Natural language generation system for producing natural language instructions
AssigneeMotorola, Inc.
InventorsJerry L. Morgan, Alan M. Frisch, Erhard W. Hinrichs
Life of this patent
Estimating the term…
Abstract
In certain computerized systems, computer-generated spoken instructions are made more natural sounding by using recursion to identify significant points in a series of similar instructions when an utterance should be made. No instructions are issued when such instructions would provide little or no information to a user. In a vehicle navigation system, instructions to a driver are typically issued at points along the drivers route where the driver should perform some critical action.
Claims8
A method of generating instructions to navigate along a route between a starting location and a desired destination location, said route being comprised of at least one road segment, a plurality of the characteristics of a plurality of road segments being known and defined in a series of data records, said method comprised of the steps of:
- a) planning a speech action for a user, that describes at least one operation to be performed by the user, to traverse at least one segment between the starting location and the destination location;
- b) determining which speech act, from a predetermined set of possible speech acts, is required to convey an instruction to the user to perform said at least one operation;
- c) for the speech act of step b:
- forming a subset of speech clauses from a set of possible speech clauses to convey said instruction;
- clausifying said subset of speech clauses prior to generating an audible speech clauses:
- generating an audible speech clause from said subset of speech clauses.
An apparatus for generating instructions to navigate along a route between a starting location and a desired destination location, said route being comprised of at least one road segment, a plurality of the characteristics of a plurality of road segments being known and defined in a series of data records, said apparatus comprised of:
- means for planning a speech action for a user, that describes at least one operation to be performed by the user, to traverse at least one segment between the starting location and the destination location;
- means for determining which speech act, from a predetermined set of possible speech acts, is required to convey an instruction to the user to perform said at least one operation;
- means for forming a subset of speech clauses from a set of possible speech clauses to convey said instruction;
- means for clausifying said subset of speech clauses prior to generating an audible speech cluases:
- means for generating an audible speech clause from said subset of speech clauses.
Dependencies are inferred from the claim text.