Speech to Text
Live recognition — meeting notes made easy
⚠️ Audio is processed by your browser vendor (Chrome uses Google). For confidential content, use a desktop solution.
Quick guide
What can Speech to Text help you do?
Free speech-to-text using browser Web Speech API. Live transcription in English, Chinese, Japanese, Korean and more.
Best for
Dictate a short, nonconfidential draft when typing is inconvenient. Recognition Language offers Chinese for Taiwan or China, US or UK English, Japanese and Korean. This is live microphone transcription, not an audio-file uploader, speaker-identification system or a way to capture every participant in an online meeting.
How to use it
Choose Recognition Language before starting, press Start Recording and respond to the microphone permission prompt. Speak one short sentence and check the text before continuing. Stop ends listening; Clear also stops recognition and removes the displayed draft. You can edit the text, copy it or download transcript.txt.
How to read the result
Text in square brackets is an interim recognition result and may change. Check names, amounts and punctuation against what was said rather than assuming the draft is exact. If the service disconnects or permission is refused, use the displayed error to decide whether to retry; this tool has no measured accuracy guarantee.
Privacy and limits
After you start and allow microphone access, Web Speech API may send audio to a browser-provided server. This tool does not force on-device recognition. Do not dictate passwords, personal records, or confidential meetings. Downloaded transcript.txt files remain on your device; clearing the text does not delete any external service records.
Browser API availability does not guarantee a working recognition service or offline support. Test your language and microphone first. Treat the transcript as an editable draft, not a certified record, and obtain permission before transcribing another person.
Frequently Asked Questions
How accurate?
There is no measured accuracy guarantee for this tool. Language, accent, microphone quality and background noise affect recognition. Dictate a short sample, then check names, numbers and punctuation against what you said.
Which browsers work?
Support depends on your browser, operating system and available recognition service. The page checks for Web Speech API, but detecting the API does not guarantee the service will connect. Test a short phrase before relying on it.
Is audio uploaded?
Audio is sent to your browser vendor's cloud (Chrome → Google) for recognition. For sensitive content, use local tools like Whisper.