Google has introduced some new updates that will allow the Assistant to better announce complicated names and understand the context of the conversations with which you share. First of all, you will soon have the opportunity to teach him to pronounce your contact’s name, just as you can already instruct him to pronounce his name correctly. According to Google, “the assistant will listen to your pronunciation and remember it”, and the company does not need to keep the recording.
Google said the feature will be available in English first, although it “hopes” that it will also be available in other languages. After receiving the new update on your phone, click the “Record your own content” option under the “Contact” field in the “Wizard” settings menu to get started.
I can only speak for myself, but I found that I ended up mostly using the Wizard to set the timer, and Google is improving this too. The company said it completely rebuilt AI’s natural language understanding model to more accurately understand the context. Thanks to Google’s BERT technology, you can now process words as a whole, rather than one word at a time. Google claims that by adding these features, “the assistant can now respond almost 100% accurately to alarm and timer tasks.” In practice, this means that you can do something similar, like change your mind in the middle of a sentence. The wizard will follow your ideas. If you live in the United States, you can immediately try this feature on an assistant smart speaker and enable it on your phone and smart display to “follow” immediately.
Finally, Google is using the same BERT technology to further enhance the Assistant’s contextual awareness. So if you start a conversation about Miami and finally make a recommendation on the best beaches, that means you mean Miami beaches. Similarly, you can also understand issues related to what you are looking at on your smartphone or screen. All these improvements should make the conversation with the assistant more natural and useful.