Speech to text

Speak, and watch your words turn into text as you talk.

How to use speech to text

  1. Pick your language and press start listening.
  2. Allow microphone access when your browser asks.
  3. Speak, then copy or download the transcript when you're done.

Browser support

BrowserWorks?Where recognition happens
ChromeYesGoogle's speech servers, needs an internet connection
EdgeYesMicrosoft's speech servers, needs an internet connection
Safari (recent macOS and iOS)YesOn the device, works without internet
FirefoxNoDoes not implement this browser feature

The 19 supported languages

  • English (United States, United Kingdom), Spanish (Spain, Mexico), Portuguese (Brazil, Portugal)
  • French, German, Italian, Dutch, Polish, Turkish
  • Russian, Ukrainian, Arabic, Hindi, Indonesian, Vietnamese, Thai
  • Japanese, Korean, Chinese

Tips for a more accurate transcript

  • Speak in a quiet room. Background noise and other voices are the most common cause of a garbled transcript.
  • Use a headset or a phone's built-in mic rather than a laptop's far-field mic when possible.
  • Speak in full sentences at a normal pace. Very fast or very choppy speech confuses the recognizer.
  • Review the finished transcript for punctuation and capitalization, since neither is added reliably.

Speech to text FAQ

Is my audio uploaded anywhere?

In Chrome and Edge, yes: your audio is sent to that browser's speech recognition servers to be transcribed, since this tool uses the browser's own built-in speech engine rather than running its own recognition. Safari performs recognition on the device on newer versions of macOS and iOS.

Why doesn't this work in Firefox?

Firefox does not support the Web Speech API's recognition feature that this tool relies on. Use the latest Chrome, Edge or Safari instead.

Why did it stop listening on its own?

Some browsers pause recognition after a few seconds of silence. This page automatically restarts listening in the background when that happens, so you rarely need to press start again mid-sentence.

Can I dictate in a language other than English?

Yes, pick from 19 languages including Spanish, French, German, Japanese and Mandarin before you start. Changing the language while listening stops the current session.

Why does the last phrase sometimes change right after I say it?

While you are still speaking, the box shows an interim result, the browser's best guess so far. Once the browser is confident in that phrase, it locks it in, and the wording can shift slightly at that point.

Does it add punctuation automatically?

Not reliably. Recognition results are not guaranteed to include commas or periods, so read back the transcript afterward and add punctuation yourself, or clean it up with the find and replace tool for repeated fixes.

Can I dictate a long document in one sitting?

Yes. Continuous listening restarts itself automatically if the browser briefly pauses after silence, so you can keep dictating for as long as you want with only short breaks between sentences.

Can I edit the transcript after I stop listening?

Yes, the text box is a normal editable field, so you can click into it and fix any misheard words before copying or downloading the result.

Will it pick up audio playing from my speakers instead of my voice?

It listens through your microphone, so audio playing from your own speakers can be picked up too if the mic hears it, but speaking directly into the mic gives far more accurate results than transcribing playback.

Is there a time limit on a single listening session?

No fixed limit in this tool, though your browser may eventually pause recognition on its own after a very long stretch of continuous listening, which this page automatically restarts.

Related tools