When Google introduced Google assistant, it was sure to give a tough competition to other personal assistants. Google has been improving it since then.
A recently granted patent by Google reveals plans of introducing another interesting feature. This can differentiate Google Assistant in the same way its search engine results differentiate themselves from that of Yahoo or Bing.
The patent discloses capturing voice command precisely using more than one device. Being the search engine pioneer, Googleknows better how voice commands will change the way people interact with devices.
Talking is the most basic form of communication and voice commands will provide natural means of interaction with our devices. Hence, an assistant cant be smart if it fails to decipher the exact words uttered by its master. And again, Google knows that hence this patent.
At present, Google Assistant captures your words correctly when you are near to the device it lives in. Problems arise when it receives a command from some distance and can’t decipher it clearly. It misses especially the last words of a command.
Increasing strength of mic is not a great solution and hence Google will be using more than one device to know the exact words you said. Confused? Keep reading:
Let say you are walking away from your room to kitchen while giving reduce the temperature by 20 command to your smartphone lying on the bed. The Google Assistant may capture reduce the temperature only!
Google Home nearby will capture the rest of the words, however. Both devices will then team up to send full command to the server to perform the right action.
Handpicked Suggestion: Why is Samsung getting so obsessed over smartphone unlocking methods?
Google will also take care of homophones words like new and knew that sound similar but have a completely different meaning that can confuse smart assistants.
Google will be using a confidence score to find these words based on high/low voice pitch.
The high pitch will have a high confidence value. For the word new, it will have a confidence value 0.90, for example. And for knew, it will have a confidence score of 0.80. Based on these values, Google Assistant will find the exact word.
In another scenario where both devices could be nearby to each other, they may capture a command simultaneously. For example, you may say print the paper, and your smartphone may decipher the command correctly while another device nearby may get tint the paper.
In situations like this, Google Assistant will analyze the timing of both devices. After finding the slightest difference in timing and words of commands, it will treat it as one command and replace the mistaken word with the right word which in this case is, print.
Another problem arises when two person give different commands simultaneously while walking away from one device to another. Here, Google will use users profile and time of commands to recognize both commands accurately.
For instance, if a first user says, raise the volume of TV by 20 and a second user says, raise the temperature by 100. The first device might capture first portion of the command which is raise the volume and the second device gets by 20. And for the second user, the first portion of the first command raise the temperature will get captured by the second device and the rest , by 100, by the first device. Google Assistant then analyzes both users profiles and time difference between words and also confidence value to find both first and second users commands.
Besides, Google Assistant also got intelligent enough to find a particular object without mentioning it. Because sometimes, a person just gives command like turn that off’ without mentioning what object to turn off TV or AC?
Google will find the right device by checking which device you gave command to turn on. This could be little problematic as per my opinion. What do you say?
And how can I forget to tell you that Google Assistant will also remind you about the things books, etc. that slipped from you memories after you lent that to someone.