And how they asked the device for voice-input for a websearch via the UI, and the device proceeded to perform voice transcription, filling the searchbox with everything that was said...