Skip to content

Speech to Text (Private Dictation)

Dictate with speech recognition that runs on your own computer. Say “comma”, “period” or “new paragraph” for punctuation, edit as you go, then copy or download your text. Your voice is never sent to a server.

Checking on-device speech recognition…

Recognition runs on your device only. Audio is never recorded, stored or uploaded.

Hearing:
0 words · 0 characters

Your transcript is kept in this browser only, so a reload doesn’t lose it. Clear removes it.

Voice commands (English)
  • comma → ,   period or full stop → .   question mark → ?   exclamation mark → !
  • colon → :   semicolon → ;   open quote / close quote → “ ”   open parenthesis / close parenthesis → ( )
  • new line → line break   new paragraph → blank line

Related Tools

Text-to-Speech Studio Word Counter Grammar & Spell Checker Private Text Translator Paraphrasing Tool Speaking Time Calculator

How to Use

  1. Choose your language — Pick the language and accent you will speak. The status box checks whether your browser can recognise it on your device. If a language pack is needed, click the download button once; after that it works offline.
  2. Start dictation — Click “Start dictation”. Your browser asks for microphone permission the first time. Speak naturally in phrases; words appear in the preview as you talk and are added to the transcript when each phrase is finished.
  3. Say punctuation and line breaks — For English, say “comma”, “period”, “question mark”, “new line” or “new paragraph” and the tool types the symbol or break instead of the word. Turn voice commands off if you need to dictate those words literally.
  4. Edit, copy or download — Click “Stop” when you are done. The transcript is an ordinary text box: fix any words by typing, place the cursor where new dictation should go, then copy the text or download it as a .txt file.

About Speech to Text (Private Dictation)

Most dictation websites stream your microphone audio to a cloud speech service. Even Chrome’s long-standing speech recognition for web pages works that way by default. WritePadPro’s Speech to Text tool only uses on-device recognition: it tells the browser that recognition must happen locally, checks that the language is installed on your computer, and refuses to start otherwise. Your voice is turned into text on your own device, and the page never records or uploads audio.

How it works

The tool uses the Web Speech API’s SpeechRecognition interface with three newer features that Chrome added for desktop in version 139:

Recognition runs continuously with interim results: you see words in the preview line as you speak, and each phrase is added to the transcript when the recognizer marks it final. Browsers end a session after silence, so while you are dictating the tool starts a new session automatically. Your microphone is used only between Start and Stop, and the browser shows its usual recording indicator.

Voice punctuation commands

Many recognizers do not add punctuation, so the tool understands spoken commands when an English language is selected:

SayYou get
comma,
period / full stop.
question mark?
exclamation mark / exclamation point!
colon / semicolon: / ;
new linea line break
new paragrapha blank line
open quote / close quote“ / ”
open parenthesis / close parenthesis( / )

For example, saying “thanks for the update comma I will reply tomorrow period new paragraph best wishes” produces “Thanks for the update, I will reply tomorrow.” followed by a blank line and “Best wishes”. Punctuation attaches to the previous word, an opening quote or parenthesis attaches to the next one, and the first word of each sentence is capitalised when “Auto-capitalize sentences” is on. If you need the literal words (for example when dictating “a comma splice”), turn voice commands off for a moment. For Chinese, Japanese and Thai, phrases are joined without spaces.

Editing while you dictate

The transcript is a normal text box. You can correct words at any time, and new phrases are inserted wherever your cursor is, with spacing adjusted around them. The word and character counts update as you go, and a timer shows how long you have been dictating. The text is kept in this browser’s local storage so that a reload does not lose it, and Clear removes it.

Who uses dictation

Most people speak far faster than they type, which makes dictation useful for first drafts, meeting notes, emails, journaling and brainstorming. It also helps people with repetitive strain injury, dyslexia or limited mobility. Because recognition happens on the device, it is suitable for confidential material such as patient notes, legal drafts and personal diaries, and it keeps working without an internet connection once the language pack is installed.

Requirements and limitations

You need Chrome 139 or later on Windows, macOS or Linux, a working microphone, and the language pack for your language (downloaded once). Firefox does not provide speech recognition to web pages, and other browsers’ recognition may rely on servers, so the tool shows an explanation there instead. Accuracy depends on your microphone, background noise and accent; proper names and specialist terms often need a quick manual correction.

Frequently Asked Questions

Is my voice sent to a server?

No. This tool only starts speech recognition with on-device processing required (the Web Speech API’s processLocally option), so the browser must recognise your speech on your computer or refuse. If on-device recognition is not available, the tool explains why instead of using a cloud service. The page never records or stores audio; only the recognised text is kept in the text box.

Which browsers and devices does it work on?

On-device speech recognition for web pages arrived in Chrome 139 on desktop computers (Windows, macOS and Linux). Firefox does not offer speech recognition to web pages, and the speech recognition in other browsers may send audio to a server, which this tool will not use. On a phone, the microphone key on your keyboard is usually the simplest way to dictate; whether it runs on the device depends on your phone and its settings.

Why do I need to download a language pack?

On-device recognition needs a speech model for each language. Chrome downloads it once when you click the download button and then keeps it, so later dictation starts quickly and works offline. The status box tells you whether the selected language is ready, needs a download, or is not supported on this device.

Which voice commands can I use?

With voice commands on and an English language selected: “comma”, “period” or “full stop”, “question mark”, “exclamation mark” or “exclamation point”, “colon”, “semicolon”, “new line”, “new paragraph”, “open quote” and “close quote”, and “open parenthesis” and “close parenthesis”. Punctuation attaches to the previous word, and the next word after a full stop, question mark, exclamation mark or line break is capitalised.

Why does dictation pause or stop by itself?

Browsers end a recognition session after a stretch of silence or after a while of continuous use. While you are dictating, the tool restarts recognition automatically, so you can pause to think. If it keeps stopping immediately, for example because the microphone was disconnected or permission was withdrawn, it stops and tells you why. Click Stop when you are finished.

Is my transcript saved?

The transcript is kept in this browser’s local storage on your device, so a page reload or accidental tab close does not lose your dictation. It is never uploaded. Clear removes it (and Undo brings it back if you clicked by mistake).

How can I get more accurate results?

Use a headset or a microphone close to your mouth, reduce background noise, and speak at a steady pace in complete phrases rather than word by word. Choose the language variant that matches your accent, such as English (India) or English (United Kingdom). Names and technical terms are the most common errors, so read the transcript once and correct them by typing.

Can I dictate into the middle of my text?

Yes. Click in the transcript where you want new words to go, then keep dictating. New phrases are inserted at the cursor with sensible spacing, and punctuation commands attach to the word before them.