Google announced an update that is aiming to solve the most annoying and long-standing problems in voice assistants: mispronunciation and context awareness. This is a great update (assuming it works as expected), not only can fix the recognition of your sentence, but also find that Google Assistant has improved over time, but it is not out of date.
Take India as an example, where the accent, pronunciation and language are different. Sometimes, explaining the actual function of the virtual assistant can be frustrating. At this point, it is up to the user to change the way he speaks, not the helper to adapt to the user. There is no flexibility. Google wants the assistant to accurately recognize and pronounce, especially unusual names.
This makes sense, because it will make you feel frustrated when you try to make a phone call with your voice to save time, and eventually you need to do more work to make the voice assistant understand what you want. The previous one assumes that you use Android Auto while driving and ask the assistant how to get to the local store, but you don’t know the name.
Finally, you stop the car and manually enter the location. This technique is useless in this situation. Fixed the problem when setting multiple timers at the same time.The wizard used to have trouble defining the timer you are talking about, but in the future you will be able to find the answer. This means that if you ask “OK Google, cancel my second timer”, you know which one it is.
Apply Transformer Bidirectional Encoder Rendering (BERT) to use previous interactions to improve the quality of the conversation and understand what is currently displayed on the device screen to answer any other questions and achieve more natural communication. Back in the next few days, new updates will be released only for smart speakers tuned in American English.
The company said that they will soon be applicable to smart phones and smart displays. Details of availability in this area have not yet been reported.