attacker that reaches its asset destroys it; the rubble stays where it stood so the row doesn't shift under you, and any other attacker aimed at the same asset turns back. The assets are your lives: when the last one falls, the game is over.

Where the callsigns come from

From my characters (the default) spells four-character, call-like groups only from your active Koch set — a digit in the second position when your set has one (K5MR), letters throughout when it doesn't (KMRS) — so nothing is ever sent that you haven't met. Real callsigns uses genuine US-style calls in the everyday shapes (K1AB, AB1C, K1ABC, AB1CD). No two assets on the field ever share a callsign.

The speed ramp

A game opens 8 WPM below your character speed, never below 15, and steps up 2 WPM after every four hits (four in total, not in a row) until it reaches your character speed. Losing an asset steps the speed back 2 WPM, never below where the game started; a wasted shot leaves it alone. Because a callsign is several characters, Farnsworth spacing applies when it's switched on: the characters go at the ramp's speed and the gaps between them stretch to your effective speed. The setup card says where a game will open and where it climbs to, the score line shows the speed right now, and the end card shows the highest speed you reached.

How it's scored

The end card shows your score, the wave you reached, accuracy, best combo and the speed you reached. Your input, callsign source and difficulty are remembered for next time.

It counts. Every route feeds the same per-character stats and confusion matrix as the drills do, character by character: a correct route is a hit on each character of the callsign; a wrong tap or a wrong typed copy is compared position by position against the callsign of the attacker nearest arrival, so routing K1AB when K1AR was sent records B confused with R; an attacker that gets through is a miss on each character it sent. The session record counts one attempt per callsign, and its chart shows the characters.

Copy on the way in, send on the way out: a roguelike where every spell is a word you hear and every counter is a word you key.

You stand in a dungeon room facing one or two monsters. Each one casts a spell by sending a word in Morse (FIRE, STORM, CURSE). Copy it, look it up in the spell book on screen, and key the counter word (WATER, PORT, BLESS) on the on-screen key or a hardware key before the attack lands. The word you key is decoded live; it casts the moment it matches the counter, or, if it doesn't, as soon as a word gap of silence follows it. Tap a monster to hear its spell again; the clock keeps running.

• A counter keyed in time takes one hit off the monster and scores. A counter marked ♥ in the spell book also heals you one life, up to the three you started with.
• A wrong word, or the window closing with nothing keyed, costs a life. Keying the spell itself back instead of its counter is a wrong word; the game tells you so.
• When every monster in the room is down the room is cleared and the next opens. Rooms 1 and 2 hold one monster, later rooms two, and every fifth room is a dragon that takes four hits instead of two.

The spell book

Twenty-four fixed pairs, each spelled from the letters you'd know at some point on the Koch ladder, from TRAP → MAP (eight letters in) to HEX → CHARM (the whole alphabet). With My active characters the book holds only the pairs your current set spells, so a learner partway up the ladder gets a short book of words made from letters they know. If the set spells fewer than three pairs, the four starter spells (which need the first ten Koch letters) are used instead, and the setup screen says so. Full alphabet plays the whole book. The pairs in play are shown on the setup screen and along the bottom of the room during a run.

The speed ramp

Spells are sent at a speed that climbs as you land counters. A game opens 8 WPM below your character speed, never below 15, and steps up 1 WPM after every four counters (four in total, not in a row) until it reaches your character speed. A hit taken steps it back 1 WPM, never below where the game started. Unlike Morse Invaders, a spell is a whole word, so your Farnsworth spacing applies to it when the switch is on: the letters speed up with the ramp, the gaps between them stay at your effective speed. The setup screen says where a game opens and where it climbs to; the score line shows the speed right now; the end card shows the highest speed you reached.

How it's scored

The end card shows your score, rooms cleared, accuracy, best combo and the speed you reached. Your character set and difficulty are remembered for next time.

It counts. Every character of every counter you key feeds the same per-character stats and confusion matrix as the drills do, letter by letter against the counter word: a right letter is a hit, a wrong letter records the pair you confused, and a letter you never reached (short or late) is a miss on that letter. A run is real sending practice, not a break from it.

Arcade crossing: hear a character, find it in the traffic, and hop.

A frog starts on the near bank and has to reach the far one: three lanes of road, a median to catch your breath, then three lanes of river. You move it the classic way, one hop at a time on the four direction buttons under the board (a hardware keyboard's arrow keys work too). Morse decides what is safe.

Every vehicle and every log carries a character from your set. Each lane has a cue: as soon as you land on the row below it the game sends one of that lane's characters in Morse, and the lane is outlined so you know which one the cue is for. Tap the board to hear it again.

• On the road, the vehicle carrying the cue is harmless: you can stand in its path and it passes straight through you, and that counts as a correct decision. Every other vehicle is a hit and costs a life. Empty asphalt is safe, so you can wait for a gap; the cued vehicle is the one you never have to dodge.
• On the river, you have to land on a log, and only the log carrying the cue floats. It carries you sideways with it (and round the far edge, if it gets there). Any other log sinks, and water is water: both cost a life.

A life lost puts the frog back on the near bank with fresh cues. Reaching the far bank completes a wave: the traffic is laid out again a little faster and you start over from the bank, keeping your combo.

The label ladder

How much the labels help changes with the wave, so the same board turns from a reading exercise into a copying one:

The traffic carries either your active Koch set, so the game drills exactly what you're learning, or the full alphabet and digits.

The speed ramp

The cues start slow and speed up as you get lanes right. A game opens 8 WPM below your character speed, never below 15, and steps up 2 WPM after every four correct lane decisions (four in total, not in a row) until it reaches your character speed. Losing a life steps it back 2 WPM, never below where the game started. Farnsworth spacing doesn't apply to a single character. The setup card says where a game will open and where it climbs to, the score line shows the speed right now, and the end card shows the highest speed you reached.

How it's scored

The end card shows your score, the wave you reached, accuracy, best combo and the speed you reached. Your character set and difficulty are remembered for next time.

It counts. Every lane decision feeds the same per-character stats and confusion matrix as the drills do, exactly as Morse Invaders does: a correct decision is a recognition of the cue, timed from the end of its tone; a wrong vehicle or log records the pair you confused (the cue you heard, the character you chose); the water is a miss on the cue with no partner.

Arcade sending: the other half of Morse Invaders. There you name what you hear; here you key what you see.

Your ship sits at the centre of the field and asteroids drift in from the edges, each carrying a visible label. Destroy one before it reaches the ship. Two ways to play:

• See it, send it: key an asteroid's label on the on-screen key or a hardware key. Each decoded character is added to what you've sent so far (the readout under the field shows it): the moment it spells a label, the nearest asteroid carrying that label is destroyed; while it could still be the start of a label, the game waits for the next character; when it can't be, that's a miss and the readout clears. A pause of 2.5 s with a partial label pending clears it silently. You choose which asteroid to send next, so with several on screen, take the one nearest the ship.
• Hear it, tap it: the game sends one asteroid's label in Morse, and only the asteroid carrying it is the target. Tap it; tap the ship (or Hear it again) to hear the label once more. Tapping an asteroid with a different label is a miss; tapping empty space does nothing. When the target is gone, the next one is sent at once.

Asteroids carry either your active Koch set or the full alphabet and digits. The first two waves are single characters. From wave 3 some spawns are larger asteroids carrying a short word (two to five characters from the top 300 common words, spelt entirely from your set) or a generated callsign; the share grows 15% a wave up to half of spawns. A word asteroid splits when hit: one small asteroid per character, fanning out from where it was, each still heading for the ship. In hear-it mode the whole word is sent as one unit.

The speed ramp

In hear-it mode the game starts slow and speeds up as you land hits, more gently than Morse Invaders does. A game opens 8 WPM below your character speed, never below 15, and steps up 1 WPM after every four hits (four in total, not in a row) until it reaches your character speed. An asteroid that reaches the ship steps the speed back 1 WPM, never below where the game started; a miss leaves it alone. Farnsworth spacing doesn't apply. The setup card says where a game will open and where it climbs to, the score line shows the speed right now, and the end card shows the highest speed you reached. In see-it mode nothing is sent, so there is no ramp: the decoder listens at your character speed.

How it's scored

The end card shows your score, the wave you reached, accuracy, best combo and, in hear-it mode, the speed you reached. Your input, character set and difficulty are remembered for next time.

It counts. Every hit and miss feeds the same per-character stats and confusion matrix as the drills do. A destroyed asteroid is a correct recognition of every character in its label; a wrong character keyed, or the wrong asteroid tapped, records the pair you confused (the character that was expected against the one you sent or picked); an asteroid that reaches the ship is a miss on each of its characters. The end card's accuracy counts asteroids; the session in Progress counts characters, so a word is worth its length there.

The jump from copying letters to copying language.

Whole ham-radio words sent as single units: CQ, DE, ANTENNA, RIG. The goal is to stop spelling words out letter by letter and start recognizing their overall shape, the same way you read English without sounding out syllables.

Your own word list

Turn on Use my word list in Settings and paste in whatever you want: your callsign, your local repeater, club names, the vocabulary of whatever you actually talk about on the air. One entry per line, or separated by commas or spaces; the list needs at least two words before the switch will take. The quiz then draws from your list instead of the built-in ranked pool.

The built-in pool is the 1000 most useful words, ranked, and it's size-adjustable: top 100, 300, 500 or 1000. A smaller pool means more repetition per word, a larger one means more variety.

Copying it isn't enough. You have to know what it means.

The everyday shorthand of CW: TNX, HW, OM, FB, 73. The question here isn't "what did you hear?". It's "what are they saying?". You pick the meaning, not the letters.

That distinction matters. On the air you don't have time to copy HW?, then translate it. It has to arrive already meaning "how do you copy me?".

The three-letter codes that carry most of an operating conversation.

QRZ, QSY, QTH, QSB, QRM. You answer with the meaning, same as Abbreviations.

The punctuation of a QSO: run-together patterns that aren't letters at all.

<AR> (end of message), <SK> (end of contact), <BT> (break), <KN> (go ahead, specific station). Each is sent as one continuous character with no gap, which is exactly what makes them hard: <AR> is not "A then R", it's its own sound.

Learn these before your first real QSO. They're how you tell someone you're finished talking, and copying them wrong makes for an awkward contact.

Continuous copy. No prompts, no choices, just text flowing past at speed.

This is the closest thing to real on-air copy. Text is sent continuously and you follow along. Three sources:

• Fables: 32 bundled public-domain short stories. Each is a few minutes of copy, self-contained.
• Longer classics: Sherlock Holmes and friends, sent in parts, with a bookmark that keeps your place between sessions. Genuinely enjoyable once you can copy at pace, because you want to know what happens next.
• Today's news: real RSS headlines, sanitized into sendable Morse and hidden until you reveal them. Decoding is the only way to read them, which is a strong motivator and a nice daily ritual.

Copying without writing anything down, the skill that makes conversation possible.

A whole word is sent. You hold it in your head, then self-grade whether you got it before the answer is shown. No typing, no tapping. The point is to build the mental buffer that lets you hold a callsign or a name while the other station keeps sending.

Auto-repeats

After the prompt plays, Head Copy can replay it a few times so you can re-hear it without mentally replaying it yourself, then counts down to the reveal with a live "Revealing in 3…". Both are in Settings: Auto-repeats from 0 to 3 (default 2) and Auto-reveal after 0 to 10 seconds (default 5; at 0 it waits for you). The manual Repeat button is always there too.

Be honest with the self-grading. It's the only mode where you're the judge, and marking a near-miss as correct only fools yourself.

Free recall. No multiple choice to lean on.

Hear it, type it. Multiple choice lets you work backwards by elimination. Type It removes that entirely, so it's a much truer test of whether a character is actually learned. Expect your accuracy to drop when you first switch over; that drop is the honest number.

A paired hardware keyboard works here (and in the choice quizzes and Journey): just type and the answer submits.

High-speed copy for when 33 WPM has stopped feeling fast.

Free-recall typing at 35, 40, 50 or 60 WPM. At these speeds there is no time for anything but instant recognition, which is precisely the point. It's the mode that proves whether a character is truly automatic or merely fast.

QRQ runs on its own speed setting and does not touch your global WPM, so you can push here without disturbing the speed the rest of your practice uses. 50 and 60 are there for operators who already work QRQ; if 35 is a struggle, that is the right place to stay.

Don't start here. Come back to it once your recognition times at 33 WPM are consistently under a second.

One five-letter word a day, sent in Morse, the same for every player on the same date. The ten-minute habit, with a scoreboard.

Tap the card on the home screen, pick a starting speed, and play the word. Type a five-letter guess and every letter is scored: green is the right letter in the right place, amber is a letter that's in the word but somewhere else, and grey is not in the word at all. Letters you've ruled out are listed underneath so you don't burn a guess on them twice.

There is no limit on guesses or listens. Both are counted, and both bring the code down to you. Each guess has to be a real word, checked against a full dictionary, so CRANE and SLATE are fine even though the answers themselves come from a shorter list of common words.

The speed ladder

This is what makes it a CW game rather than a word game. The word plays at the speed you chose, and the code slows by 5 WPM for every three listens and, separately, 5 WPM for every three wrong guesses, down to a floor of 10. A listen is every play of the word, the first and every replay, so hearing it again is never free, but it is cheap. The two counts step the speed independently and the steps add: four listens and three wrong guesses from 60 WPM puts you at 50. A word you can't catch at 75 comes back within reach, and the number that matters is the lowest speed you heard the word at before the guess that got it, so nudging the dial up just before you answer doesn't inflate the result.

Sharing it

When you get it, the result card shows the word, the lowest speed you heard it at, and what it cost in guesses and listens, and the app builds a pasteable result: a headline with the puzzle number, that speed and the two counts, then one emoji row per guess, then a link back here. The headline reads like this:

Daily Dit #245 — 50 WPM · 4 guesses · 4 listens

Play with the chart hidden and "(no reference)" is added to it. It's plain text on purpose, so it survives being pasted into Discord or a group chat. Copy it or share it straight from the result card; replays after the win are free and don't change the count.

The word is chosen from the date on your own device, not from a server, so it works offline and everyone playing on the 3rd gets the 3rd's word, on iPhone and Android alike. Puzzle #1 was January 1st, 2026. The next puzzle arrives at midnight, your time, even if you leave the app open.

Back-to-back items with no gap to recover in: contest conditioning.

A continuous stream sent at your chosen pace. Choose what to copy and how to answer:

Then pick how to answer: type each one as you hear it, head-copy then type, key each one back, or just listen and review the transmitted list afterwards. The pace between items is adjustable, so tighten it as you improve, and call signs come in the shapes you pick (US-only or with DX prefixes, from chips on the setup screen). All of it, including the two toggles, is remembered between sessions.

The full simulator: you call CQ, stations answer, and you work them. This is the mode that turns copying into operating.

Set your callsign in Settings (it starts as W1AW) and it's used throughout, sent when you call CQ and work stations. With Key my side in Morse on, your own side is keyed on the air, so you hear the whole exchange as it would sound; off, your CQ, calls and TU are logged silently.

Eight exchange flavours

Realism controls

This is where it stops being a quiz. All of these are adjustable:

• Max callers: a cap of 1 to 8 stations answering at once. Lower it mid-run and the surplus callers are sent away straight away, never the one you're working.
• Speed band and tone spread: callers at different speeds (12 to 60 WPM) and different pitches across the band, with their own Farnsworth switch. Zero-beat stacks them all on your exact pitch, which is far harder.
• Wait between callers: the minimum and maximum pause before the next station calls.
• QSB (fading): signals rise and fall mid-exchange.
• QRN (noise): atmospheric noise at selectable levels.
• Cut numbers: digits sent as letters to save time (0→T, 9→N), with a switch per digit for which ones get cut. You can type either form.
• Bust behavior: what happens when you get a call wrong, and whether stations give up and drop out after you keep busting them.
• Who got away: when a caller runs out of patience and leaves, this decides whether you hear about it. At the end (the default) lists the ones who walked in the run summary, along with what you had them as. As it happens also raises a banner mid-run naming the call. Off keeps the run uninterrupted.
• Keep partial call: a partly-copied call stays in the box so you can send ? and add to it rather than retyping from scratch.
• Callsign shapes: US-only, or mix in DX prefixes.

A live log and rate readout run alongside. With Pileup re-calls after TU on, the stations still waiting call again on their own once you log a contact, so the run keeps moving; off, you send AGN or CQ yourself.

When your copy is close but wrong

The pileup answers a miscopy the way an operator would rather than treating everything short of perfect as a total bust:

• A near miss gets corrected. Send N9HS for N9HO and that station comes back with its own call again, and keeps sending it until you get it right. It still counts as a bust and still spends the station's patience, so a caller you never resolve will eventually walk, but you get the chance to fix it.
• An ambiguous copy brings back both. If what you sent sits equally close to two calls, both answer, because neither knows which you meant. Refusing to guess is your job, not the pileup's. Neither one gives up its exchange until you actually name one, so you cannot end up working a station you never copied.
• A partial works from anywhere in the call. Querying the middle or the end is ordinary contest practice: 9H? is how you ask N9HO to come back. The fragment no longer has to be the start of the call.
• A partial can have a hole in it. A ? anywhere in what you send marks a gap, so N?O brings back every call that starts with N and ends with O. And a fragment doesn't have to be consecutive: send NO? and any call carrying those letters in that order answers.

A timed run at a real weekly event, scored the way the real one scores.

Five actual contests, each with its authentic exchange, speed range and scoring:

Pick your run length, and a live score and rate update as you work stations. At the end you get a scorecard.

Why bother: a real contest is unforgiving of hesitation, and there's no way to practise that pressure by copying single characters. A 20-minute SST run before your first real one is worth a great deal of nerve.

The historical FCC/ARRL code test, reproduced faithfully.

The US code requirement was dropped in 2007, so this is a personal benchmark rather than a licence requirement, but it's a well-defined, honest one. Roughly five minutes of plain-language text styled as an on-air QSO, with callsigns, names, locations and rig details.

Two ways to pass

• Solid copy: one minute of unbroken copy, meaning 25 consecutive correct characters.
• Content questions (the default): about ten fill-in questions about what the message actually said.

You can also choose a built-in passage instead of a freshly generated one, if you'd rather practise a known text.

Live CW with real people over the internet. No licence, no radio, no antenna.

Connects to the Vail network, where anyone with a key can call. What you send goes out to everyone on the channel; what they send comes back to you. It's the closest thing to being on the air without being on the air.

• Server picker and a private-channel option: set a channel name and you and a friend have the frequency to yourselves. Ideal for scheduled practice with someone at your level.
• Signal timeline: a visual trace of what's arriving, so you can see the rhythm as well as hear it.
• Chat alongside the CW, for when you need to say something in plain text, with an unread badge so you notice a message that arrived while you were keying.
• Who's on: a roster of the stations on the channel. A station that's gone quiet drops off it after a minute.
• Vail Adapter support: MIDI key input and output, keyer modes and speed, sidetone, and the RX piezo buzz (which you can turn off). The sidetone ducks other audio rather than taking it over, so a radio app can keep running underneath.

Hear it, key it back. Sending and receiving reinforce each other.

A dedicated mode on the adaptive ladder: the app sends a character or word, you key it back, and it decodes what you actually sent and tells you. Not what you meant to send, but what you sent. That feedback loop is what fixes sloppy timing.

• Live decode as you key, so you see your sending rendered back as text in real time.
• Always-on replay: re-hear the target as often as you want.
• Connected MIDI key readout: shows what hardware the app can see, so you know the key is actually talking to it, and says when MIDI itself is unavailable on the device as opposed to no key being connected.

Works with the on-screen touch key or a hardware key. Most people's sending is worse than they think, and the live decode is a fast, slightly humbling way to find out.

Printable practice sheets, for keying away from the app.

Generates a sheet of random character groups you can print or share, then key from at your own bench with a real key and no screen involved. Three kinds:

• Studied: drawn from the characters you've actually learned, evenly weighted.
• Personalized: weighted toward the characters you're currently weakest on.
• Numbers & punctuation: the parts everyone neglects until a contest exchange goes wrong.

Point the microphone at received Morse and read it as text.

Open it from the waveform icon in the top bar. Hold the phone near your rig's speaker, or play a WebSDR through anything, and it decodes what it hears. Live WPM and pitch telemetry show what it's locked onto, and a two-core pitch-lock rescue helps it hang on when the signal wanders.

Noise

Static crashes (QRN) are what confuse a decoder most: each crash lands in the tone detector as a short spurious mark, splitting a character or adding a stray E. Both apps run a noise blanker ahead of the decoder. It catches each crash where it is unmistakable, in the wideband signal rather than at the tone's pitch, and while the crash lasts it substitutes the tone's own continuation, so a crash inside a dah fills the hole instead of breaking it, and a crash in a gap stays a gap. It is always on and has no setting. Steady hiss is left to the decoder, which copes with it on its own down to a poor signal-to-noise ratio. What the blanker cannot save is a band so busy that a third of the time is crash, or another station on the same pitch. Hold the phone close enough that the code sits well above the room, and if your rig has a noise blanker of its own, use it too.

The lookup table you're allowed to use, because looking something up isn't the same as learning by chart.

A browsable, tap-to-hear chart. Tap any row and it's sent, so you can attach a sound to it rather than a written pattern. Six categories:

Practice with your hands and eyes busy: driving, walking, washing up.

Code plays, then the answer is spoken aloud. No interaction at all. It keeps playing with the screen locked, with full background audio and lock-screen controls, so it works with the phone in your pocket. The lock screen, CarPlay and a Bluetooth car display show the current item as the track title (the revealed answer once it has been spoken, "Listening…" before that) with the app logo as the artwork, so a head unit no longer keeps the last podcast's cover up.

It deals the whole pool like a deck of cards, every item once before anything repeats, and the word source is the full 1000-word list, so a long commute doesn't circle back to the same dozen words.

• What it announces: choose the content it draws from. Characters is your active Koch set; QSO elements · Top 20 and QSO elements · Top 100 are the on-air vocabulary below; Words is the 1000-word list; Abbreviations & Q-codes is the same table the Abbreviations and Q-Codes drills use.
• Gap before the spoken answer: how long you get to answer in your head before it tells you. Four tiers: Standard 1.3 s, Rapid Fire 1.0 s, Warp 0.5 s, ICR-Territory 0.2 s.

QSO elements

Dictionary words are poor practice for the air: nobody sends GOVERNMENT in a contact. The two QSO elements sets are one curated list of the hundred things you actually hear, ordered most common first, so Top 20 is the skeleton of every contact and Top 100 adds the rest of a ragchew, a DX exchange and the weather. The first twenty, in order:

CQ  DE  <K>  <KN>  <AR>  <SK>  <BK>  TU  73  RST  599  5NN  UR  NAME  QTH  HR  ES  HW?  R  TNX

Prosigns are sent run-together and shown in angle brackets; a few items are short phrases (CQ CQ CQ, TNX FER CALL, 5NN TU) sent with their word gaps. The reveal shows the token and its meaning, and speaks the token letter by letter and then the meaning, exactly as the Abbreviations set does: "q t h, my location is". Both apps have both sets.

Passive practice isn't as effective as active drilling, but it's dramatically better than not practising, and it turns dead time into reps. A commute a day adds up quickly.