10 ANDROID ADVISOR • ISSUE 62
GOOGLE I/O 2019
The secret ingredient? Moving Google Assistant’s
processing from the cloud to your device. CEO Sundar
Pichai claims that the company managed to condense
Assistant’s speech recognition models from 100GB in
size all the way down to a mere 50MB, letting it run
directly on your phone. “This breakthrough enabled
us to create a next-generation Assistant that processes
speech on-device at nearly zero latency, with
transcription that happens in real time, even when
you have no network connection,” Google revealed
in its announcement post.
Google says the shift allows Assistant to process
your commands and requests up to 10 times faster
than before. Paired with the software’s Continued
Conversation feature, which allows you to converse
naturally with Assistant without the need to
repeatedly utter wake words, and enhancements
to how Google’s technology can understand the
context of your requests through the lens of the
people, places, and events most important to you,
this core Android feature now operates much faster
than you’d ever expect it to.
In an extensive on-stage demonstration, a
Google employee blew through a rapid-fire list of
requests and commands, and Assistant handled
them all masterfully, leaping from app to app and
understanding the context of requests such as “Get
an Uber to my hotel” and “when’s my flight?” The
company also showed how the next-generation
Assistant can help you quickly find specific photos
to include in texts, or even help composing and
sending emails for you, all in a few seconds.