Speech-to-text setup at a glance
Use your device's native dictation shortcut or keyboard microphone to enter text. A website that records audio has its own recording control and needs separate microphone permission.
| Platform | Native tool or option | How to start | Where to configure it |
|---|---|---|---|
| Windows 11 | Voice typing |
Windows + H |
Settings → System → Sound; voice typing settings |
| Windows 10 | Dictation |
Windows + H |
Settings → Privacy → Speech |
| macOS | Dictation |
Fn twice, if assigned; a Dictation key; or your configured shortcut |
System Settings → Keyboard → Dictation |
| iOS & iPadOS | Apple Dictation |
Tap the microphone on the onscreen keyboard |
Settings → General → Keyboard → Enable Dictation |
| Android | Gboard voice typing |
Tap the microphone on Gboard |
Gboard settings → Voice typing |
| Web browsers | Website voice input or a browser-based note-taking tool |
Click the website's microphone or recording button |
Website microphone permissions and browser settings |
Two distinctions matter: browser dictation has no universal activation shortcut, and installing a language pack does not automatically make speech recognition work offline.
Before You Start: Microphone, Permissions, and Language
Speech-to-text setup depends on a working microphone, permission to access it, and a supported recognition language. Check those first; offline use also requires a feature that supports local recognition.
Check that the microphone actually hears you
Record a ten-second clip in your device's recording app, then play it back. Your voice should sound clear, without heavy hiss, distortion, or long stretches of silence.
On Windows and Mac, watch the microphone's input-level meter while speaking. It should respond to your voice without staying at its maximum. There is no single correct input-volume percentage for every microphone.
Using a Bluetooth headset? Check which microphone is active. Your computer might still be listening through the laptop microphone across the desk.
Understand the permission layers
Native dictation and browser recording do not follow identical permission paths:
- Native dictation needs the relevant system feature and microphone access.
- Browser recording needs microphone access for the browser and permission for the individual website.
- Managed devices can have additional restrictions imposed by a workplace or school.
Allowing a website to record does not override a microphone block in the operating system.
Separate keyboard languages from speech models
Choose the language you will actually speak, including the regional variant where available. A keyboard layout and a speech-recognition download are not necessarily the same thing.
- Dictation
- Speech recognition that turns spoken words into text, usually in the active text field.
- Keyboard language
- The typing layout and related language features. Installing it does not prove that offline voice input is available.
- Speech model
- The component that recognizes spoken words. Some devices download local speech models separately.
Offline support: compare the actual features
Offline recognition belongs to a specific speech feature and language model. Download any required components while connected, then test the feature with network access disabled.
| Feature | Offline use | What to check |
|---|---|---|
| Windows + H | No |
Standard Windows voice typing needs internet access, even with Speech components installed. |
| Windows 11 Voice access | On supported systems |
Download a supported speech model during initial setup. |
| Mac Dictation | Hardware and language dependent |
Check Dictation settings and complete any prompted speech downloads. |
| iPhone & iPad Dictation | Device and language dependent |
Test in Notes with both Wi-Fi and mobile data turned off. |
| Android voice typing | Configuration dependent |
Check the phone, speech service, language, and selected voice-typing feature. |
How to Set Up Speech to Text on Windows
On Windows 11 and Windows 10, place the cursor in a text field and press Windows + H to start dictation. Check your microphone and privacy permissions first; standard Windows voice typing requires an internet connection.
Windows 11 setup
- Select your microphone. Open Settings → System → Sound. Under Input, select the microphone you intend to use.
- Test the input. Open the microphone's properties and use the available test controls. Speak at your normal volume. Adjust input volume if the recording is too quiet or distorted.
- Allow microphone access. Go to Settings → Privacy & security → Microphone. Turn on microphone access and the relevant app-access switches. Desktop applications have a separate access control.
- Choose your input language. Open Settings → Time & language → Language & region. Add the required language if it is missing. Use Windows + Space to switch between installed input languages.
- Click into a text field. Start with Notepad or a blank document, not a password box.
- Press Windows + H. Follow any first-use prompts, then speak.
- Configure punctuation. Open the voice typing settings from its panel. Turn on Automatic punctuation if available for your language.
Say this to test your setup"My microphone is working and this is my first dictated note."
Check the result before dictating a full page.
Windows 10: the same shortcut, different settings paths
The activation shortcut is still Windows + H, but the settings paths differ:
- Microphone permissions: Settings → Privacy → Microphone
- Online speech recognition: Settings → Privacy → Speech
- Language settings: Settings → Time & Language → Language
Windows 10 dictation uses online speech recognition. If prompted, enable it to continue.
Can Windows speech to text work offline?
Windows + H voice typing requires internet access. Downloading optional Speech components under a language's settings does not change that.
For offline dictation on supported Windows 11 systems, check Voice access under Settings → Accessibility → Speech. It is a separate feature for controlling the computer and entering text by voice. Initial setup requires downloading a supported speech model; recognition can then run on the device.
Do not confuse Voice access with the older Windows Speech Recognition feature. Their setup paths and availability differ.
How to Set Up Speech to Text on a Mac
Turn on Dictation in System Settings → Keyboard, choose a language, and check the displayed shortcut. Pressing Fn twice works only when assigned; some Apple keyboards have a dedicated Dictation key instead.
Enable and test Mac Dictation
- Check the microphone input. Open Apple menu → System Settings → Sound → Input. Select your microphone and watch the input-level indicator while speaking.
- Open Dictation settings. Go to System Settings → Keyboard and find Dictation. Older macOS releases use the name System Preferences.
- Turn Dictation on. Review any privacy or speech-processing information shown during setup.
- Set the language and microphone. Choose the recognition language and the correct microphone source where those controls are available.
- Check the shortcut. Use the shortcut displayed in settings. Assign Fn twice if that option suits your keyboard and is available.
- Test in a document. Open Notes or TextEdit, click into the document, activate Dictation, and speak.
You can also start Dictation through Edit → Start Dictation in applications that expose that menu command.
Downloading languages and using Dictation offline
On-device Dictation is available on supported Mac hardware and languages. Other configurations require an internet connection.
Let any requested speech downloads finish before testing offline. Check the information displayed in Dictation settings, then disconnect the network and try a short sentence.
App microphone permissions and Voice Control are separate
For third-party recording tools, check System Settings → Privacy & Security → Microphone. Native Dictation is not necessarily listed as a separate application there.
Voice Control is a separate option under Accessibility for users who need spoken computer controls as well as dictation. It downloads speech files during initial setup. It is not simply another shortcut for keyboard Dictation.
How to Set Up Speech to Text on iPhone and iPad
Enable Dictation under Settings → General → Keyboard, select the correct keyboard language, and tap the microphone on the onscreen keyboard. Offline recognition depends on the device and language, not just whether Dictation is switched on.
Enable keyboard Dictation
- Open Settings → General → Keyboard.
- Turn on Enable Dictation. Accept the activation prompt if one appears.
- Add your language. Open Keyboards → Add New Keyboard if the required language is missing. Check Dictation Languages too, where that option appears.
- Open Notes or another text app. Tap the text field to display the keyboard.
- Switch languages if necessary. Use the globe or keyboard-switching control.
- Tap the keyboard microphone and speak. Tap the available stop control when finished.
Use the keyboard microphone, not the voice-message button. The microphone inside a messaging app can record an audio message instead of entering text.
Check microphone access and restrictions
Record a short clip in Voice Memos if Dictation hears nothing. A poor recording points to an audio-input problem, not a punctuation setting.
For third-party recording apps, check Settings → Privacy & Security → Microphone. If Enable Dictation is missing or unavailable, inspect Screen Time restrictions and any work or school management policies.
Can iPhone Dictation work offline?
Supported iPhones and languages offer on-device Dictation. There is no universal "download offline dictation" button across all iOS versions.
Add the required language, complete any prompted downloads, and test with both Wi-Fi and mobile data turned off. A successful offline test in Notes confirms that your current setup works locally; it does not prove that every voice feature on the phone works offline.
How to Set Up Speech to Text on Android
Enable Voice typing in Gboard, allow microphone access, and tap its microphone while a text field is active. Offline recognition depends on the phone, installed speech services, language, and selected voice-typing feature.
Enable Gboard voice typing
- Install or update Gboard. Use Google Play if Gboard is not already installed.
- Make Gboard available. On Pixel-style Android menus, look under Settings → System → Keyboard → On-screen keyboard. Samsung devices commonly place keyboard controls under Settings → General management → Keyboard list and default.
- Open Gboard settings. Use the gear icon on the keyboard or open Gboard through system settings.
- Select Voice typing and turn it on.
- Choose your language. Add the language under Gboard's language settings, then select the appropriate keyboard while typing.
- Grant microphone access. Accept the permission prompt. If access was previously denied, check Settings → Apps → Gboard → Permissions. Some configurations also depend on the Google app's microphone permission.
- Tap the Gboard microphone. Speak into the phone and check the words appearing in the active field.
Android menu names vary by manufacturer and release. Search Settings for "keyboard" or "microphone" if your menus differ.
Check the system microphone switch
Recent Android versions include a system-wide Microphone access control, often available in Quick Settings or privacy settings. If it is off, an app-level permission alone will not fix voice typing.
Use the phone's Recorder app to test audio input. Disconnect a Bluetooth accessory if Android is listening through the wrong microphone.
Download offline speech languages where available
Some Google speech configurations expose Offline speech recognition through the Google app's Settings → Voice menu. If that menu is present, download the required language while connected.
Other devices manage speech models through their voice-typing components without exposing the same menu. Follow the download prompts available on your device.
A downloaded Gboard keyboard language is not proof that offline speech recognition is installed. Test the actual voice-typing feature with network access disabled.
How to Set Up Speech to Text in a Web Browser
Use either your operating system's dictation feature inside a text field or a website that records your microphone. Website recording needs browser and site permissions; there is no universal browser shortcut, and support differs between browsers and speech services.
Option 1: Use native dictation inside a website
Click into a normal text field, then activate your device's dictation:
- Windows: Windows + H
- Mac: your configured Dictation shortcut
- iPhone or Android: the keyboard microphone
Here, the operating system supplies the text. The website does not necessarily need microphone permission because it is not recording audio itself.
Option 2: Use a browser-based voice tool
- Open a trusted HTTPS website that offers recording or dictation.
- Click its microphone or recording control.
- Allow microphone access when the browser asks.
- Choose the correct microphone if a device selector appears.
- Check system permissions if recording remains blocked.
- Record a short test, then confirm that the transcript appears.
For an existing document workflow, Google Docs offers Tools → Voice typing in supported desktop browsers. Chrome is a straightforward starting point.
If you previously blocked a website, reopen its site-permission controls near the address bar, change the microphone setting, and reload the page. Browser and website support still matter; permission alone cannot add a missing speech feature.
Essential Voice Commands for Dictation
Common dictation commands insert punctuation and paragraph breaks while you speak. Their exact wording depends on the platform, language, and recognition mode, so test each command before relying on it in a longer document.
| Say this | Intended result |
|---|---|
| "Period" | Inserts . |
| "Full stop" | Alternative for . in supported English variants |
| "Comma" | Inserts , |
| "Question mark" | Inserts ? |
| "Exclamation point" or "exclamation mark" | Inserts !, depending on the system |
| "Colon" | Inserts : |
| "New line" | Starts another line |
| "New paragraph" | Starts another paragraph |
| "Open quote" / "close quote" | Inserts quotation marks where supported |
What you say
"Please send the report comma then call me period new paragraph Thank you period"
Result on a compatible system
Please send the report, then call me.
Thank you.
Automatic punctuation is not the same as voice editing
Some dictation tools insert punctuation automatically. Windows 11, recent Apple devices, and certain Android configurations offer automatic punctuation, with availability depending on language and settings.
Editing commands are less consistent. Do not assume that "delete that" or "select the previous sentence" works everywhere. Full voice-control systems and basic keyboard dictation have different capabilities.
Where Built-In Dictation Becomes Frustrating
Built-in dictation handles short text entry, but most native tools type into the active field without creating a separate saved session. Whether your work survives a crash depends on the destination application, not the dictation shortcut.
Useful for quick text
- Short messages and everyday notes
- Entering text directly into an app
- Voice input generally included with the device
Limits to plan around
- Commands differ across platforms
- Words follow the active cursor
- Unsaved text has no independent dictation archive
Settings are scattered
A single dictation problem can involve the input device, system privacy controls, keyboard language, online recognition settings, and app permissions. That explains a familiar annoyance: the microphone works in a video call but not in a note-taking tool. Each app can have different access, and each service can select a different input.
Permission bugs and stale permission states add another wrinkle. After changing access, you sometimes need to close and reopen the affected application.
Spoken commands interrupt natural speech
Saying "comma" and "new paragraph" is workable for a short message. During brainstorming, it breaks your train of thought.
Automatic punctuation helps, but it is not universal. Nor does it always capture the structure you intended. A long pause can become a period when you wanted a comma.
The cursor controls where your words go
Native dictation generally types wherever the cursor sits. Switch windows accidentally, and your next sentence can land in a search box, chat field, or the wrong document.
Watch the destination. Especially before dictating sensitive information.
Unsaved text can disappear
Native keyboard dictation is not usually a recording archive or transcript-history system. If a browser tab crashes or an application closes before saving, dictated text in an unsaved field can disappear. The dictation tool generally has no independent session you can reopen.
This is not inevitable in every app. A document editor with autosave, recovery, or version history provides its own protection. But that protection belongs to the editor, not to the dictation feature.
For longer notes, the saving workflow matters as much as recognition quality.
Use Browser-Based AI for Natural Speech and Saved Notes
A browser-based AI note taker can combine microphone capture, transcription, punctuation, and saved-note storage without native OS dictation setup. Microphone permissions still apply, and saved history protects completed notes, not audio that never reached the service.
AI Voice Note Taker: speak naturally without installing a dictation app
The SpeechText.AI AI Voice Note Taker is a practical option for capturing thoughts without working through platform-specific dictation menus.
Open the tool in a web browser and grant microphone access. There is no native dictation feature to activate and no keyboard shortcut to memorize. Browser recording support and access permissions still determine whether a particular device can record.
Automatic punctuation
Intelligent punctuation is applied by default. Speak naturally instead of announcing every comma and period.
No separate dictation app
Start in the browser without installing a dictation application or configuring OS speech language packs.
Saved transcript history
Transcripts are automatically organized so you can return to saved notes rather than a temporary text field.
Think "zero installation," not "zero permissions." If the operating system blocks the browser's microphone, that permission still needs fixing.
File history also is not a promise that unsaved speech can never be lost. Interrupted uploads, network failures, and closing a session too early remain risks. Before leaving an important note, check that its transcript appears in history.
SpeechText.AI: use file transcription for existing recordings
Already have an interview, meeting, lecture, or recorded consultation? Uploading the file is the appropriate workflow. Playing it through a speaker into a dictation microphone adds noise and gives you less control.
SpeechText.AI provides professional audio and video transcription with domain-specific AI models, including medical, legal, and technology-focused recognition. These models address specialized vocabulary that general-purpose dictation often struggles with.
Why separate audio channels matter for transcription
SpeechText.AI's multi-channel audio processing is relevant to recordings captured on separate channels. Preserving separate speaker audio gives the transcription workflow information that disappears when everything is mixed together.
Multiple speakers on one channel are not the same thing as a multi-channel recording.
Clean audio still matters. Review names, numbers, technical terms, and overlapping speech before publishing a transcript or using it in a high-stakes record. A domain-specific model supports accuracy; it does not replace human review.
