Dictation and Voice Typing: How to Talk Instead of Type

Dictation is speaking your words so that they are written down as you say them. Today that means software typing live into whatever app has the cursor, and every Mac, Windows PC, iPhone and Android phone ships a free version. Speech is fast: in a controlled phone study, English text entered by voice reached 153 words per minute against 52 words per minute on the keyboard ¹. But fast speech is not the same as fast writing. The skill lies in three habits: compose before you speak, keep the microphone close, and draft by voice, then edit by eye and keyboard.

The evidence behind those habits is old and consistent. In IBM's 1978 study of letters, about two-thirds of composing time was planning, whichever way the letter was produced ², and novices learned within a few hours to dictate letters as fast and as well as they wrote them ³. Software adds one cost that a human secretary did not: recognition errors. In clinical notes, the raw speech recognition draft had 7.4 errors per 100 words ⁴, and the notes physicians signed after transcriptionist review had 0.3 percent errors ⁵.

This page covers the whole subject in order: what dictation is, how to start, when it beats typing, how to dictate, punctuation, commands, mistakes, fixes, microphones, noise, building the habit, what speaking does to your writing, dictation machines and the history. Each section gives the short answer and links to the guide that covers it in depth.

Dictation is speaking so that software writes your words as you say them

Dictation, voice typing, talk to text and speech to text all name the same act: you speak, and text appears at the cursor. The names follow the platform. Apple calls it Dictation, Windows calls it voice typing, and phone users often say talk to text. The full explanation, with the names each platform uses, is in What Is Dictation?, and the short answer to whether the terms differ is in Is speech to text the same as dictation?.

Dictation is easiest to understand next to the three things it is often confused with. Transcription converts a recording that already exists, so you can do it hours later and with several speakers. Voice control operates the computer itself, opening apps and clicking buttons. Typing is the baseline everyone compares against.

The line between dictation and transcription is mostly about time: dictation writes while you talk, transcription works on audio after the fact. What is the difference between dictation and transcription? explains where the line blurs, for example when you dictate into a recorder and run the file through speech recognition later.

Whether dictation needs the internet depends on the engine. Windows voice typing uses online speech recognition powered by Azure Speech services ⁶, while Windows 11 Voice Access writes text by voice without an internet connection ⁷. On a Mac, Apple's Keyboard settings say whether your dictation is processed on the device or sent to Apple's servers. The questions Does dictation work without internet? and Is voice typing free? cover both in short form, and the voice privacy section covers what happens to your audio.

Every computer and phone already has free dictation, one setting and one shortcut away

You do not need to buy anything to start. The built-in tools are free, need no voice training, and take a few steps to turn on. How to Turn On Voice Typing on Every Device gives the steps for each platform; the table below is the short version, checked against the platforms' own help pages.

Windows voice typing needs three things: a working microphone, an internet connection and the cursor in a text box ⁸. On a Mac with Apple silicon you can keep typing while dictation is running; the cursor pauses while you type and resumes when you stop ⁹. Mac Dictation stops by itself after 30 seconds with no speech ¹⁰, which can make it seem to switch off in the middle of a thought.

The platform sections hold the details: dictation on Mac, dictation on Windows and dictation in your apps, including Google Docs voice typing and dictating in Word. Short answers live in What is the shortcut for voice typing?, Can you use talk to text on a computer? and How do you turn off voice typing?. If you want to try dictation without changing any settings, How do you voice type online for free? explains the browser options, and our free online dictation page works in a browser tab.

Speaking is about three times faster than typing on a phone, but composing sets the real pace

For raw words, speech wins clearly. In the Stanford and University of Washington phone study, English speech reached 153 words per minute against 52 on the iPhone keyboard ¹¹, and Mandarin speech reached 123 words per minute against 43 with the Pinyin keyboard ^{12 13}. Typing speed is a different baseline: a study of 136 million keystrokes from an online typing test found that the average participant typed 51.56 words per minute ¹⁴.

Those numbers measure copying text you were given, not writing your own. Two costs pull dictation back toward typing speed when you compose:

• Thinking. Planning took about two-thirds of composing time in the IBM letter study, regardless of method ², so a faster output channel speeds up only the remaining third. People with years of dictation experience were only 0 to 25 percent faster than novices, depending on how complex the letter was ¹⁵.
• Correcting. Speech left slightly more errors in the final text than the keyboard in the phone study (1.30 percent against 0.79 percent uncorrected) ¹⁶. In an early study of dictation software, users spent over 65 percent of their time correcting errors ¹⁷, and Vertanen reports that correction takes up most of the task time with a large-vocabulary recognizer ¹⁸.

The practical answer is to split the work. Dictate first drafts, messages, notes and anything long, and use the keyboard for precise edits, code, tables and rooms where speaking aloud is awkward. Dictation vs Typing weighs speed, accuracy, effort and setting with the research, and the short answers are in Is dictation faster than typing?, When is typing better than dictation? and Can you dictate and type at the same time?. The typing speed section has the full speed data, and How Much Time Dictation Saves at Work puts it in working hours.

Good dictation means composing each sentence before you say it

The single habit that most improves dictation is deciding what a sentence will say before you start saying it. Because planning dominates composing time ², the way to dictate faster is to plan in short bursts and then speak each sentence as one even phrase. Recognizers reward this: Nuance's Dragon guidelines recommend longer phrases over single words, because hearing a whole phrase helps the software predict what you say ¹⁹.

A workable routine for anything longer than a message looks like this:

1. Outline first. Three to five points keep a long dictation on track and stop you from circling.
2. Compose, then speak. Pause before the sentence, not in the middle of it. Mid-sentence restarts end up in the text.
3. Keep going. Leave the fixing for the end. Editing while you dictate breaks the train of thought that dictation is good at keeping.
4. Edit in a separate pass, by eye and keyboard.

Filler words are a planning symptom, not a microphone problem. When you start speaking before you know where the sentence ends, "um" and "uh" fill the gap, and the software may type them. Some tools now remove them: Microsoft says Fluid dictation on Copilot+ PCs corrects grammar, punctuation and filler words as you speak ²⁰. The rest of the fix is the habit above.

How to Dictate explains each technique with examples for letters, notes and long drafts. The common situations have short answers: How do you dictate a letter?, How do you dictate notes quickly?, Should you edit while you dictate or afterward? and How do you stop saying um when you dictate?. The glossary defines filler words and oral composition, the name for composing aloud. For dictating whole books or emails, see Writing a Book by Dictation and How to Dictate Emails.

You punctuate either by saying the mark or by letting the software infer it

There are two ways to get punctuation into dictated text. Spoken punctuation means saying the name of the mark: on iPhone you say the mark's name, such as "exclamation point" ²¹. Automatic punctuation means the software guesses commas, periods and question marks from your phrasing. In supported languages, iPhone Dictation inserts commas, periods and question marks as you speak, without a command ²², Mac Dictation adds certain punctuation automatically ²³, and Windows voice typing has an Automatic punctuation setting ²⁴.

Most tools do both, and the trouble starts when you mix them. If automatic punctuation is on and you also say "period", you can get a doubled mark or a stray word. Pick one style per session. Automatic punctuation can miss marks, so people who care about exact punctuation often say the marks. Researchers made the same point years ago: for dictation systems, where punctuation matters most, a style with explicit dictation of the punctuation marks was foreseen ²⁵.

Line breaks and formatting have their own commands, and they differ by platform:

Dictation Punctuation gives the full spoken-punctuation table across Apple, Windows, Google and Microsoft 365, and explains when automatic punctuation gets it wrong. Short answers: Does voice typing add punctuation automatically?, How do you say a question mark in talk to text?, How do you start a new line or paragraph by voice?, How do you dictate numbers, dates and symbols? and How do you capitalize a word when dictating?. The glossary entries for spoken punctuation and automatic punctuation define both terms.

Voice commands handle small edits, and the keyboard is faster for heavy ones

A voice command is a phrase the software acts on instead of typing. Every dictation tool has a few: Windows voice typing uses commands such as "delete that" and "select that" to edit what you just dictated ³¹, and in Windows Voice Access, "Delete that" removes the selected text or the last dictated text ³². Commands are worth learning for the edits you make constantly: deleting the last phrase, undoing, and starting a new paragraph.

Commands have two limits. First, a command spoken in the middle of a sentence is often typed as text, because the software cannot tell an instruction from a sentence that happens to contain the same words. Apple's answer in Voice Control is a separate Command mode, so that it responds only to commands and does not type them ³³. Second, plain dictation tools stop short of full editing. Microsoft says corrections in Word dictation cannot be made fully by voice ³⁴, and in a 1999 study, correcting by speech alone was slower than multimodal correction, which mixes speech with keyboard and mouse, and multimodal correction was still slower than typing for people with good typing skills ³⁵.

Full voice control is a different mode built for people who cannot use a keyboard or mouse, with hundreds of commands. On a Mac, turning on Voice Control replaces standard Dictation ³⁶. If your hands work, the fastest editing workflow is mixed: voice for drafting, keys for fixing. Commands still pay off for formatting: a 1992 study found word-processing tasks took 18.67 percent less time when speech activated the commands ³⁷.

Dictation Commands compares the commands side by side across Apple Dictation, Voice Control, Windows voice typing, Voice Access, Google Docs and Word. Short answers: What is the difference between dictation and voice control?, How do you delete a word by voice?, How do you correct a misheard word by voice? and Why does dictation type my command instead of doing it?. For full hands-free control, see Voice Control on Mac and Voice Access in Windows 11. The glossary defines voice command.

Dictation errors are predictable, and proofreading removes most of them

Speech recognition is good, not perfect, so every dictated draft needs a read-through. A large study of real dictation is clinical. In dictated clinical notes, the speech recognition draft had an error rate of 7.4 percent ⁴, and 96.3 percent of drafts had at least one error ³⁸. After a transcriptionist edited the drafts and physicians signed them, the rate fell to 0.4 and then 0.3 percent ⁵: review cut errors about 25-fold ³⁹.

Errors tend to come from five causes:

1. Sound-alike words ( homophones), which a spell checker never flags. Mac Dictation underlines ambiguous words in blue so you can click one and pick an alternative ⁴⁰.
2. Names, brands and jargon the engine has not learned. Custom vocabulary or replacement rules fix these where the tool supports them.
3. Word-by-word speech, which removes the context the recognizer uses to choose between candidates.
4. Distance and noise, covered in the microphone and noise sections below.
5. Who is speaking. Engines are not equally accurate for everyone: across five commercial systems tested in 2020, the average word error rate was 0.35 for Black speakers and 0.19 for white speakers ⁴¹. The languages and accents section covers this in depth.

Proofreading dictated text needs its own method. Read it with your eyes rather than listening to it: in one study, people missed over 50 percent of recognition errors when they checked by listening ⁴². Some changes to your words are also deliberate. Windows voice typing has a Filter profanity setting ⁴³ that lets you block or allow profanity being typed.

Dictation Mistakes sorts the errors by cause with the fix for each. Short answers: How do you improve dictation accuracy?, Why does talk to text censor words?, How do you get dictation to spell names correctly? and How do you proofread dictated text?.

When dictation fails, five checks in a fixed order find the cause

When dictation does nothing at all, the cause is nearly always one of five things, and checking them in order is fastest:

1. The feature is off or restricted. No microphone icon at all points here.
2. The app has no microphone permission, or the operating system blocks it.
3. The wrong microphone is selected. An icon that listens but types nothing usually means this. Bluetooth headsets often switch microphones on their own.
4. The engine is missing something it needs: a network connection for cloud dictation (Windows voice typing needs one ⁴⁴), or the right dictation language. A wrong language produces gibberish rather than silence.
5. The field refuses dictated text. Password boxes and some apps block it by design.

Dictation that starts and then stops is a different problem: on a Mac it ends after 30 seconds of silence ¹⁰, so a long pause to think ends the session.

Dictation Not Working? walks through the order with the symptom each step points to. Short answers: Why is dictation on but not typing anything?, Why is the dictation shortcut not working?, Why does dictation work in some apps but not others? and Why does dictation keep stopping?. Platform fixes are in Why Mac Dictation Is Not Working, our Mac dictation troubleshooting page and Windows Dictation Not Working?.

A close microphone matters more than an expensive one

Distance is the strongest lever you control. Sound level drops by about 6 dB every time the distance from the source doubles ⁴⁵, so a microphone at 12 inches picks up your voice about 21.6 dB more weakly than one at 1 inch, under ideal conditions ⁴⁶. Room noise does not drop in the same way, so a close microphone raises your voice above everything else. Nuance's guidance for a headset puts the microphone about an inch from the side of the mouth ⁴⁷.

The research shows the same thing in error rates. Close-talking sentences recorded in a studio gave a word error rate of 3.7 percent ⁴⁸; in a different study, real recordings made with a single distant ceiling microphone in an apartment gave 12.0 percent ⁴⁹.

Four kinds of microphone suit dictation: a wired or USB headset (the most consistent for long sessions), a handheld speech microphone, a desk microphone, and the laptop's built-in array (fine in a quiet room). Bluetooth earbuds are the least predictable. When their microphone is in use, many switch to the Bluetooth Headset Profile, which carries voice at 8 kHz, or at 16 kHz with the wideband codec ⁵⁰, while dictation and other large-vocabulary recognition mainly uses at least 16 kHz ⁵¹.

Best Microphone for Dictation explains the four types and a two-minute test for your own setup. Short answers: Do AirPods work for dictation?, Is a laptop microphone good enough for dictation?, Is a wireless headset good for dictation? and How close should the microphone be when you dictate?. The glossary defines a noise-canceling microphone.

Noise and echo lower accuracy, so a quiet spot beats any setting

A recognizer has to separate your voice from everything else in the room, and the harder that is, the more words it gets wrong. In a study of noisy, accented oral-history recordings, signal-to-noise ratio and syllable rate were the two dominant factors behind the error rate ⁵². Apple's own advice for Dictation in a noisy room or one with a lot of echo is that a headset microphone may help ⁵³.

In practice, three things help in this order: move the microphone closer, move somewhere quieter or softer (hard, empty rooms add echo), and speak at a steady volume rather than raising your voice. Open-plan offices add a social problem: colleagues hear you. Professionals who must dictate where others can hear use speech privacy masks, and some hospitals set aside dictation rooms for the same reason. Court voice writing uses a mask microphone held over the mouth so that the reporter can repeat testimony without being heard. Cars and public transport work for short notes, not long drafts.

Dictation and Background Noise covers each setting. Short answers: Can you dictate in a noisy room?, Can you whisper to dictation software?, What is a voice dictation mask? and What is a dictation room in a hospital?.

Composing aloud takes hours to learn, and the software habit takes weeks

The two parts of learning to dictate run on different clocks. Composing aloud is quick to learn: in IBM's study, novices dictated letters as fast and as well as they wrote them within a few hours, although they did not believe they were doing that well ³. That gap between performance and confidence is the awkward feeling most beginners report. A 1983 study found that some versions of an imperfect listening typewriter could be at least as good as handwriting and dictating for composing letters, even on first use ⁵⁴.

Working smoothly with recognition software takes longer, because the hard part is correcting. Novices keep fixing errors by voice; users who keep dictating for about 20 hours start to mix voice correction with keyboard and mouse ⁵⁵. In a hospital study, the rate of words the recognizer did not know leveled off after about 10 weeks for new users ⁵⁶. Users become more efficient at dictating, navigating and correcting as they gain experience ⁵⁷.

The habit itself follows ordinary habit research. In a 2010 study of daily behaviors, the time to reach 95 percent of the automaticity ceiling ranged from 18 to 254 days between people ⁵⁸. A short daily session tied to a task you already do (messages in the morning, notes after meetings) is an easy place to start. Modern built-in tools need no voice training, unlike older desktop software.

How to Get Used to Dictation turns this into a plan. Short answers: How long does it take to get used to dictation?, Why does dictating feel so awkward at first?, Do you need to train dictation software to your voice? and How do you practice dictation?. The glossary defines push-to-talk, the hold-a-key style of starting dictation that keeps stray speech out of your text.

Speaking changes the first draft more than the quality of the final text

The research points one way: composing is the core skill, and the output method is secondary. Because planning takes about two-thirds of composing time whatever the method ², switching from keys to voice changes the speed of the last third and the texture of the draft, not the quality of the thinking. Dictated drafts tend to come out longer and more conversational, closer to speech, and they need a revision pass by eye to read as written prose.

Dictation can raise quality, not just speed, for some writers. In a study of high school students, those with learning disabilities wrote better essays by dictating to speech recognition than by hand, and better still dictating to a human scribe, while students without learning disabilities showed no difference ⁵⁹. Both groups kept total word errors under 10 percent ⁶⁰ while they learned the software. Lowering the cost of getting words out helps the writers for whom that cost was highest.

Thinking aloud helps some writers and distracts others. Speaking makes it easy to keep going, which helps against a blank page, and it also makes it easy to ramble. The fix is the same as for speed: outline, compose each sentence, then revise.

Does Dictation Change How You Write? reads the studies in detail. Short answers: Is dictated writing worse than typed writing?, Why does dictated text sound different from typed text? and Can dictation help with writer's block?. For dictation as an aid for dyslexia, ADHD or pain, see the accessibility and health section, starting with Speech to Text for Dyslexia.

Dictation machines still serve medicine and law

A dictation machine records your voice so that a typist or speech recognition can turn it into text later. The workflow is old: from the earliest dictation machines, typists transcribed recordings on playback machines with foot pedals that controlled speed and let them reverse and repeat ⁶¹. Today's version is digital. A Philips foot pedal starts, pauses, rewinds and fast-forwards with a slight movement of the foot, leaving both hands on the keys ⁶², and the DSS file format was designed for digital voice recorders, with high compression and little loss of quality ⁶³.

The modern kit has four parts: a handheld recorder or a desk microphone with record, play and rewind buttons, a dictation foot pedal for the typist, and workflow software that routes files from author to typist and back. That system is called digital dictation. Doctors and lawyers still use it because it fits a division of labor where someone else produces the final document. Live speech recognition has taken over much of that work: at one community hospital, the median radiology report turnaround fell from 24 hours to 1 hour after speech recognition was introduced ⁶⁴, and US employment of medical transcriptionists is projected to decline 4 percent from 2025 to 2035 ⁶⁵.

For personal dictation, a phone or a laptop with a close headset now covers what a recorder did. Dictation Machines explains each device and who still needs one. Short answers: What is a dictation foot pedal?, Can you use a voice recorder for dictation?, Is there a device that types what you say? and What is a SpeechMike?. For clinical and legal workflows, see Medical Dictation and Legal Dictation Software.

Dictation moved from scribes to shorthand, to machines, to software

People dictated to scribes long before anyone dictated to a machine. Each change since has moved the writing closer to the person doing the thinking.

Shorthand let a trained secretary write as fast as an executive spoke. The skill survives in court reporting: the Registered Professional Reporter (RPR) certification requires transcribing dictation at 180, 200 and 225 words per minute with 95 percent accuracy ⁶⁶. The glossary covers shorthand, and What is shorthand dictation? and What does it mean to take dictation? answer the common questions.

Machines came next. Thomas Edison believed the phonograph's main use would be in business, as a dictation machine ⁶⁷. Graphophone dictating machines began using the Dictaphone name around 1907, and by 1908 at the latest ⁶⁸. The name became the common word for dictation machines ⁶⁹. Uptake was slow: by 1945, Dictaphone estimated it had captured only 15 to 25 percent of its potential market ⁷⁰. Wax cylinders gave way to the plastic Dictabelt in 1947, which held about 15 minutes ⁷¹, and to cassette tape in the 1960s ⁷². For most of the twentieth century, business people recorded letters and a secretary typed them up ⁷³.

Speech recognition began as a lab problem. IBM's Shoebox of 1961 recognized 16 spoken words ⁷⁴. Early dictation software either had a small vocabulary or made you pause between every word ⁷⁵, until Dragon Systems unveiled NaturallySpeaking in 1997, the first general-purpose, large-vocabulary continuous speech recognition product ⁷⁶. The latest change is scale: OpenAI's Whisper models were trained on 680,000 hours of audio ⁷⁷, and their authors report that they approach human accuracy and robustness ⁷⁸.

History of Dictation traces each stage with dates and primary sources. Short answers: What is a Dictaphone?, Which famous authors dictated their books? and When did computers first type what you say?. The glossary defines Dictaphone. For Dragon's later history, see Dragon NaturallySpeaking.

Start with the free tool you own, and choose software only for a reason

For most people, the built-in tool is enough to learn the skill: Apple Dictation, Windows voice typing, or the microphone on your phone keyboard. Move to separate software when you hit a specific limit, such as needing dictation without an internet connection on Windows, dictation that works the same in every app, a custom vocabulary, or a different shortcut style. The dictation software section compares the categories, and How to Choose a Dictation App lists what to test before paying.

Yap, which we make, is one option in that category: an on-device dictation app for Mac and Windows that types into any app. You can see what it does on the download page and the prices on pricing. Whatever you use, the habits on this page matter more than the app: compose first, keep the microphone close, draft by voice and proofread by eye.

Fourteen guides cover each part of dictation in depth

Each guide covers one part of the subject in depth:

1. What Is Dictation? Voice Typing and Talk to Text Explained
2. How to Turn On Voice Typing on Every Device
3. How to Dictate: Techniques for Clean, Usable Text
4. Dictation Not Working? How to Fix It on Any Device
5. Dictation Punctuation: How to Say Commas, Periods and More
6. Dictation Commands: How to Edit and Format by Voice
7. Dictation Mistakes and How to Make Voice Typing Accurate
8. Dictation vs Typing: When Speaking Beats the Keyboard
9. Dictation Machines: What They Do and Who Still Uses Them
10. Best Microphone for Dictation: What Actually Matters
11. How to Get Used to Dictation and Build the Habit
12. Dictation and Background Noise: Where and How to Dictate
13. Does Dictation Change How You Write?
14. History of Dictation: From Shorthand to Speech Recognition

Related subjects: dictation on Mac, dictation on Windows, dictation in your apps, dictation for work and study, transcription, languages and accents, accessibility and health and voice privacy. The glossary terms dictation and voice typing are the place to start for definitions.