Interactive speech recognition system
AssigneeNuance Communications, Inc.
InventorsLars König, Rainer Saam, Andreas Löw
Life of this patent
Estimating the term…
Abstract
An interactive speech recognition system includes a database containing a plurality of reference terms, a list memory that receives the reference terms of category “n,” a processing circuit that populates the list memory with the reference terms corresponding to the category “n,” and a recognition circuit that processes the reference terms and terms of a spoken phrase. The recognition circuit determines if a reference term of category “n” matches a term of the spoken phrase.
Claims18
A method of interactively recognizing a spoken phrase using a database in which is stored reference terms, the reference terms being organized into categories, the method comprising:
- (a) forming, by an interactive speech recognition system, an initially empty path in an output memory of the interactive speech recognition system, receiving terms identified in the spoken phrase into an input memory of the interactive speech recognition system, and loading one or more reference terms of a first category from the database into a list memory of the interactive speech recognition system;
- (b) determining by the interactive speech recognition system when one or more of the reference terms in the list memory matches one or more terms of the spoken phrase in the input memory, and continuing with process (a) when there is no match;
- (c) determining by the interactive speech recognition system, that the spoken phrase has been recognized when one or more paths in the output memory uniquely identify a database entry;
- (d) determining by the interactive speech recognition system when the list memory has sufficient capacity to load the reference terms of a set of subsequent categories, each category in the set being determined by a respective path stored in the output memory, and when the list memory has insufficient capacity, eliminating by the interactive speech recognition system one or more paths from the output memory based on supplemental user input;
- (e) replacing, by the interactive speech recognition system in the output memory, each given path with a set of extended paths for disambiguating the spoken phrase, each path in the set of extended paths being formed by the interactive speech recognition system by concatenating to the given path a different reference term in the list memory that matches the given path, wherein when the given path matches no such reference terms, the set includes no extended paths and the given path is deleted from the output memory by the interactive speech recognition system;
- (f) loading the list memory by the interactive speech recognition system with the reference terms of the set of categories, and continuing with process (b);
- and
- (g) controlling, by the interactive speech recognition system after determining that the spoken phrase has been recognized, a system in communication with the interactive speech recognition system.
A non-transitory computer-readable storage medium having processor executable instructions to interactively recognize a spoken phrase using a database in which is stored reference terms, the reference terms being organized into categories, by performing the acts of
- (a) forming, by an interactive speech recognition system, an initially empty path in an output memory of the interactive speech recognition system, receiving terms identified in the spoken phrase into an input memory of the interactive speech recognition system, and loading one or more reference terms of a first category from the database into a list memory of the interactive speech recognition system
- (b) determining by the interactive speech recognition system when one or more of the reference terms in the list memory matches one or more terms of the spoken phrase in the input memory, and continuing with process (a) when there is no match;
- (c) determining by the interactive speech recognition system that the spoken phrase has been recognized when one or more paths in the output memory uniquely identify a database entry, thereby indicating that the spoken phrase has been recognized;
- (d) determining by the interactive speech recognition system when the list memory has sufficient capacity to load the reference terms of a set of subsequent categories, each category in the set being determined by a respective path stored in the output memory, and when the list memory has insufficient capacity, eliminating, by the interactive speech recognition system, one or more paths from the output memory based on supplemental user input;
- (e) replacing by the interactive speech recognition system in the output memory, each given path with a set of extended paths for disambiguating the spoken phrase, each path in the set of extended paths being formed by the interactive speech recognition system by concatenating to the given path a different reference term in the list memory that matches the given path, wherein when the given path matches no such reference terms, the set includes no extended paths and the given path is deleted from the output memory by the interactive speech recognition system;
- (f) loading the list memory by the interactive speech recognition system with the reference terms of the set of categories, and continuing with process (b);
- and
- (g) controlling, by the interactive speech recognition system after determining that the spoken phrase has been recognized, a system in communication with the interactive speech recognition system.
An interactive speech recognition system for interactively recognizing a spoken phrase, the system comprising:
- a database containing a plurality of reference terms, the reference terms being organized into categories;
- an input memory configured to store terms identified in a spoken phrase;
- an output memory configured to store a plurality of paths, wherein a path is a concatenation of matched reference terms of different categories;
- a list memory configured to store reference terms of a set of categories;
- a processing circuit configured to populate the list memory with the reference terms of the set of categories;
- a recognition circuit configured to determine when a reference term in the list memory matches one or more terms of the spoken phrase in the input memory;
- the recognition circuit or the processing circuit further including:
- means for determining that the spoken phrase has been recognized when one or more paths in the output memory uniquely identify a database entry;
- means for determining when the list memory has sufficient capacity to load the reference terms of a set of subsequent categories, each category in the set being determined by a respective path stored in the output memory;
- means for eliminating one or more paths from the output memory based on supplemental user input when the list memory has insufficient capacity;
- means for replacing, in the output memory, each given path with a set of extended paths for disambiguating the spoken phrase, each path in the set of extended paths being formed by concatenating to the given path a different reference term in the list memory that matches the given path, wherein when the given path matches no such reference terms, the set includes no extended paths and the given path is deleted from the output memory;
- and
- means for loading the list memory with the reference terms of the set of categories.
Dependencies are inferred from the claim text.