Firefox 160 Runs Speech Recognition on the Device, Which Changes the Privacy Pitch

Firefox 160 shipped on 10 November 2026, and the change we would flag to clients is in the Web Speech API. Firefox now supports on-device speech recognition on desktop, in 40 languages. MDN's notes are clear about the limits: it is not supported on Android, and remote, cloud-based recognition is not supported in Firefox.
That matters for one reason: privacy. Voice search and voice notes on a website have always come with an awkward question about where the audio goes. With on-device recognition, the honest answer in Firefox is that it stays on the machine. For healthcare, legal and finance clients, that is a much easier conversation.
It is still not something to build a whole interface around. Support differs between browsers, and Firefox users on Android get nothing. So treat voice as a progressive enhancement. Feature-detect speech recognition, show the microphone button only when it is available, and keep a normal text field as the main way in.
Test with real accents and real background noise before launch. A voice feature that mishears a postcode is worse than none at all.