There are many ‘great divides’ in technology, Android vs iOS (clearly Android is better), Xbox vs Playstation etc, but none may be as contentious as voice input vs text input. I myself sit firmly in the voice input camp, often preferring to voice type than to hen peck on my phone to bludgeon out some text full of typos.
(Editor: It’s true, Duncan can’t type well.)
For fellow lovers of voice input, Google’s AI division has some good news – they’re bringing speech detection to Gboard on-device for Pixel users. What does this mean? Faster recognition, less data use and of course offline usage as well.
The research team have an example below comparing the speed of online and on device:
At launch, the new feature is limited to US English on all Google Pixel variants but in the blog post they do say that with future convergence of specialised hardware (read dedicated chips on devices) and improvement in the algorithms that they hope the feature could roll out to more languages and devices in the future.
Hopefully Google is working with other chip makers, such as Huawei with their Kirin 980 SOC with on board AI processors, to ensure that features like these can roll out beyond Google-controlled silicone.
For those interested in very detail descriptions of AI training I suggest you read the Google AI blog post link below!