← Back to Blog

Short answer: To dictate an email address, say the name, then "at sign", then the domain, then "dot com", with no pauses. Dictation adds spaces around spoken symbols, so you will usually need to close the gaps afterwards. For addresses you use often, a text replacement or vocabulary rule is far more reliable.

Dictation is genuinely good at sentences and genuinely bad at strings. That is the whole problem in one line, and understanding why makes every workaround in this guide obvious rather than arbitrary.

Speech recognition works by predicting the most likely sequence of words given the sounds you made and the context around them. Sentences are full of context. "I'll send the contract on Tuesday" contains a dozen clues that make every word in it easy to guess. An email address contains almost none. It is an arbitrary sequence of characters, often including a surname that is not in any dictionary, a domain that might be a made-up word, and punctuation that has to land in exactly the right places with no spaces anywhere.

So when people search for how to dictate an email address, they are usually looking at something like "john smith at acme dot com" sitting in a To field that will not accept it, and wondering what they said wrong. Usually nothing. They asked a sentence engine to produce a string.

This guide covers the exact phrases that do work, the spacing problem and how to deal with it, the cases where you should stop and type, and the setup that makes the addresses you use constantly come out perfect every time.

The core problem: spoken symbols get spaces

When you say "at sign" or "dot" or "underscore", dictation inserts the character and treats it as a word. Words get spaces around them. That is correct behaviour for prose, where you genuinely want "the cost was 40 percent" to have spaces, and it is exactly wrong for an email address, where you want no spaces at all.

The result is that a dictated address tends to arrive looking like this: john smith @ acme . com. Every character is right. The spacing is wrong in four places. And an email field will reject it outright, which feels like a total failure even though the transcription was near-perfect.

Once you know this is what is happening, the fix stops being mysterious. You are not fighting accuracy. You are closing gaps.

The phrases to say

Here are the spoken forms for the symbols that appear in addresses and URLs. These are broadly consistent across Mac and iPhone dictation:

Two habits improve your hit rate considerably.

Do not pause between the parts. Say the whole address as one continuous run: "john dot smith at sign acme dot com". Pausing invites the engine to treat each fragment as its own phrase, and fragments get capitalised, punctuated and second-guessed independently. A continuous run gives it the context to recognise that this is an address-shaped thing.

Say "dot com" as a unit. Common top-level domains are frequent enough in speech that they are often handled well as a chunk. "Dot com", "dot org", "dot net", "dot co dot uk" tend to come out better than spelling around them.

Spelling out the hard part

The local part of an email address, the bit before the @, is where things break down, because it is frequently a surname or a made-up handle. If you say "Siobhan" and expect it typed correctly, you are relying on luck.

For those, spell it letter by letter, slowly and with a slight separation between letters. This is the one place where the "never speak word by word" rule is inverted, because there is no word to speak. Letter-by-letter dictation is slow, and it is still faster than what most people do instead, which is retype the whole thing twice.

If the letters come out as the wrong ones, the usual culprits are the acoustically similar sets: B, P, D, E, G, T, V, C, Z all sound alike over a mediocre microphone. Slowing down helps more than speaking louder does. If a specific letter keeps failing, that is a microphone problem rather than a recognition problem, and our guide to fixing inaccurate Mac dictation covers the audio-path causes.

How to dictate a URL

URLs have the same spacing problem plus a structural one: the protocol prefix. Say "www dot example dot com" and you will usually get something usable after closing spaces. Say "h t t p s colon forward slash forward slash" and you have spent eight seconds producing five characters that the browser would have added for you.

Skip the protocol. Every browser and nearly every application will resolve example.com without help. Dictate the domain and let the software do the rest.

For deep links with paths, query strings and tracking parameters, stop. A URL with a path segment, a question mark, three parameters and an encoded value is not a dictation task in any meaningful sense. It is a copy-and-paste task, and it always was. There is no technique in this guide that beats Command-C.

The one URL case where voice genuinely wins is speaking a plain domain into a prose sentence, where a stray space is easy to spot and fix and the surrounding words give the engine context. "Send them to acme dot com slash pricing" inside an email body works fine.

How to dictate acronyms and initialisms

Acronyms split into two groups, and they behave very differently.

Acronyms you pronounce as words — NASA, SCUBA, RADAR, and the industry ones like SaaS — are just words to the engine. Say them normally. Whether they come out capitalised correctly depends on whether the engine knows the term. Well-known ones usually do.

Initialisms you spell out — API, CRM, SLA, ROI, KPI, PDF — are harder, because the engine has to decide whether you meant the letters or a similar-sounding word. Say the letters with slight separation and without inserting "dot" between them. If your field uses one constantly and it keeps coming out wrong, that is not something to keep fighting manually.

This is exactly what Smart Vocabulary in Voice Keyboard Pro is for. It is a personal dictionary with replacement rules, so you can teach it the initialisms your work actually uses and have them come out right permanently rather than correcting the same six terms every day. Our post on building a custom dictation vocabulary goes through how to choose which terms earn a slot.

The better approach: stop dictating addresses at all

Everything above is technique for the case where you have to speak a string once. But most of the addresses and URLs you type are not one-offs. They are the same twenty strings over and over: your own work email, your personal email, your two or three most-emailed colleagues, your company domain, your booking link, your support address.

For those, character-by-character dictation is the wrong tool entirely. Use a replacement rule.

Set up a short trigger phrase that expands to the full string. Say "my work email" and have the full address appear, correctly spelled, with no spaces, every single time. Say "booking link" and get the whole URL. This is not a workaround for weak dictation; it is strictly better than typing, because it is faster than typing and it cannot produce a typo.

Smart Vocabulary handles this on Mac. iOS also has a built-in Text Replacement feature in Settings under General, Keyboard, which works alongside dictation and syncs across your devices. Either route gets you to the same place: the strings you repeat become one spoken phrase, and the strings you say once stay a typing job.

Spend ten minutes setting up rules for the fifteen strings you use most and you will have removed the majority of address dictation from your life permanently. That is a far better return than getting incrementally better at spelling out loud.

Dictate the sentences. Expand the strings. Type the one-off URLs. Every mixed-input problem in dictation resolves along that line.

Closing the spaces afterwards

When you do dictate an address and it arrives with gaps, you have three options, in ascending order of sensibility.

You can delete the spaces by hand, which on a Mac takes about four seconds and is completely fine. You can, on iPhone, use Voice Edit to speak the fix rather than trying to place a cursor precisely between two characters on a phone screen, which is the single most annoying interaction on a touch device. Or you can avoid the situation by using a replacement rule for anything you say more than twice.

On iPhone specifically, Voice Edit is worth knowing about here because cursor placement is the real cost on mobile. Selecting the text and saying what you want changed avoids the pinch-and-magnify dance entirely. It is not faster than a replacement rule, but it is much faster than tapping.

Where dictation genuinely should not go

An honest guide names its limits. Do not dictate:

What these share is that a mistake produces a wrong value rather than an awkward phrase. Prose errors are visible; string errors are not. Keep voice on the visible side.

A quick reference table

If you want one thing to remember from this page:

Frequently asked questions

Why does dictation put spaces in my email address?

Because spoken symbols are treated as words, and words get spaces. This is correct for prose and wrong for addresses. There is no setting that changes it, which is why replacement rules are the real answer for addresses you use repeatedly.

What do I say for the @ symbol?

"At sign." Saying just "at" will usually produce the word "at" instead, since that is a far more common word in ordinary speech.

How do I dictate a hashtag?

Say "hashtag" followed by the word, without pausing between them. Because the tag itself is often a compound with no spaces, you may still need to close a gap. For tags you post regularly, a replacement rule is more reliable than saying it each time. Social posting is one of the more common places this comes up.

Can I dictate into the To field of an email?

You can, but the autocomplete in your mail client is almost always faster. Type two or three letters of the person's name and let the client complete it from your contacts. Dictating an address that your software already knows is doing work that has already been done for you.

Does spelling letter by letter actually work?

Yes, with a caveat. Speak the letters with slight separation rather than running them together, and expect the acoustically similar letters to be the ones that fail. It is reliable enough for a one-off address and too slow to be a habit.

Is there a way to make dictation stop capitalising the first letter?

Capitalisation follows sentence structure, so a dictated fragment at the start of a field usually gets a capital. For an address, lowercase it after the fact or use a replacement rule that produces the exact casing you want. Our post on Mac dictation capitalisation explains the underlying rules.

The bigger picture

Most adults type at around 40 words per minute, professional typists reach 80 to 100, and comfortable speech runs at 130 to 150. That gap is why dictation is worth using at all. But the gap only exists for prose. On a fourteen-character email address, voice has no advantage and several disadvantages, and pretending otherwise is how people end up concluding that dictation "does not work" after a bad first afternoon.

The people who get the most out of voice are not the ones who dictate everything. They are the ones who dictate the paragraphs, keep their hands for the strings, and set up a handful of replacement rules so the strings they repeat all day stop being typed at all. That combination is genuinely faster than either pure typing or pure dictation.

Voice Keyboard Pro works system-wide on Mac, hold a hotkey and speak into any application, and as a keyboard with a built-in mic button on iPhone. Smart Vocabulary handles the replacement rules described here. There is a free tier with daily limits if you want to test the setup on your own most-typed addresses first; Pro is $4.99 a month or $34.99 a year.