Google to Upgrade Gboard with Faster, Reliable Offline AI Dictation

Google Will Initially Release the Update on Its Pixel Phones Only

Remember the days of Smartphones with keypads? The kind of devices that made typing a lot more sense? Then they came up with touchscreens and typing became a living nightmare for Mobile Phone users. Mostly the millennials saw this quick evolution of Smartphones and adapting was never easy. Perhaps the phrase, ‘Fat-Fingering’ is the invention of this transitionary period.

With changing communication styles, both on professional and personal fronts, there had to be a technology that acted as a golden middle, between bad typing and calling. Voice-to-text that caught on in the early part of the second decade of this century was hailed to be the solution of modern times to by-pass, well, typing.

Read More: Hyundai Mobis Accelerates Global Open Collaboration with Strategic Investment

Voice-to-text trends are still catching on! The technology has helped several users to communicate professionally and personally, on-the-go. However, as an increasing number of users complain about the accuracy and dysfunctionality of this very popular Smartphone feature, Google has decided to pay heed.

What Is Google Doing?

Google will be upgrading its Gboard— they are integrating AI-Powered offline dictation in the application. This is going a step further wherein users will not necessarily need to be connected to the internet while dictating their emails and/or texts.  Google did mention though that the initial release will be rolled out on the Pixel phone and in American English only.

Read More: Gartner Survey Shows 37 Percent of Service Leaders Are Piloting or Using AI Bots and Virtual Customer Assistants

Google’s AI blog confirms that the new ‘powered by AI’ feature will put an end to slow and unreliable speech-to-text translations. The upgrade, they say, in a nutshell, is powered by an end-to-end, all-neural, on-device speech recognizer. Although the best part is that, now, Gboard’s output will work on a character level. What this means is that the text will populate character-by-character as the user speaks.

How Google is doing this without an internet connection is through their ‘decoder’ device. It’s no secret that the ‘decoder graph’ is installed on the phone and as such does not require the internet to operate. What’s even better now that the size of the decoder now is only 80 MB as compared to what it was previously (2 GB).

Google is confident that the feature is versatile and multiple industries can leverage the same after appropriate adoption. Various languages and devices that use Gboard will be encapsulated when the release is published on a macro scale.

Read More: CEVA Computer Vision, Deep Learning and Long Range Communication Technologies Power DJI Drones