Dictation Commands: How to Edit and Format by Voice

A dictation command is a spoken phrase that the software carries out instead of typing. The useful ones delete the last phrase, select it, correct it, undo a change, or start a new line. The words differ by platform. On Windows, "delete that" works in voice typing and in Voice Access. On a Mac, Apple's Dictation list does not include a delete command, because Apple's editing commands belong to Voice Control. Microsoft Word has its own set. When a command appears as typed words, the software was in a mode that types everything you say. For a misheard word, the evidence favors speaking the correct words rather than describe the edit, but for good typists the keyboard is often faster.

Commands are spoken words the software acts on instead of typing

A dictation engine turns speech into text. A command is a phrase that the engine maps to an action, such as a line break or a deletion, rather than to words. Two kinds matter in practice. Formatting commands insert punctuation, symbols, line breaks and capital letters. Editing commands act on text that is already on the page, such as deleting, selecting and undoing.

The two kinds are often separate lists. Apple's Dictation page says that its listed commands "are always available while you’re dictating text", and it covers punctuation, typography, formatting, capitalization, math, currency and emoticons ( Apple: Commands for dictating text on Mac). Apple adds that Dictation "also adds certain punctuation automatically as you dictate" for supported languages, so some punctuation needs no command at all. Editing is handled elsewhere, which is the first thing to know before you look for a phrase.

Platform behavior also changes with language. Microsoft's Windows voice typing page says that the commands for other languages are listed separately, and that you change the dropdown to see them ( Microsoft: Use voice typing to talk instead of type on your PC). A phrase that works in English may not exist in French or Japanese, so check the list for your language, not the one for another.

Each platform uses different words for the same command, and one table compares them

The table below lists the phrases that the Microsoft and Apple pages state, for the English versions read on 2026-10-16. A blank cell means the page I read does not list that command for that tool. It does not prove that the feature is missing from every version.

Sources: Apple, Commands for dictating text on Mac; Microsoft, Use voice typing, English (United States) table; Microsoft, Voice access command list; Microsoft, Using dictation in Word.

Three things stand out. Deletion is the most consistent command, because every Microsoft tool uses some form of "delete that". Mac Dictation has almost no editing commands, and its Voice Control mode uses numbers instead: Apple's Voice Control page says that you can say a command such as "Delete" followed by the number of a text item ¹ ( Apple: Voice Control commands). Word has a broad set for formatting, including "Bold that" and "Start list".

Google Docs is not in the table because I could not open Google's voice typing help page while writing this guide. Its commands are covered in the Google Docs voice typing guide, and Word's dictation is covered in the Word dictation guide.

The table covers the built-in tools only. Third-party dictation apps may use different phrases, and so may Yap, which we make, so check their own documentation before you rely on a phrase.

Delete, undo and select are the commands you will use most

Deleting by voice has two forms that are worth knowing. The first deletes the last thing you dictated. The second acts on a selection. Microsoft's Windows voice typing page gives the example "delete that" or "select that" and then notes: "If a word or phrase is selected, speaking any of the "delete that" commands will remove it" ^{2 3}. In other words, if you have selected a phrase, "delete that" removes the selection rather than the last phrase you spoke.

Voice Access gives finer control. Its Edit text table has "Delete that" for the selected or last dictated text ⁴, and separate commands for the previous or next character, word and line ( Microsoft: Voice access command list). It also offers "Select that" for the last dictated text ⁵. Use these when you want to remove a phrase and not the whole dictated paragraph.

Undo works the same way on Windows and Word. Windows voice typing lists "Undo that" and "Revert" as ways to reverse the previous change, and Voice Access lists "Undo that" and "Redo that" ( Microsoft: Voice access command list). In Word, "Undo" is in the editing list, alongside "Delete last word" and "Delete last sentence" ( Microsoft: Using dictation in Word).

Voice is not always the quicker tool for deletion. For one character, pressing Backspace is usually quicker than saying a command. The command earns its place when the phrase is long, the cursor is far from it, or your hands are busy. This is a practical judgment, not a measured result. The Windows list includes "Press Backspace" and "Press Enter" for the same reason, so you can keep the keyboard action in the same flow.

Correcting a misheard word works best with select and say again

Correction takes more time than dictation. In a preliminary 2001 study of dictation with a recognizer of that time, participants, including people with spinal cord injuries, "spent less than 35% of their time dictating and over 65% of their time correcting errors" ⁶. The same study found that participants spent over 32 percent of their time moving from one place to another within the document, which here means reaching the places that need correction ⁷. The figures come from an early, small study, so treat them as a sign of how much editing can take, not as a current average. A later paper by Keith Vertanen reached the same qualitative conclusion, saying that "correcting errors dominates task time" ⁸.

Given that, the question is how to fix a word quickly. The evidence points to speaking the correct words rather than describing the change. In a 2021 study of eyes-free text editing, the authors' editing techniques "decrease the average failure rate by 54.1%" compared with descriptive commands or re-speaking ⁹. That study used earphone buttons and not a desktop, so it supports the principle more than any one product.

In practice, the method is the same on each platform. Select the wrong words, then dictate the correct ones. The steps differ by tool:

1. Windows voice typing. Say "Select that" to select the last dictated phrase, then say the replacement. The Microsoft list gives "Select that" only for the last phrase, so a misheard word in the middle of a sentence needs a mouse or keyboard selection first.
2. Windows Voice Access. Say "Correct that" for the last dictated text ¹⁰, or "Correct" followed by the words you want to fix. Voice Access also accepts "Select" followed by the text to select, and "Spell that" to spell out the last dictated text letter by letter ( Microsoft: Voice access command list).
3. Word. Microsoft says that "At the moment, corrections cannot be made fully with voice" ¹¹. Use the suggestions that Word marks under words it may have misheard, or the keyboard, without turning off the microphone.

Most editors replace a selection when you start typing, so dictating over a selection often works. Test this in your own app, because the Microsoft pages I read do not describe the behavior. Speech is not always the fastest correction method. A 1999 study that compared several correction methods found that multimodal correction, which combines speech with other input, was slower than correction by typing for users with good typing skills ¹². Vertanen's 2006 paper also reports that a correction spoken again is often misrecognized as well, which costs time ( Vertanen, 2006) ¹³.

If a correction still fails, the usual cause is that the recognizer heard the replacement as a command or as a different word. The dictation mistakes guide covers the causes of recognition errors. For the mechanics of selecting text in each app, see the Mac dictation commands guide and the Windows voice typing commands guide.

Formatting commands work in some apps and not others

Formatting commands are the second family. They insert line breaks, paragraphs, symbols and capital letters, and the Microsoft and Apple lists show that they are not the same on every platform. Apple's Mac Dictation has "caps on", "all caps" and "numeral" for formatting ( Apple). Windows voice typing has "new line" and "new paragraph". Word has a broad set, including "Bold that", "Underline last sentence", "Start list" and "Next line". Microsoft's page labels this part as a preview: "Dictation is more powerful with punctuation and voice commands (preview)" ( Microsoft: Using dictation in Word).

The app you dictate into decides which commands exist. Word's Dictate button is part of Word, so its commands only apply there. Windows voice typing works in any text box, but its phrases follow the language setting, as noted above. Apple's Dictation works in any text field on the Mac, but its list stops at formatting and symbols. This is why a formatting phrase that works in Word may not work in an email.

Commands are also a matter of setting. Word lists "Voice Commands" in its dictation settings, and says that it is "turned on for supported languages" by default ( Microsoft: Using dictation in Word). Windows voice typing has an "Automatic punctuation" option, so you may not need the punctuation phrases at all ( Microsoft: Use voice typing). The spoken punctuation and automatic punctuation terms explain the difference, and the punctuation guide covers the phrases in more detail.

Learn a few formatting commands at a time. A 1992 study of speech and mouse commands in word processing found that speech activated commands faster: on average, task time fell by 18.67 percent ¹⁴. The same study found that participants made significantly more memorization errors when they used speech for command activation ¹⁵. The speed came with a cost in remembering the phrases, so a short list you use every day is worth more than a long one you forget.

A command spoken mid-sentence is often typed as text

This is the most common complaint, and the cause is simple. The software has to decide whether a phrase is a command or ordinary speech, and in a mode that types everything, it types. Apple says so directly: "Any words you say that aren’t Voice Control commands are entered as text" ¹⁶. Voice Control's default is Dictation mode, and Apple states that "Dictation mode is on by default" ¹⁷.

Apple also gives the fix. To keep a command from being typed, "have Voice Control respond only to commands (with Command mode)" ¹⁸. On a Mac, turning on Voice Control also replaces standard Dictation, and Apple notes that "standard macOS Dictation isn’t available" while Voice Control is on ¹⁹. So the typed-command problem and the mode problem are the same problem on a Mac.

Windows Voice Access has a deliberate prefix for this. Its Dictate text table has the phrase "Type" or "Dictate" to insert a Voice Access command as text, so a command word can be entered on purpose ( Microsoft: Voice access command list). Voice typing offers a "Wait time before acting" setting, which Microsoft describes as the time that voice typing waits "before carrying out a voice command", useful for speakers who pause differently ( Microsoft: Use voice typing).

Three practical habits help:

• Pause briefly before and after a command, so that it stands alone. This is advice, not a measured effect.
• Use a command mode or a "Type" prefix when a command word is also a normal word in your text.
• If you keep getting a phrase typed as text, check the mode first, then the language setting, then the wait time.

The reason is the same for every platform: a phrase such as "delete that" is also ordinary English. Once the software is listening for commands, the ambiguity mostly disappears. The question on why dictation types a command goes through the checks in order, and the dictation not working guide covers the cases where nothing is typed at all.

Full voice control is a separate accessibility mode with its own command set

Dictation commands are a small list. Full voice control is a different mode with a much larger command set, built for people who use the computer by voice all day. The two are often confused, so it is worth keeping them apart.

On a Mac, Voice Control is the mode that lets you open apps, click, scroll and dictate. Apple says that you can speak commands to "navigate the desktop and apps, interact with what’s on the screen, dictate and edit text" ²⁰, and that when it is on, standard Dictation is not used ¹⁹ ( Apple: Voice Control commands). The question on the difference between dictation and voice control explains the distinction, and the Mac voice control commands guide lists the phrases.

On Windows 11, the equivalent is Voice Access. Microsoft describes it as a feature that lets you "control their PC and author text using only their voice and without an internet connection" ²¹. The offline point matters for privacy and for places with no connection. Voice typing, by contrast, needs the internet: Microsoft says that "To use voice typing, you'll need to be connected to the internet" ²². The Windows voice access commands guide covers the full list, and the guide to dictation for RSI and carpal tunnel covers who uses these modes and why.

Use the full modes when you need hands-free control of the whole computer. Use dictation commands when you are writing and only need to fix text. Control modes are built for hands-free navigation of the whole computer, and dictation commands are built for editing one document. Using either for the other job is a poor fit.

For heavy editing, voice and keyboard together beat either alone

The evidence does not support voice-only editing for heavy work. Dictation itself is fast. In a controlled study on an iPhone, English text entered by speech reached 153 words per minute against 52 on the built-in keyboard ²³. The same study found that speech made fewer errors during entry but left slightly more errors in the final text: an uncorrected error rate of 1.30 percent, against 0.79 percent on the keyboard ²⁴. So the drafting gain is real, and the cleanup cost is also real.

On the editing side, a 1999 study grouped participants by typing skill. Its participants' average typing rates on plain text were "23, 35, and 40 wpm for the slow, average, and fast typists" ²⁵. Among good typists, correction by typing was faster than multimodal correction, which combines speech with other input ¹². The study is old and small, but it explains why the advice differs by typing skill.

The chart above puts the typing rates from the 1999 study next to the keyboard and speech entry rates from the 2018 iPhone study. The two sets come from different studies, with different tasks and devices, so read them as separate reference points and not as one comparison. Even the fastest typists in the 1999 group were well below the 2018 speech rate, which is why dictation is worth learning for drafting. The cleanup time covered in the sections above is the cost that the chart does not show.

For most people, a workable pattern is this. Dictate the draft without stopping to edit. Use "delete that" or "select that" for whole phrases you spoke badly. Use the keyboard or mouse for precise corrections in the middle of a sentence, where a voice command would need an exact selection. Use Word's suggestions or the keyboard for words the recognizer marks. This is a judgment from the evidence above, not a single tested method, and it works whether you use Apple, Windows or Word.

If you use Yap on Mac or Windows, the same principle applies: the built-in commands come with your operating system, and a paid app is worth it only for what the built-in tools cannot do. The overview of dictation software compares the options, and the guide to how Apple dictation works explains Apple's engine. You can also check the download page for the current version.

Related questions

If you searched for one of these, the short answers are on their own pages: how to delete text by voice, how to correct a misheard word by voice, and why dictation types my command. For the overview, start at Dictation and voice typing. The voice command and voice typing terms define the wider vocabulary.