Speech to Text Dictation
Dictate text using your microphone — supports English, Hindi, Tamil, Telugu, Bengali, and more.
Written by Anil Gautam · Updated
Which languages can I dictate in?
The language menu has twelve options: English (US), English (India) and English (UK), then Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam and Punjabi, each with its India locale code such as hi-IN or ta-IN. English (US) is selected when the page loads. Indian languages come back in their own script, so Hindi appears in Devanagari and Tamil in Tamil script. If you speak English with an Indian accent, try English (India). Pick the language before pressing Start listening; to switch midway, stop, change the menu and start again.
How does the transcript build up?
Recognition runs continuously and shows interim results. Words the engine has not settled yet appear in grey italics at the end of the box; once it settles a phrase, that phrase turns black and gets a space after it. Copy takes only the black text, so wait for the grey part to settle before copying. The transcript area cannot be edited, so fix mistakes after pasting the text elsewhere, and Clear empties it.
Browsers end a recognition session on their own from time to time, for example after a stretch of silence or a network error. When that happens the button switches back to Start listening and your text stays. Press it again and new words are added to the end: if the first session finished with "send the invoice" and the second adds "by Friday", the box holds "send the invoice by Friday". Nothing is saved between visits, so copy the text before you reload or leave.
Why does listening stop as soon as I start?
If the button flips back to Start listening straight away, the usual cause is a blocked microphone. The tool shows no error message in that case, so check the microphone permission for the site in your browser's address bar and try again. It uses whichever microphone the browser is set to, and you can switch devices in the site settings or your system sound settings. A notice saying speech recognition isn't supported means the browser lacks the Web Speech API; Chrome, Edge and Safari have it.
The page has no speech engine of its own. It hands your audio to the browser's recognition service, which is why results can differ between browsers. Paste the finished text into Word Counter to count it, or use Text to Speech to hear text read back.