Speech-recognition browser support

This is part of the reference hub. Browser dictation uses the Web Speech API, specifically the SpeechRecognition interface (exposed as webkitSpeechRecognition in Chromium and WebKit browsers). Recognition usually requires a network connection and microphone permission.

Support matrix

Browser	SpeechRecognition	Notes
Chrome (desktop)	Yes	Prefixed as webkitSpeechRecognition; reliable
Edge	Yes	Chromium-based; same support
Safari (macOS / iOS)	Partial	Prefixed; from Safari 14.1 (macOS) / 14.5 (iOS); routes audio to Apple, behavior less consistent
Firefox	Off by default	Behind the dom.webspeech.recognition.enable flag; not available to end users out of the box
Chrome (Android)	Yes	Works with mic permission

Feature detection

Because the interface is prefixed in most browsers, detect it before showing a dictate control:

const Recognition = window.SpeechRecognition || window.webkitSpeechRecognition;
if (!Recognition) {
  // No support (e.g. Firefox) — fall back to typing
}

What this means in practice

• Offer a typing fallback always. Firefox users have recognition off by default, so the dictate button won't work for them out of the box. The typed path must produce the same result.
• Expect cloud round-trips. Recognition commonly streams audio to a server, so it needs connectivity and can drop mid-session.
• Don't block the workflow on voice. Dictation is a convenience layer; see dictation isn't working for the user-facing troubleshooting version of this table.

Support reflects current stable browser versions as of 2026.

Frequently asked questions

Why doesn't dictation work in Firefox?

Firefox keeps SpeechRecognition disabled by default. The interface sits behind the dom.webspeech.recognition.enable flag in about:config, so it isn't available to end users out of the box. Use Chrome or Edge for voice, or type.

What's the difference between SpeechRecognition and webkitSpeechRecognition?

They're the same interface; webkitSpeechRecognition is the vendor-prefixed name that Chromium and WebKit browsers expose. Code that wants broad support checks for both: window.SpeechRecognition || window.webkitSpeechRecognition.

Is recognition done on-device or in the cloud?

In most browsers it streams audio to a server for transcription, so it generally needs an internet connection and results can stop on a dropped connection.