Speech to Text
Dictate into your microphone and watch the transcript appear live in 19 languages. Uses your browser's speech service, which in Chrome means Google's servers.
The interactive tool loads here when JavaScript runs.
About this toolWhat is Speech to Text?
Speech to Text turns what you say into editable text while you speak. Press the microphone, allow access, and words appear in the Transcript box: grey interim guesses first, then the final wording once the recognizer settles. You can edit the transcript, copy it or clear it at any time.
It uses the SpeechRecognition interface of the Web Speech API, which is available in Chrome, Edge and Safari but not in Firefox. Be clear about where the audio goes: Chrome sends your voice to Google's speech servers for recognition, Edge uses Microsoft's online service, and Safari uses Apple's speech recognition, which may run on Apple servers. This is not offline transcription.
While you dictate, the page can also keep a recording of your microphone through MediaRecorder, which you can download afterwards as WebM, Ogg or M4A depending on what your browser records.
- Live interim and final results
- 19 languages including English for the US, UK, India, Australia and other markets, Spanish, Japanese and Arabic
- Editable transcript with copy and clear
- Optional microphone recording in a format your browser supports
- No account; this site does not store your transcript
Updated
How to use Speech to Text
- Open the page in Chrome, Edge or Safari and pick your language in Settings.
- Choose a Recording Format if you want to keep the audio.
- Click the large microphone button and allow microphone access when the browser asks.
- Speak normally. Final text is added to the Transcript box; the grey "Real-time" line shows the current guess.
- Click the microphone again to stop, then use Copy, Clear or Download Recording.
Worked example
Interim versus final results
Input
Spoken: "schedule the deploy for nine thirty tomorrow"
Output
Interim guesses such as "schedule the deep" are replaced by a final result such as "schedule the deploy for 9:30 tomorrow"
The recognizer revises its guess as more audio arrives and may format times and numbers itself. Only final results are added to the transcript, so the grey text can change while you watch.
Background and insight
Chrome shipped speech recognition behind the webkitSpeechRecognition name in 2013, and the API has kept that prefix in Chrome and Safari ever since. The tool sets continuous to true, so recognition keeps going across sentences, and interimResults to true, so you see partial guesses that get replaced when a result is marked final.
Because the recognizer is a cloud service in most browsers, it behaves like one. It needs a network connection, it decides on its own when a session has gone quiet and ends it (Chrome typically stops after several seconds of silence), and its punctuation is limited. You will usually add commas and full stops by editing the transcript afterwards.
The recording is separate from recognition. MediaRecorder can only write the containers a browser supports: WebM with Opus audio in Chrome, Edge and Firefox, Ogg in Firefox, and MP4 with AAC in Safari. No browser records WAV or MP3 directly, so the format menu only lists what your browser can actually produce.
Common use cases
Drafting by voice
Talk through a first draft of an email, ticket or meeting summary faster than you can type it, then tidy it up.
Quick notes
Capture an idea hands-free and copy it into your notes app or issue tracker.
Language practice
Check whether the recognizer understands your pronunciation in Spanish, French or Japanese.
Accessibility
Give people who find typing painful a simple way to enter longer text.
Tips and limitations
- Do not dictate passwords, medical details or anything confidential; in Chrome and Edge the audio leaves your device.
- If recognition stops on its own after a pause, click the microphone again. The transcript is kept.
- Pick the exact language variant you speak. en-GB and en-US models handle accents and spellings differently.
- A headset microphone in a quiet room makes more difference than speaking slowly.
- Firefox has no SpeechRecognition support, so the page shows a notice there instead of the microphone.
Sources and further reading
- Web Speech API, W3C Audio Community Group. The specification for SpeechRecognition, continuous and interimResults
- Voice driven web apps: Introduction to the Web Speech API, Chrome for Developers. Google's 2013 announcement of webkitSpeechRecognition in Chrome
- SpeechRecognition, MDN Web Docs. Properties, events and browser support
- MediaRecorder, MDN Web Docs. The recording API and the formats it can produce
Frequently asked questions
Is browser speech to text processed locally on my computer?
No, not in most browsers. Chrome sends the audio to Google's speech servers, Edge to Microsoft's, and Safari uses Apple's speech recognition, which may also run on Apple servers. This site never receives the audio or the transcript, but the browser vendor does, so avoid dictating anything confidential.
Why does speech to text not work in Firefox?
Firefox does not ship the SpeechRecognition interface of the Web Speech API, so there is no recognizer for the page to call. The tool detects that and shows a notice. Use Chrome, Edge or Safari, or a desktop dictation feature such as Windows Voice Typing or macOS Dictation.
Why does dictation stop by itself after I pause?
The recognizer decides when a session has ended. Chrome closes it after a stretch of silence even in continuous mode, and the microphone button returns to blue. Click it again to keep going; the text you already have stays in the transcript.
Can I upload an audio file to transcribe?
No. The Web Speech API only listens to a live microphone and has no way to accept a file, so this tool does not offer upload. For recorded audio, use a transcription service or an offline model such as Whisper.
Why can I not download the recording as WAV or MP3?
Browsers record through MediaRecorder, which writes WebM or Ogg with Opus audio, or MP4 with AAC in Safari. None of them records WAV or MP3. Download the WebM or M4A file and convert it to WAV with the Audio to WAV Converter if another program needs it.
How do I add punctuation when dictating?
Edit the transcript afterwards. Chrome's recognizer adds little punctuation through the Web Speech API and does not reliably turn spoken words like "comma" into symbols. Speaking in clear sentences with short pauses helps it split results, but expect to add commas and full stops yourself.
Is there an in-depth guide to Speech to Text?
Yes. “Text to Speech and Speech Recognition in the Browser: What Really Happens” covers the background, how it works, common pitfalls and worked examples. Read the full guide.
Related Tools You May Like
Free, in-browser tools that pair well with this one.