Gboard Gets All-Neural On-Device Speech Recogniser, Boosting Dictation on Pixel Phones

Google is adding an artificial intelligence (AI) offline dictation feature on its Gboard keyboard for Pixel phones that would allow users to speak out emails and texts even without an Internet connection.

"We're happy to announce the rollout of an end-to-end, all-neural, on-device speech recogniser to power speech input in Gboard which is always available, even when you are offline," Johan Schalkwyk, Speech Team, Google wrote in a blog post on Tuesday.

Google has designed the feature to work at the character level. The company added that it has developed a model that has been trained with RNN Transducer (RNN-T) technology that is compact enough to reside on a phone.

"This means no more network latency or spottiness — the new recogniser is always available, even when you are offline. The model works at the character level, so that as you speak, it outputs words character-by-character, just as if someone was typing out what you say in real-time, and exactly as you'd expect from a keyboard dictation system," Schalkwyk said.

To increase use-parameters of the speech recognition feature, Google said it has hosted the new model on device in order to avoid the latency and inherent unreliability of communication networks.

For now, the on-device Gboard speech recogniser has been made available in American English language on all Pixel devices.

"Our new all-neural, on-device Gboard speech recogniser is initially being launched to all Pixel phones in American English only. Given the trends in the industry, with the convergence of specialised hardware and algorithmic improvements, we are hopeful that the techniques presented here can soon be adopted in more languages and across broader domains of application. We are hopeful that the techniques presented here can soon be adopted in more languages and across broader domains of application," Schalkwyk added.