Speech to Text
Speak and watch your words appear: choose a language, press Start dictation, and the browser's speech recognition types what you say into an editable transcript, with live previews as you talk. Say "comma" or "new line" for punctuation in English, then copy or download the text. Your audio goes to your browser's speech service, not to CitationEasy — the details are below.
Checking whether this browser supports speech recognition…
Not listening
Where your voice goes
Speech is turned into text by your browser's speech service, not by CitationEasy. In Google Chrome, the audio is sent to Google's servers to be transcribed; Microsoft Edge sends it to Microsoft's online speech service; Safari uses Apple's speech recognition, which can also process audio on Apple's servers. CitationEasy never receives your audio or your transcript, and the text stays in this page until you copy or download it. Don't dictate anything you wouldn't want those services to process.
How to Use Speech to Text
- 1. Choose the language you'll speak. Recognition works far better when it knows the language — and the accent: pick English (United Kingdom) or English (India) rather than US English if that's how you speak.
- 2. Press Start dictation and allow the microphone. The browser asks once per site. A red dot shows when it is listening.
- 3. Speak in phrases. Grey text under the transcript is the recogniser's running guess; when you pause, the finished phrase is added to the transcript. With Keep listening through pauses on, dictation carries on after a silence instead of stopping.
- 4. Edit, then copy or download. The transcript is an ordinary text box: fix names and mistakes as you go, then copy the text or save it as a .txt file.
How Browser Dictation Works
This page uses the speech recognition part of the Web Speech API, which browsers provide to websites. The page hands the browser your chosen language, the browser captures your microphone and sends the audio to its own recognition service, and the service streams back words: first provisional guesses, then a final version of each phrase. The quality, the languages and where your audio goes all depend on the browser, not on this page.
| Browser | Works here? | Who turns speech into text |
|---|---|---|
| Google Chrome (computer, Android) | Yes | Google's speech servers |
| Microsoft Edge | Yes | Microsoft's online speech service |
| Safari (Mac, iPhone, iPad) | Yes | Apple's speech recognition, which can use Apple's servers |
| Firefox | No | — |
| Some other Chromium browsers, such as Brave | Often not | The API is present but may have no service behind it |
What happens to your voice
CitationEasy never receives your audio or your transcript. But the audio does leave your device: in Chrome it is sent to Google to be transcribed, in Edge to Microsoft, and Safari may send it to Apple. The transcript itself stays in this page until you copy or download it, and nothing is saved when you close it. Don't dictate passwords, medical details or anything else you wouldn't want those services to process.
Spoken Punctuation
Recognition services return words, so saying "comma" usually gives you the word comma. With spoken punctuation on and an English language selected, the tool replaces a small set of spoken commands with the marks they name, removes the space before each mark, and capitalises the next sentence.
| Say | You get |
|---|---|
| comma | , |
| period, or full stop | . |
| question mark | ? |
| exclamation mark, or exclamation point | ! |
| colon / semicolon | : / ; |
| new line | a line break |
| new paragraph | a blank line |
Worked example
You say
dear professor comma new line could I have an extension on the essay question mark thank you period
Transcript
Dear professor, Could I have an extension on the essay? Thank you.
The commands are deliberately literal: "the Victorian period" becomes "the Victorian." too. Switch spoken punctuation off for text like that, or fix it in the transcript. Some recognition services add punctuation of their own; if yours does, turn the option off to avoid doubled marks.
Frequently Asked Questions
- Which browsers support speech to text?
- Google Chrome and Microsoft Edge on computers and Android, and Safari on Mac, iPhone and iPad. Firefox does not offer speech recognition to websites, and some Chromium-based browsers, such as Brave, include the feature but cannot reach a recognition service, which shows up as a network error.
- Is my audio processed on my device?
- Usually not. In Chrome the audio is sent to Google's servers to be turned into text, Edge uses Microsoft's online speech service, and Safari uses Apple's recognition, which can involve Apple's servers. CitationEasy itself never receives the audio or the transcript, and nothing is stored after you close the page.
- How do I add punctuation when dictating?
- Select an English language, keep spoken punctuation on, and say the mark's name: "comma", "period" or "full stop", "question mark", "exclamation mark", "colon", "semicolon", "new line" or "new paragraph". The tool converts them as each phrase is finished. In other languages, add punctuation by editing the transcript.
- Why does dictation stop on its own?
- Browsers end a recognition session after a pause or after a while of continuous use. With Keep listening through pauses on, the tool starts a new session straight away, so you can keep talking; after several silent restarts in a row it stops and tells you no speech was heard. Microphone and permission problems also stop it, with a message saying what went wrong.
- Can I transcribe an audio file or a recorded lecture?
- Not with this tool. It listens to your microphone live; it cannot open audio or video files, and playing a recording into the microphone usually transcribes poorly. For recorded lectures and interviews, use a transcription service designed for files, and check its privacy terms first.
- How accurate is browser speech recognition?
- Accurate enough for drafts and notes in a quiet room with a decent microphone, and noticeably worse with background noise, fast speech, technical terms and names. Always read the transcript through before you use it; a grammar checker catches some of the slips, but not wrong words that are spelled correctly.