Data processing: speech signal processing – linguistics – language – Speech signal processing – Application
Reexamination Certificate
2001-06-04
2003-11-04
{haeck over (S)}mits, T{overscore (a)}livaldis Ivars (Department: 2641)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Application
C379S088010, C704S246000
Reexamination Certificate
active
06643622
ABSTRACT:
TECHNICAL FIELD
The present invention relates generally to methods, software, and apparatuses for providing requested information from a database, and more particularly to providing a speech recognition interface between a live data retrieval assistance operator and a computer database.
BACKGROUND ART
The telecommunications industry is constantly seeking ways of reducing the high labor costs associated with operator assisted calling services while simultaneously increasing customer satisfaction. Because of the need to reduce labor costs, the trend in operator assisted call management is to automate as much of the task as practicable without sacrificing customer satisfaction. For instance, in some cases the caller is greeted and prompted for information by a recording made in the voice of a live operator, who comes on line shortly before the caller utters his or her request. Such a system improves efficiency and lowers labor costs because the automated response reduces operator work time and reduces operator fatigue by automating this initial greeting phrase. This type of automated response increases customer satisfaction because the partial automation of the system is transparent to the caller; the caller hears only one voice and is unable to distinguish the recorded message from the live operator.
In most current day systems, the operator interprets the calling customer's request and then communicates with a computer-based search algorithm and stored database via a conventional keyboard. The computer forwards the search results to the operator's screen, and the live operator releases the selected listing to the caller via an audio response system if an exact match has been found by the computer. If the computer finds multiple listings that match the search request, the live operator must then question, suggest, or identify the number with the caller's assistance. If the customer accepts one of these numbers, the call is released to the audio response system. If the customer does not accept one of these numbers, the operator often must give the caller a nothing found report, or engage in a second search if additional listing details are provided by the customer.
In an effort to further reduce labor costs, there have been numerous attempts to fully automate a telephone directory assistance system through the use of voice recognition technology and other technological advancements. For instance, U.S. Pat. No. 5,479,488 to Lennig et al., U.S. Pat. No. 5,659,597 to Bareis et al. and U.S. Pat. No. 5,638,425 to Meador, III et al. all teach automated directory assistance systems that rely upon a computer to play recorded prompts to the caller and then interpret the caller's responses with a speech recognition system. While such systems can conceivably lower labor costs, they unfortunately can result in a large decrease in customer satisfaction because speech recognition technology is not yet capable of effectively recognizing the voices of the entire pool of calling customers. In addition, even if the words spoken by the caller are recognized, there remains a significant probability that the computer will be unable to understand and translate the recognized words into a searchable query. Because of these shortcomings, automated systems must often rely upon live operators to act as backups when the automated system fails. Even when these systems perform correctly, customer satisfaction is often reduced because it is generally difficult to make the operation of such a system so transparent that the calling customer is unable to realize that they are communicating with a machine rather than a live operator assistant.
It should therefore be understood that the overall success of a directory assistance system requires an appropriate balancing of a variety of competing interests including calling customer satisfaction, job satisfaction of live operator assistants, labor and equipment costs associated with the system, and the satisfaction of subscribers who often pay to have their number listed with the director assistance service (e.g. Yellow Pages). Thus, the primary dimensions by which the success of a directory assistance system may be measured include: increasing the number of correct listings given by the system to caller requests; correspondingly decreasing the number of incorrect listings given out and thereby lessening the costs to the subscribers who pay for the service, if applicable; improving the interface between calling customers and the operator so that the utilization of new technology is transparent to the calling customer; and finally, improving the qualitative aspects of the live operator's job by introducing appropriate technology. Finally, in the telephone industry, every second, or a portion of every second saved per call can mean a literal savings of hundreds of thousands of dollars annually.
Apart from directory assistance, many industries maintain computer databases that contain information needed by requesters for some purpose. These requesters may be in house customers, or might be outside customers seeking information for payment of a fee. Because access to these databases is often restricted, and because the database often includes a relatively complex structure rendering it difficult to search, a live operator with knowledge of the database contents and structure is often positioned between the information requestor and the computer database. These live operators interpret identifying information provided by a requester and translate that information into a searchable inquiry that is communicated to a searching means having access to the information database. Like the directory assistance systems discussed above, there is a desire to further automate these data retrieval systems while maintaining or increasing requestor satisfaction and/or the accuracy and/or timeliness of information provided to the requester.
The present invention is directed to improving upon these and other aspects of data retrieval assistance systems.
DISCLOSURE OF THE INVENTION
In one embodiment, a data retrieval assistance system provides requested information to a requester based upon identifying information provided by the requester to the system. The system includes means for conveying the identifying information to a live operator, such as by a direct communication link or by playing back recorded requester information to the live operator. A computer has access to an information database, and a means for searching the information database in response to an inquiry. A voice communication interface from the live operator to the computer includes a speech recognition system capable of converting a spoken inquiry from the live operator into an interpreted searchable inquiry that is available to the means for searching. Preferably, the speech recognition system is personalized and trained to recognize the specific live operator, and the operator preferably has some insight into the database structure and the searching algorithms used by the computer. In this way, overall system reliability can be improved while retaining the perceived live contact that requesters desire and have come to expect.
In another embodiment of the present invention, a method of data retrieval assistance provides requested information, which capable of being stored and retrieved from a computer database, to a requestor based upon identifying information provided by the requester. The method includes the initial steps of establishing a communication link with the requestor, and then conveying the identifying information provided by the requestor to a live operator. A speech recognition system is positioned between the live operator and a computer that has access to an information database. The database is searched at least in part in response to a spoken inquiry by the live operator to the computer via the speech recognition system, wherein the spoken inquiry is an interpretation of the identifying information provided by the requester. The method is completed by providing the search res
Hornback Don
Stuart Robert O.
Stuart Scott P.
Armstrong Angela
{haeck over (S)}mits T{overscore (a)}livaldis Ivars
LandOfFree
Data retrieval assistance system and method utilizing a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Data retrieval assistance system and method utilizing a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Data retrieval assistance system and method utilizing a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3156427