Google has added an AI-powered offline dictation on its Gboard keyboard. The new update will now allow users to dictate emails and texts without using the internet; making the process more fast and reliable.
The company worked hard for five years to launch the new feature. The new feature replaces the older version of the software; replacing the traditional three-step process for speech recognition for a single-step solution.
Previously, the software utilized different components to use audio waveforms, match sound with phonemes and then mix those phonemes into a written form. The new version does all these in a single task. Similarly like an index in a book, the decoder graph matches audio waveforms in the written form.
The new version also made its decoder graph smaller by 25 times. Previously, the size of the graph was 2GB and now it’s only 80MB.
Currently, the new launch is only available for Pixel Phones for in English and will be rolled out worldwide in the near future- spreading the reach of AI voice recognition. The company aims to make the feature available in other languages as well.
According to a member of Google’sspeech recognition and mobile input group, the company can roll out the new feature in other smartphones as well. It will happen once the company rolls out the new feature in other languages as well.