As 9to5Google was able to discover in a new teardown of the current APK of the Google app, Google is probably working on a kind of personalized speech recognition for the Google Assistant.
The whole thing is intended to make it easier to recognize frequently used words and names – certainly also helpful when dealing with nicknames and phrases that are used regularly. The settings of the current APK, therefore, contain the following lines as an explanation of the function:
“Save audio recordings on this device to help the Google Assistant better recognize what you’re saying. Audio recordings remain on this device and can be deleted at any time by disabling personalized voice recognition.”
Accordingly, audio data of corresponding statements should then also be saved on the device and analyzed in order to be able to improve the necessary transcription in the future. An aspect that, in view of data security, will certainly not appeal to everyone.
Nest Hub and also the Nest Mini of the 2nd generation already do the request processes in this way. In the future, smart home devices and even Android as an operating system could also use this concept. It can be assumed that this could then be an opt-in function, so you have to decide for yourself whether to activate and use it.
As is so often the case, it is not yet possible to say exactly when this function will ultimately be distributed to us consumers. At the last I/O, the company gave at least oncean outlook on all sorts of improvements that we might see in the future.