The recently announced smartwatches based on Android Wear extensively rely on voice recognition for user interaction.

Is the voice recognition performed on-device, on the paired smartphone, or in Google's cloud? Or is it a mix of those three, e.g. online with an offline fallback?


While I don't have any official source, from testing it's clear that it's cloud based.

  • Watch paired, handheld with internet connection -> Works.
  • Watch paired, handheld without internet connection (e.g. airplane mode) -> Doesn't work.
  • Watch not paired (e.g. handheld turned off) -> Doesn't work.

the voice recognition is performed on device; information of the voice is persisted in Google's cloud

  • 1
    Do you have any sources? What do you mean by "information of the voice is persisted"?
    – onik
    Jul 4 '14 at 5:24
  • Most likely our good old "language barrier", and what was meant is "recorded voice data is processed", i.e. audio analysis is done on Google servers. I would wonder if it were different, as this is the way it's done on Android phones/tablets as well (and I doubt Wear devices have more power, so they wouldn't need that).
    – Izzy
    Jul 4 '14 at 15:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.