Voice typing used to mean a dedicated app, a paid transcription service, or a virtual assistant tied to one platform. A capable version of the same thing now runs directly inside some browsers, for free, with no account and no audio ever leaving your device.
What's actually happening under the hood
This is built on the Web Speech API's recognition feature — a browser-native capability, not a library or a service our site calls out to. When you start it, the browser itself listens to your microphone and converts speech to text in real time, streaming back results as you talk rather than waiting until you stop. It distinguishes between "interim" results (a live, still-changing guess at what you just said) and "final" results (locked in once the browser is confident), which is why text sometimes appears to adjust itself slightly a moment after you say it — that's the interim result being replaced by the final one, not an error.
Where support actually stands
This is the part worth knowing before relying on it: browser support for the Web Speech API's recognition side is inconsistent. Chrome and Edge support it well. Firefox has essentially no support for it, and Safari's support is limited and inconsistent across versions. This isn't a small footnote — if voice typing seems to silently not work at all, the browser is very likely the reason, not your microphone or a bug.
What affects accuracy
Like any speech recognition system, results are better with clear audio — minimal background noise, a reasonably close microphone, and normal speaking pace. Accents, uncommon proper nouns, and technical vocabulary are the most common source of misrecognized words, the same limitation any voice assistant has. It's genuinely good for drafting and note-taking; it's not a substitute for careful proofreading on anything that needs to be exact.
Why "runs in the browser" is the actual headline feature
Because recognition happens as a browser capability rather than a network call to a transcription service, your voice recording itself is never transmitted anywhere by this tool — a meaningfully different privacy posture than a cloud-based dictation service, and worth knowing if you're using it for anything even mildly sensitive.
Using the tool
Our Speech to Text tool wraps this browser capability in a simple start/stop interface with a live transcript and one-click copy — works best in Chrome or Edge, per the browser support notes above.
Try it here: Speech to Text.