Mitigating false positives and/or false negatives in hot word free adaptation of automated assistant

A system comprising:

• memory storing instructions;
• and
• one or more processors operable to execute the instructions to:
• identify interactions that are each between a particular user and an automated assistant;
• for each of the interactions:
• generate a respective active speech measure based on processing respective image data and audio data from the interaction;
• generate, based on the respective active speech measures, a personalized active speech parameter for the particular user;
• store the personalized active speech parameter in association with an account of the particular user;
• subsequent to storing the personalized active speech parameter in association with the account of the particular user:
• determine, based on processing at least one image frame generated by a camera of a client device, that a face, of the particular user and that is captured by the at least one image frame, corresponds to the account;
• generate a current active speech measure for the particular user based on processing a plurality of images, generated by the camera, along with temporally corresponding audio data;
• in response to determining that the face corresponds to the account:
• use the personalized active speech parameter, stored in association with the account, along with the active speech measure in determining whether the particular user is engaging in hot word free interaction with the automated assistant;
• and
• when it is determined that the particular user is engaging in hot word free interaction with the automated assistant:
• automatically adapt processing performed by the automated assistant.

A non-transitory computer readable storage medium configured to store instructions that, when executed by one or more processors, cause one or more of the processors to:

• identify interactions that are each between a particular user and an automated assistant;
• for each of the interactions:
• generate a respective active speech measure based on processing respective image data and audio data from the interaction;
• generate, based on the respective active speech measures, a personalized active speech parameter for the particular user;
• store the personalized active speech parameter in association with an account of the particular user;
• subsequent to storing the personalized active speech parameter in association with the account of the particular user:
• determine, based on processing at least one image frame generated by a camera of a client device, that a face, of the particular user and that is captured by the at least one image frame, corresponds to the account;
• generate a current active speech measure for the particular user based on processing a plurality of images, generated by the camera, along with temporally corresponding audio data;
• in response to determining that the face corresponds to the account:
• use the personalized active speech parameter, stored in association with the account, along with the active speech measure in determining whether the particular user is engaging in hot word free interaction with the automated assistant;
• and
• when it is determined that the particular user is engaging in hot word free interaction with the automated assistant:
• automatically adapt processing performed by the automated assistant.