Telephonic communications – Audio message storage – retrieval – or synthesis – Voice activation or recognition
Reexamination Certificate
2001-01-16
2003-06-24
Tsang, Fan (Department: 2645)
Telephonic communications
Audio message storage, retrieval, or synthesis
Voice activation or recognition
C704S275000, C704S251000, C704S252000, C379S088160
Reexamination Certificate
active
06584180
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to automatic voice interaction systems such as call centers and computer systems which include voice recognition means and voice response means and to methods of their use.
Highly precise and inexpensive voice recognition means such as voice recognition software which operates in computers such as “ViaVoice” developed by IBM, Ltd. has recently become available, and an automatic voice response system using such voice recognition means has been suggested. An example of such an automatic voice interaction system includes a call center system which deals with customers who call the system to perform telephone shopping. Instead of a human operator, the shopping customer interacts with a computer through a voice recognition means and a voice response means to place an order for goods and/or services using voice communication.
In such telephone shopping systems in particular, large numbers of telephone calls to the call center are often made on the same day, and hence a large number of operators are required. By the use of the apparatus, computer program and method of the instant invention, the number of operators required for telephone shopping can be advantageously reduced. Generally, in such an automatic voice interaction system, a telephone shopping customer for example, herein after referred to as the caller, responds to an order question generated by a voice response means of a computer that took the call from the caller. The callers response to the system order question is provided to a voice recognition means and a recognition result is generated as text or an equivalent form of order information in the computer. This text is then synthesized into a voice form by the voice response means and sent to the caller in the form of an order confirmation question. Then, a caller is allowed to confirm the order to the computer through the voice recognition means.
The voice recognition steps in this method may fail owing to various kinds of causes such as the level of precision of the voice recognition means, tone quality and intonation of the caller, or a voice interpretation mistake by either the caller or the voice recognition means, perhaps due to noises generated around the customer. In this case, the caller is requested to utter the order information again. When in the example voice interaction system, the voice recognition fails repeatedly for a predetermined number of times, for example, three times, a private branch exchange switches the caller telephone connection so that the caller can speak with a human operator. Because the precision of presently employed voice recognition means is not 100% reliable, and because some callers can not interact effectively without the amenities of personal human conversation, sometimes called small talk, a human operator who can pick up on a telephone shopping call to an automatic voice interaction system is essential.
Current voice recognition means show a reasonable recognition accuracy in obtaining one item of order information from a caller when the machine has the opportunity to perform repeated recognition operations interleaved with voice response confirmation questions to the caller. When a larger number such as for example, four different order information items are required, the information capture accuracy is degraded. For example, when a caller must enter four items including a name of a bank to which money is transferred, a name of a bank's branch, a sum of money and a transfer date, all by telephone voice communication to voice recognition means, the probability of completing the information input without switching from an automatic voice interaction system to a human operator is 0.9×0.9×0.9×0.9=0.6561. In other words, nearly 35% of the callers drop out from the automatic voice interaction process before they reach the completion of the order processing, and a human operator must deal with the caller's order instead of the automatic system. To be more specific, although there is some advantage to be obtained in adopting an interactive automatic voice response system, the automatic voice response system can not achieve total unmanned operation. In other words with the accuracy assumed above, about 100 lines can be supported by 34 operators. One reason that a better automated rate can not be achieved is that once an operator intervenes in a call being inadequately processed by an automated system, it is difficult for the operator to return the call to the automated system without creating caller dissatisfaction. Customer satisfaction seems to require that once an operator starts dealing with a caller, the call must continue to be handled on a manual basis and that requires a certain amount of time. Intervention by an operator who takes over a call when the caller is having difficulty with an automated system only seems to confirm in the callers mind that the automated system does not or can not work properly and further reduces the callers patience when requested to deal with repeated confirmation questions asked by an automated system using voice response means.
U.S. Pat. No. 6,044,142 issued to Hammerstrom et al attempts to solve this problem by switching a call to an operator for special assistance and then returning the call to their “intelligent network services”. This sounds good but the caller of Hammerstrom is aware of the operator intervention and will resist being shunted back into the computer system.
SUMMARY OF THE INVENTION
An advantage of the present invention is that it reduces the number of times that an operator must take over a call being inefficiently handled by an automated system while at the same time reducing the number of times that a caller is asked to repeat uttering information needed by the automatic system.
Another advantage of the present invention is that it allows human intervention into the automated information gathering process without alerting a caller that such intervention is occurring. Accordingly the caller is not in a position to expect further manual communication with a human operator and caller satisfaction is not degraded. Actually, caller satisfaction is more likely to be enhanced because an order for example may be expeditiously completed without the delay required for the call to be transferred to a human operator.
These and other advantages of the present invention are obtained by a novel automatic interactive voice communication system apparatus, computer program and method of operation that provides for automatic screening of voice recognition accuracy and for limited human intervention in the automatic interactive voice communication process without transferring the call. The accuracy of the automatic process is thereby improved without the need to abort the automatic process and transfer the call to a human operator.
An automatic voice system of a first aspect of the present invention for receiving a voice input from a caller and for transmitting a voice message to the caller includes:
automatic means for receiving a call and storing a callers voice input;
voice recognition means for analyzing the voice input and generating a voice recognition result;
screening means for recognizing when the voice recognition means has difficulty in recognizing intelligible information in the callers voice input, the screening means including screener interface means for reproducing the voice request when the voice recognition means has difficulty in recognizing an intelligible request in the callers voice request and for receiving a recognition result entered by a human voice inspector (hereinafter referred as a screener) of the callers voice input; and switching means for switching the call from the automatic means for receiving a call to an operator interface means by which an operator and the caller directly talk with each other.
In one embodiment such a system, the automatic voice response means transfers the received voice of the caller to the voice recognition means, and the voice recognition mea
Chow Ming
Hesse kARL O.
International Business Machines Corp.
Tsang Fan
LandOfFree
Automatic voice response system using voice recognition... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Automatic voice response system using voice recognition..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Automatic voice response system using voice recognition... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3094473