Typing is the slow part of writing for a lot of people: the ideas arrive faster than the fingers, or a sore wrist, a phone keyboard and a long email all conspire against getting words down. Our free voice dictation tool turns speech into live text in your browser — no app to install, no account, no signup — and the transcript stays editable, ready to copy to your clipboard or download as a .txt file. This guide explains how to use it, the spoken punctuation commands that turn a wall of dictated words into usable text, the microphone habits that decide accuracy more than any setting, and the honest limits: which browsers it works in, which languages it covers, and what happens to your voice along the way.
Step one: open the tool in a supported browser and allow the microphone. Press Start dictation and grant microphone access when your browser asks — without that permission there is nothing to transcribe, and the permission prompt is the browser's, not the site's. Pick the recognition language that matches how you speak before you begin; the dropdown covers English (US), English (UK), Spanish (Spain), French, German, Hindi, Urdu and Arabic. Step two: speak naturally and watch the transcript build. Your words appear live — grey italic text is the browser's live guess, and confirmed words land in the transcript as solid text. Listening restarts automatically if the browser pauses after a silence, so a long pause does not end your session. Step three: use your notes. The transcript is an editable text area, so you can fix a word while the mic is still hot, without stopping. When you are done, press Stop, then copy the text or download it as a .txt file — or paste it into the text-to-speech reader to hear it read back, which is a surprisingly good proofreading pass: errors you skim past on screen jump out when spoken aloud.
Raw dictation without punctuation is a run-on paragraph nobody wants to read, and this is where most people give up on voice typing. The fix is to say the punctuation out loud as you speak: say "period", "comma", "question mark", "exclamation point", "colon", "open quote" and "close quote", and the recognizer converts them into the marks themselves. "New paragraph" starts a fresh paragraph, which matters more than it sounds — dictated notes without paragraph breaks are genuinely hard to reuse. Try it on a short test first: dictate two sentences with a comma, a question and a new paragraph, then read the result back. Once that works, it becomes automatic, and your dictated drafts start arriving in a shape you can actually send. Two cautions from the tool page itself. First, punctuation commands work best with English recognition; for Hindi, Urdu and Arabic, results vary, so test a short sentence before dictating something long. Second, common words in your own language are recognised far better than mixed-language sentences — pick one language in the dropdown and stick to it for the session rather than switching mid-sentence.
Accuracy starts with the microphone, and this is the single most useful fact in this guide. Laptop and phone built-in microphones pick up room echo and keyboard noise, so a cheap headset or a pair of earbuds with a mic will immediately outperform them. Sit in a quiet room, keep the mic a few inches from your mouth and off to the side so your breath does not blast it, and speak at a steady, conversational pace — rushing makes words blur together, and long pauses make the browser think you are finished. If accuracy is poor, the first thing to change is always the mic, not the settings. Next, choose the exact language variant you speak: en-US versus en-GB makes a real difference in vocabulary, and the same applies to the other languages on offer. Dictate in full sentences rather than fragments, because the recognizer uses context to disambiguate words that sound alike, and correct the text live as you go — the textarea stays editable while the mic is hot, so fixing a word costs seconds, while fixing a whole page at the end costs far longer. If you see the live grey text stuck without confirming, pause for a second and continue; the engine catches up. Long sessions work fine, because the tool automatically restarts the recogniser when the browser pauses it after silence.
Picture a realistic use: a short work email you would otherwise type on a phone keyboard. You open the tool in Chrome, pick English (US), put in earbuds, and press Start dictation. You say: "Hi Sara comma thanks for sending the report period I have reviewed the first section and I have two questions question mark new paragraph First comma could we move the deadline to Friday question mark Second comma is the budget figure final question mark Thanks period." What lands in the transcript is a short, punctuated email with two paragraphs — not a perfect draft, but one that needs a thirty-second tidy rather than a rewrite. That is the right expectation for dictation generally: it replaces the typing, not the editing. Read it once, fix the two or three words the recognizer guessed wrong, and send it. For longer pieces — meeting notes, a journal entry, a first draft of an article — the same rhythm holds: dictate in chunks, correct live, and keep the mic close. People who try to dictate a flawless final draft in one breath are disappointed; people who treat dictation as fast first-draft capture, with a short edit afterwards, keep using it every day.
Dictating in the wrong browser and blaming the tool. Firefox does not currently support the Web Speech API this tool needs, so the page shows a clear notice there — open it in Chrome, Edge or Safari instead. Skipping punctuation commands. A transcript without periods and paragraphs is not a document; say the marks as you speak. Dictating far from a built-in mic in a noisy room. Distance and echo cost more accuracy than any setting can recover; a headset in a quiet room is the fix. Mixing languages mid-sentence. Pick one recognition language and stick to it for the session. Waiting until the end to correct. The transcript is editable while the mic is hot — fix words as they land, not after a page of them. Pasting sensitive notes into a chatbot instead. ToolNest itself never receives, stores or uploads your audio or transcript; a chatbot, by contrast, receives everything you paste by design. For private material, that distinction is the whole point of dictating here.
The honest limits are four, and the tool page states them plainly. First, browser support: dictation needs the Web Speech API, available in Google Chrome on desktop and Android, Microsoft Edge on desktop, and Safari on macOS and iOS from version 14.1 — Firefox is not supported. Second, privacy has a nuance worth understanding: transcription is done by your browser's built-in speech service, which may process audio on your device or through the browser vendor's servers depending on your platform. ToolNest itself never receives, stores or uploads your audio or transcript — everything you dictate stays in this page as far as ToolNest is concerned — but the browser's own service is part of the chain, so for highly sensitive material, check your browser's speech settings too. Third, punctuation commands and accuracy vary by language: English recognition is the most reliable, and Hindi, Urdu and Arabic results vary, so test short first. Fourth, dictation is a capture tool, not an editing tool: it will not fix your grammar, structure your argument or choose your words. Within those limits, it removes the slowest part of writing — the typing — and hands you an editable transcript in seconds. The companion guides are our text-to-speech guide, for hearing a dictated draft read back as a proofreading pass, our character counting guide, for fitting a dictated post to a platform limit, and our word count guide, for the essay and application limits dictated drafts most often have to meet.
Hear a dictated draft read back — errors you skim past jump out when spoken aloud.
Fit a dictated post to a platform limit when every character is audited.
The essay and application limits dictated drafts most often have to meet.