System and method for activation of voice interactive services based on user state
AssigneeSonde Health, Inc.
InventorJames D. Harper
Life of this patent
Estimating the term…
Abstract
The present invention provides a system for activating personal assistance services. The system includes an audio data collector adapted to collect a sample of speech, a processing module, and a service activator couple to an output device. The processing module further includes an audio feature extractor that extracts a plurality of acoustic features from the sample of speech, and a classification unit that classifies a status of a user from the plurality of acoustic features. The Service activator activates a personal assistance service according to the status of the user classified by the classification unit.
Claims16
A system for activating personal assistance services to a user, said system comprising:
- a processing module comprising an audio feature extractor, a service activator, and a classification unit;
- said service activator adapted to activate said system based on contextual health data derived by a contextual data collector coupled to the system, wherein said contextual health data is derived from non-linguistic content of a speech sample;
- an audio data collector adapted to collect said speech sample;
- said audio feature extractor adapted to extract a plurality of acoustic features from said speech sample;
- said classification unit adapted to rank said extracted plurality of acoustic features using Random Forest, Gini Impurity, and feature importance score threshold;
- said classification unit adapted to classify a health state of the user by using top-ranked features amongst said ranked plurality of acoustic features and said contextual health data derived by said contextual data collector;
- and
- said service activator adapted to present a ranked list of services to said user based on said classified health state of said user.
A method of activating personal assistance services to a user, said method comprising:
- providing a system comprising an audio data collector and a processing module, wherein said processing module comprises an audio feature extractor, a service activator, and a classification unit;
- activating said system, by said service activator, based on contextual health data derived by a contextual data collector coupled to said system, wherein said contextual health data is derived from non-linguistic content of a speech sample;
- collecting said speech sample, by said audio collector;
- extracting a plurality of acoustic features from said collected speech sample, by said audio feature extractor of said processor module;
- ranking said extracted plurality of acoustic features using Random Forest, Gini Impurity, and feature importance score threshold, by said classification unit of said processor module;
- classifying one or more of a plurality of health states of said user, by said classification unit of said processor module, using top-ranked features amongst said ranked plurality said plurality of acoustic features and said contextual health data derived by said contextual data collector;
- and
- activating a personal assistance service, by said service actuator, to present a ranked list of services to said user based on said health state of said user as classified by said classification unit.
Dependencies are inferred from the claim text.