Semi-supervised training of destination map for call handling applications
First Claim
1. A method, comprising the steps of:
- receiving a speech signal from a caller;
producing a speech recognition string based on the received speech signal;
determining if there is destination information that matches the speech recognition string;
wherein if there is destination information that matches the speech recognition string, the method proceeds with any ofoutputting the destination information to the caller, orconnecting the caller to a destination that corresponds to the matching destination information; and
wherein if no destination information matches the speech recognition string, the method proceeds withdirecting the caller to an operator to initiate an interaction,providing the operator with request information that comprises any of the received speech signal or a request by the caller for the destination during the interaction;
recording destination identifying information input from the operator in response to the request information, andthe method proceeds with any ofoutputting the recorded destination identifying information to the caller, orconnecting the caller to the matched destination using the recorded destination identifying information.
1 Assignment
0 Petitions
Accused Products
Abstract
A method of semi-supervised synonym inference for a call handling application, such as automated directory assistance or call routing, is described. In one embodiment the method comprises examining a database of caller interaction results from a directory assistance system that includes an automated directory assistance engine, detecting a specified characteristic in the caller interaction results, and using the detected characteristic to automatically train a destination map, where the destination map is for use by the automated directory assistance engine in automatically mapping human speech to a destination. The detecting of the specified characteristic in the caller interaction results may include a statistical analysis of the caller interaction results for each of one or more speech recognition strings.
21 Citations
22 Claims
-
1. A method, comprising the steps of:
-
receiving a speech signal from a caller; producing a speech recognition string based on the received speech signal; determining if there is destination information that matches the speech recognition string; wherein if there is destination information that matches the speech recognition string, the method proceeds with any of outputting the destination information to the caller, or connecting the caller to a destination that corresponds to the matching destination information; and wherein if no destination information matches the speech recognition string, the method proceeds with directing the caller to an operator to initiate an interaction, providing the operator with request information that comprises any of the received speech signal or a request by the caller for the destination during the interaction; recording destination identifying information input from the operator in response to the request information, and the method proceeds with any of outputting the recorded destination identifying information to the caller, or connecting the caller to the matched destination using the recorded destination identifying information. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10)
-
-
11. An apparatus, comprising:
-
a storage for storing a destination map comprising a plurality of destination information; an interface for receiving a speech signal from a caller; a recognizer module for producing a speech recognition string based on the received speech signal; and at least one processor programmed to determine if there is destination information that matches the speech recognition string; wherein if there is destination information that matches the speech recognition string, the processor is programmed to any of output the destination information to the caller, or connect the caller to a destination that corresponds to the matching destination information; and wherein if there is no destination information that matches the speech recognition string, the processor is programmed to direct the caller to an operator to initiate an interaction, provide the operator with any of the received speech signal or a request by the caller for the destination during the interaction, record destination identifying information input from the operator in response to the request by the caller, and proceed to any of output the recorded destination identifying information to the caller, or connect the caller to the matched destination using the recorded destination identifying information. - View Dependent Claims (12, 13, 14, 15, 16, 17, 18, 19, 20, 21)
-
-
22. A method, comprising the steps of:
-
providing a destination map, wherein the destination map comprises destination identifying information corresponding to a plurality of destinations, and a plurality of stored speech recognition strings associated with the destination identifying information; receiving a voice signal from a caller, wherein the voice signal comprises a request for a destination; producing a speech recognition string based on the received voice signal; calculating a listing hypothesis for a selected one of the plurality of destinations using the produced speech recognition string; directing a caller to an operator when any of the system has low confidence with a listing hypothesis for the selected destination, or when the caller responds negatively to confirmation of the selected destination; receiving information from an interaction between the caller and the operator; and updating the destination map based on the received information.
-
Specification