What Is Dictation? Voice Typing and Talk to Text Explained

Dictation is speaking your words so that they are written down as you say them. For most of its history a person did the writing: a secretary, a stenographer, or a typist working from a recording. Today the writer is usually software. You press a key or tap a microphone, you talk, and the words appear in whatever app has the cursor. Apple calls this Dictation, Microsoft and Google call it voice typing, many phone users say talk to text, and engineers call the technology underneath speech to text. All four names describe the same act.

Three things are worth knowing before you start. Dictation is fast: in a controlled study on an iPhone, people entered English text by speech at 153 words per minute ¹, while in an online study of 37,370 volunteers, people typing on their own phones averaged 36.2 words per minute ². Dictation is not transcription (turning a recording you already have into text) and it is not voice control (running the computer by voice), and mixing them up sends people to the wrong tool. And dictation already comes free on every Mac, Windows PC, iPhone and Android phone, although each free tool stops somewhere: a language it does not cover, a need for an internet connection, or an app it cannot reach.

This guide opens the Dictation section. It defines the word, maps the names each platform uses, separates dictation from its neighbors, explains how the software works, and shows what the free tools can and cannot do as of October 2026.

Dictation means speaking your words so they are written down as you say them

The word is far older than the software that now writes it. Merriam-Webster defines dictation as "the act or manner of uttering words to be transcribed," and also as the material that is dictated ³. The word comes from Latin dictare, to say aloud words to be transcribed by another, and its first known use in English dates to 1624 ⁴. The same Latin root gives the verb "to dictate" its other meaning, to give orders, which is why "dictate terms" and "take dictation" share a word.

The noun has three everyday senses, and people searching for "dictation meaning" are usually after one of them:

The classroom sense is a test: Cambridge defines it as one "in which a piece of writing is dictated to students learning a foreign language" ⁵. The business sense already allowed for machines. The Cambridge Business English Dictionary describes dictation as speaking aloud "for a person or machine to record" ⁶. Before software, that machine was usually a dictaphone or a tape recorder whose tape went to a typist, and the fast alternative was a person writing shorthand as you spoke. That history, from shorthand pads to speech recognition, is told in History of Dictation, and the office phrase is explained in what does take dictation mean.

Machines that understand speech are older than most people expect. In 1952, researchers at Bell Laboratories built a system that recognized isolated spoken digits from a single speaker ⁷. It took much longer for software to take dictation of ordinary sentences at conversational speed. That is the sense used throughout these guides: dictation is live speech recognition that writes your own words where you are writing.

"What is dictation in writing" usually points at the other half of the definition: composing a text out loud instead of reading out a text that already exists. Dictating a letter means forming each sentence in your head and saying it, a skill called oral composition. It feels awkward for the first few hours, which is normal and covered in How to Get Used to Dictation. The technique itself, from planning a paragraph to speaking punctuation, is in How to Dictate.

Voice typing, talk to text and speech to text name the same act on different platforms

The differences between these names are branding, not technology. Each company picked its own label, and people search with all of them:

The product names come straight from the makers. Apple's Mac guide says that with Dictation "you can speak to enter text anywhere you can type it" ⁸. On Windows 11 the feature is called voice typing and starts with Windows key + H on a hardware keyboard ⁹. Inside the Word desktop app the button is called Dictate, and Microsoft makes it available only to Microsoft 365 subscribers ¹⁰. Samsung says its Galaxy phones ship two voice-to-text apps, Samsung voice input and Google Voice Typing ¹¹.

"Speech to text" is the widest of the names. It covers dictation, but also live captions, meeting transcripts, voicemail previews and voice search: any system that turns spoken words into written ones, which engineers call automatic speech recognition. So all dictation is speech to text, but not all speech to text is dictation. The question is speech to text the same as dictation works through the edge cases, and speech to text vs text to speech covers the opposite direction, where the computer reads text aloud. The glossary has one-paragraph definitions of dictation and voice typing.

"Talk to text" deserves one note. It is the name most people use for dictating a text message on a phone, and it is often searched as if it were a separate app. It is not: it is the microphone button on the phone's keyboard, which is the same voice typing feature under its plain-English name. On a Samsung phone that button runs Samsung voice input or Google Voice Typing, on an iPhone it runs Apple's Dictation.

Dictation writes live while transcription converts a recording that already exists

The difference is timing and purpose. Dictation happens while you speak, the speaker is the author, and the text lands where the author is writing, so mistakes can be fixed at once. Transcription starts from audio that already exists (a meeting, an interview, a voice memo, a lecture), often with several speakers, and produces a written record of what was said.

The two used to be halves of one job. A doctor dictated into a recorder and a medical transcriptionist typed it up. The US Bureau of Labor Statistics still lists the first duty of medical transcriptionists as listening to "the recorded dictation of a physician or other healthcare worker" and transcribing it into reports ¹². Speech recognition has pulled the halves apart: the BLS says that advances in speech recognition and natural language processing allow physicians to "document patient encounters in real time" ¹³, and it projects employment of medical transcriptionists to decline 4 percent from 2025 to 2035 ¹⁴. Clinical software calls the live version front-end speech recognition and the record-then-edit version back-end speech recognition.

The old split survives in medicine because raw speech recognition still makes mistakes that matter. A study of 217 clinical notes dictated in 2016 found an error rate of 7.4 percent in the draft produced by speech recognition ¹⁵. After a transcriptionist edited the drafts the rate fell to 0.4 percent, and to 0.3 percent in the notes the physicians signed ¹⁶.

The lesson carries over to everyday dictation: somebody has to read the text. With dictation that somebody is you, right after you speak, which is why the same software feels accurate for a text message and risky for a contract. The full comparison is in dictation vs transcription. If what you have is a recording rather than something to say, read Transcribe Audio to Text or use the free in-browser audio to text converter; whether a person or software should do it is weighed in Human vs AI Transcription.

Dictation enters text while voice control operates the computer

Dictation writes words; voice control clicks, scrolls, opens apps and edits by voice. Both listen to the microphone and both understand a few commands, so they are easy to confuse, but they solve different problems. Voice control exists so that someone who cannot use a keyboard or mouse can still use the whole computer, and dictation exists so that anyone can write without typing.

Apple makes the boundary explicit. On a Mac, "when Voice Control is on, you use Voice Control to dictate text; standard macOS Dictation isn't available" ¹⁷. Voice Control has a dictation mode of its own in which any words you say that are not commands are entered as text ¹⁸. On Windows, voice access lets you control the PC and write text by voice without an internet connection ¹⁹, and Microsoft lists the older Windows Speech Recognition as deprecated and replaced by voice access ²⁰.

A third neighbor, the voice assistant, answers rather than writes. Asking Siri to send a message is dictation wrapped in a command; asking it for the weather is not dictation at all. The few voice commands that dictation does understand are listed in Dictation Commands, the line between the two features is drawn in dictation vs voice control, and on a Mac the three are compared in Dictation vs Voice Control vs Siri. Windows users who want full control by voice should read Voice Access on Windows 11.

Every phone and computer now ships a free dictation tool

You do not need to buy anything to start dictating. Every major operating system includes a dictation tool at no extra cost, and for messages, notes and short emails it is often all you need. Each one also stops somewhere, and those limits decide whether you need anything more.

The details behind the table, from each maker:

• Mac. You start Dictation with the Microphone key, if your keyboard has one, or with the Dictation keyboard shortcut ²¹. Apple says you can dictate text of any length without a timeout ²², but Dictation stops when it hears no speech for 30 seconds ²³, and it is not available in every language or region ²⁴.
• Windows. Voice typing needs an internet connection, a working microphone and the cursor in a text box ²⁵. Voice access works offline but is a different tool, described above.
• Microsoft 365. Dictate in Word needs a microphone and a reliable internet connection ²⁶, on top of the subscription.
• Samsung. The keyboard's Offline voice input section lets you download language packs so voice input works without a connection ²⁷.
• Browsers. Speech recognition on a web page uses a server-based engine by default, so it does not work offline ²⁸.

The limit people hit most is reach rather than accuracy. A built-in tool works where the operating system's own text input works. Word's Dictate works only inside Microsoft 365 apps, a website's microphone button works only on that website, and Windows voice typing needs a connection to work at all. People who dictate all day across many apps usually move to system-wide dictation software, compared in How to Dictate Into Any App and Free Dictation Software. Yap, which we make, is one such app: it runs on the device on Mac and Windows (see download and pricing). If you dictate only now and then, the built-in tool is the right choice, and you can try dictation with nothing installed at online dictation.

Whether you ever need to pay is answered in is voice typing free, and step-by-step setup for each device is in How to Turn On Voice Typing on Every Device. Platform guides go further: How to Use Dictation on a Mac, Windows Voice Typing, Android Voice Typing and Dictate in Word.

A dictation system listens, recognizes, punctuates and inserts text in four steps

Every dictation tool, from a 1990s desktop program to a 2026 phone, does the same four jobs in the same order. Knowing them tells you where a problem comes from when the text goes wrong.

1. The microphone listens

The tool opens the microphone when you start it and closes it when you stop or go quiet. Some tools listen until you stop them; others listen only while you hold a key, a style called push to talk. Many dictation errors start here, with a laptop microphone across the room or a noisy café. Best Microphone for Dictation and Dictation and Background Noise cover what actually helps.

2. The recognizer turns sound into words

The recognizer finds the most likely sequence of words for the sound. Older recognizers combined separate models of sounds, pronunciations and word sequences into one large search graph; Google said its production graph was almost 2 GB ²⁹. Many modern recognizers are neural networks trained on large amounts of speech. OpenAI's Whisper models, for example, were trained on 680,000 hours of audio with transcripts ³⁰.

Accuracy is measured as word error rate, the share of words a system gets wrong. In 2017 Microsoft measured professional human transcribers on a standard test of telephone conversations and found a word error rate of 5.9 percent on the Switchboard conversations and 11.3 percent on the harder CallHome conversations ³¹; its own system reached 5.8 percent and 11.0 percent on the same recordings ³².

Benchmarks flatter real life. Accuracy also depends on who is speaking: across five commercial systems tested in 2020, the average word error rate was 0.35 for Black speakers and 0.19 for white speakers ³³. Accents, noise, distance from the microphone and unusual names all raise the error rate. Dictation Mistakes covers the fixes, Word Error Rate explains the measure, and How Speech Recognition Works explains the models.

3. Punctuation and formatting are added

Speech has no commas, so something has to add them. Older tools relied on spoken punctuation: you say "comma" and "new paragraph" and the tool types the mark. Most now add automatic punctuation as well. On iPhone, Dictation inserts commas, periods and question marks as you speak in supported languages ³⁴, and Windows voice typing has an Automatic punctuation setting that inserts punctuation from what you say ³⁵.

The newest addition is a language model that cleans up the raw text after recognition. With Apple Intelligence in iOS 27, dictation uses an on-device model to improve spelling, punctuation and capitalization in English on the iPhone 17 Pro, iPhone 17 Pro Max and iPhone Air, and later models ³⁶. Windows fluid dictation, a Copilot+ PC feature available in English, corrects grammar, punctuation and filler words as you speak ³⁷. The spoken commands that still matter are listed in Dictation Punctuation.

4. The text is inserted where you are writing

Finally the text has to reach your document. Built-in tools type into the focused text field the way a keyboard would, a step called text insertion. On a Mac with Apple silicon you can type during dictation: the cursor pauses while you type, and dictation resumes when you stop ³⁸. Insertion is also where many tools fail: an app that ignores the input, a field that loses focus mid-sentence, or a tool that only works inside its own window. That is why Dictation Not Working has its own guide.

On-device and cloud dictation differ in privacy, speed and what works offline

Where recognition runs decides three things: whether your audio leaves the device, whether dictation works without internet, and how much a slow connection delays the text. Until a few years ago most dictation ran in the cloud because the models were too large for a phone. Google's old server recognizer "requires online connectivity to work properly," it wrote in 2019 ²⁹, when it moved Gboard voice typing onto the phone with an 80 MB recognizer that works even offline, first on Pixel phones in American English ³⁹.

Today the major platforms are split between the two approaches:

Apple says iPhone Dictation is processed on the iPhone in many languages and needs no internet connection ⁴⁰. On a Mac, Keyboard settings show whether your general dictation is processed on the device and not sent to Siri servers ⁴¹. Windows voice typing, by contrast, uses online speech recognition powered by Azure Speech services ⁴².

Cloud does not have to mean stored. Microsoft says Microsoft 365 Dictate does not store your audio or the transcribed text, but your speech is sent to Microsoft to produce the text ⁴³. On-device dictation goes one step further: the audio never leaves the machine, which matters for confidential client, patient or company material and makes dictation work on a plane or in a basement. The trade-off used to be accuracy, since server models were larger, but Google said already in 2019, in its own tests, that its on-device design matched the accuracy of its server models ⁴⁴. Results still vary by language and device.

The short answer to does dictation work without internet is "on Apple devices and Android phones with offline packs, usually yes; with Windows voice typing and Microsoft 365, no". Offline Voice Typing on a Phone and Offline Speech to Text on Windows list the options, Is Voice to Text Safe? covers what happens to your recordings, and On-Device Speech Recognition explains the technology. The glossary defines on-device speech recognition and cloud dictation, and Mac users can read about offline dictation on a Mac.

Dictation fits drafting, messages and notes better than precise editing

Speaking is faster than typing for getting words down, and slower for fixing them. The speed difference is large in each of the studies cited here. In the iPhone study, English speech input reached 153 words per minute against 52 words per minute on the phone's keyboard ⁴⁵, and Mandarin speech reached 123 words per minute ⁴⁶. Outside the lab, an online study of 37,370 volunteers found that people typing on their own phones averaged 36.2 words per minute ², and an online typing test of 136 million keystrokes found that the average participant typed 51.56 words per minute ⁴⁷.

Those numbers measure copying short phrases, not writing. An earlier study puts people's normal speaking speed at 125 to 150 words per minute ⁴⁸, but composing is limited by thinking: earlier research, which used older recognizers, put composition speed with speech software at only 8 to 15 words per minute ⁴⁹. Errors change shape too. In the iPhone study, with a 2016 recognizer, speech produced fewer errors during entry than the keyboard but left slightly more errors in the final text, 1.30 percent against 0.79 percent ⁵⁰. Dictation moves the effort from producing text to checking it.

That makes some jobs a natural fit and others a poor one:

The full comparison, with when each method wins, is in Dictation vs Typing, and the research on speeds is in Speaking vs Typing Speed. If you dictate because typing hurts, Dictation for RSI and Carpal Tunnel covers the setup that cuts down on keystrokes the most, and programmers can read Voice Coding.

Four questions go deeper into what dictation is

Each of these short answers expands one part of this guide:

• Is speech to text the same as dictation? Where the two terms overlap and where they do not.
• What is the difference between dictation and transcription? Live writing versus converting a recording.
• Is voice typing free? What the built-in tools cost, and when people pay.
• Does dictation work without internet? Which devices dictate offline, and how to check.

For the whole subject on one page, from setup to technique, start at the Dictation overview.