v139 · Web APIs · demo
Feature Detection & Setup
Check on-device speech support in this browser and try live transcription using the same SpeechRecognition interface — the browser handles routing to on-device or cloud automatically.
SpeechRecognition API
⟳
On-device support
⟳
Platform
⟳
Active mode
—
Configuration
// Chrome 150+ — on-device recognition via SpeechRecognition
const recognition = new webkitSpeechRecognition();
// Standard setup — same as always
recognition.lang = 'en-US';
recognition.interimResults = true;
recognition.continuous = true;
// Chrome 150 automatically uses on-device processing when:
// 1. The browser has the on-device model for the requested lang
// 2. The platform supports it (Windows/Mac/Linux in Chrome 150)
// 3. Connectivity is unavailable (graceful degradation)
// Feature detection
const hasOnDevice =
'SpeechRecognition' in window || 'webkitSpeechRecognition' in window;
// (On-device specifics are transparent — same API surface)
recognition.onresult = event => {
const transcript = Array.from(event.results)
.map(r => r[0].transcript)
.join('');
// Identical handling regardless of on-device vs cloud
};
see also
- Privacy Architecture Comparison — data flow diagram
- Back to feature index
- ChromeStatus entry