Video & Audio
Speech to Text (Voice Typing)
Dictate with your microphone and see the text appear live — in dozens of languages — then copy or download it. Works in Chrome, Edge and Safari.
How to use
- 1Pick your language and press Start dictation.
- 2Speak naturally — say 'period' or 'comma' for punctuation in many languages. Copy or download the text.
Frequently asked questions
Which browsers support it?
Chrome, Edge and Safari support speech recognition. Firefox doesn't yet.
Is my voice sent anywhere?
Speech recognition is provided by your browser — Chrome and Edge send audio to Google or Microsoft for processing. Safari may process on-device.
Can it transcribe an audio file?
No — it listens to your microphone live. Play a file near the mic as a workaround, or use a dedicated transcription service.
Why this is the one tool in this category that isn't fully private — worth reading before you dictate anything sensitive
This tool is a thin wrapper around the browser's built-in Web Speech API, not a self-contained speech model — and critically, that API's actual recognition engine runs server-side for most browsers: Chrome and Edge send your captured audio to Google's or Microsoft's speech recognition servers for processing, and only the resulting text comes back to this page. Safari's behavior may differ and lean more on-device. That's a meaningfully different privacy model from the rest of this tool category, where processing genuinely happens entirely on your device — dictating sensitive or confidential content here means it leaves your machine, even though this specific page never stores or uploads anything itself.
Why dictation doesn't stop after a few seconds of silence, even though the underlying API does
Browser speech recognition engines automatically end a recognition session after a pause — this is a built-in behavior of the API itself, not something this tool controls directly. Rather than making you repeatedly hit Start, the code tracks a separate 'wanted' flag representing whether you've asked to keep dictating, and in the recognition engine's onend handler, it immediately calls .start() again if that flag is still true — creating the experience of one continuous dictation session stitched together from multiple short-lived recognition sessions under the hood. Only clicking Stop sets the flag false and lets a pause actually end the session for good.
Why interim and final results are tracked and displayed differently
The API streams back two kinds of results as you speak: 'interim' results, which are the engine's current best guess that can still change as it hears more context, and 'final' results, which are locked in and won't be revised. The tool keeps these separate — final text accumulates permanently into the saved transcript, while interim text is shown appended temporarily (and replaced each update) until it resolves to final — which is why you'll sometimes see a word flicker or change slightly as you keep talking, before it settles into the permanent transcript once the engine is confident enough to finalize it.
Australian English opens on US English, unless the match is exact
The language menu is a fixed list of 21 languages, not every dialect a browser might be set to. It opens on your browser's language only when that language is in the list exactly. Otherwise it opens on English (US). A browser set to Australian, Canadian, or Indian English therefore starts on English (US). British English is in the list, but it is not used as a stand-in for those other English settings. You cannot change language while Listening is showing. Stop first, pick the language, then start again, and that new session uses the language selected at that moment. Closing the page stops dictation, so a session is not left running after you leave.
A new phrase gets a space only if your text does not already end on one
Each finished phrase is added to the end of what you already have. A space is placed in front of it only when your text does not already end with a space or a line break, and a space at the start of the new phrase is removed so you do not get a double gap. A guess that never settles is dropped when recognition pauses, including the automatic restart after a quiet moment, so a half-formed word does not stick. Silence does not show an error and does not stop you. Blocking the microphone does: you see Microphone access was blocked, and dictation ends. Any other failure shows Recognition error and the reason. The word count includes only text that has settled, not words still being guessed.
Related tools
- Text to Speech (Read Aloud)
Have any text read aloud with natural voices — choose the voice, language, speed and pitch. Free, private and runs in your browser.
- Online Voice Recorder
Record your voice or any microphone in the browser — pause, play back and download the recording. No sign-up and nothing is uploaded.
- Online Screen Recorder
Record your screen, a window or a browser tab — with optional microphone — right in your browser, then download the video. No install, no watermark.
- Drum Machine & Beat Maker
Make beats in your browser — a 16-step drum sequencer with kick, snare, hi-hats, clap and more, preset patterns, swing and tempo — and export your loop as WAV/MP3.
- Online Tuner (Guitar, Ukulele, Bass, Chromatic)
Tune your guitar, ukulele, bass or any instrument with your microphone — see the note, the frequency and how many cents sharp or flat you are.
- Online Notepad
A fast, distraction-free online notepad that saves automatically in your browser — multiple notes, search, word count, copy and .txt download. No account, nothing uploaded.