How to use the Speech to Text
- Choose the language you will speak (it is remembered for next time).
- Click Start dictation and allow microphone access when the browser asks.
- Speak naturally. Grey italic text is the live guess; it becomes final text in the box when you pause.
- Say “comma”, “period”, “question mark” or “new line” to add punctuation (English). Edit the text directly at any time.
- Click Stop dictation, then Copy the text or Download .txt.
Dictate instead of typing
This speech to text tool turns your voice into editable text as you talk. It is useful for drafting emails and documents, capturing ideas while they are fresh, writing when typing is uncomfortable or slow, taking notes in another language, and as an accessibility aid. Most people speak at around 120–150 words per minute but type at 40, so dictating a first draft and tidying it afterwards is often much faster.
How it works — and where your audio goes
The tool uses the Web Speech API that is built into some browsers. Your browser captures the microphone and passes the audio to its own speech recognition service, which returns text. It is important to understand who that service is:
- Google Chrome sends the audio to Google’s servers for recognition. By default it is not processed on your device, and an internet connection is required.
- Microsoft Edge uses Microsoft’s online speech service.
- Safari (Mac, iPhone, iPad) uses Apple’s speech recognition, which may run on Apple’s servers or on the device depending on the device and language.
- Firefox does not currently support the API, so the tool shows a message instead.
This website never receives your audio or your transcript: there is no upload, and the text stays in the page until you copy or download it. Still, because the browser vendor processes the audio, avoid dictating passwords, card numbers, health details or other highly sensitive information.
Voice punctuation commands
Browser speech services usually return words without punctuation. With Voice punctuation on (English languages), these spoken commands are converted:
| Say | You get |
|---|---|
| comma | , |
| period / full stop | . |
| question mark | ? |
| exclamation mark / exclamation point | ! |
| colon / semicolon | : / ; |
| new line / new paragraph | line break / blank line |
Example: saying “thanks for your help comma see you tomorrow period new line best regards” produces:
With Auto-capitalize on, the first letter after a full stop, question mark, exclamation mark or line break is capitalised. Turn commands off if you need to write the literal word “period” or “comma”.
Supported languages
The list includes 37 common languages and regional variants, such as English (US, UK, India, Australia, Canada), Spanish, French, German, Portuguese, Hindi, Tamil, Telugu, Marathi, Bengali, Arabic, Chinese, Japanese and Korean. Choosing the right regional variant (for example English (India) rather than English (US)) noticeably improves accuracy for accents, names and spellings. Availability depends on your browser; if a language is not supported you will see a message.
What to expect from accuracy
Modern cloud speech engines handle clear, everyday speech well, but results depend heavily on conditions. Background noise, several people talking, strong accents the chosen language variant does not expect, technical jargon and rare names all increase errors. The live grey text is a provisional guess that often changes as the engine hears more of the sentence, so judge accuracy on the final text. Treat the result as a fast first draft: dictate freely, then spend a minute correcting it.
Tips for better accuracy
- Use a headset or a microphone close to your mouth, in a quiet room.
- Speak in full phrases at a steady pace; pauses help the service finalise each sentence.
- Recognition stops after long silences — the tool restarts it automatically while dictation is on.
- Proofread names, numbers and homophones (“their”, “there”), which all speech engines get wrong at times.
- Count or analyse the result with the word counter, or hear it read back with text to speech.
Need to transcribe an existing recording? This tool listens to a live microphone only. You can record with the voice recorder, but transcribing a file requires playing it aloud near the microphone, which reduces accuracy.
Frequently asked questions
Is this speech to text tool free?
Is my voice processed on my device?
Which browsers support speech to text?
How do I add punctuation when dictating?
Why does dictation stop by itself?
Can I transcribe an audio file?
Last reviewed November 2026. Results are estimates for information only — see our disclaimer.