Method and apparatus for effectively receiving voice input to a voice recognition system

• US 5,033,088 A
• Filed: 12/21/1989
• Issued: 07/16/1991
• Est. Priority Date: 06/06/1988
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Voice Processing Corp. (Koninklijke Philips N.V.)
• Inventors: Shipman, David W.
• Primary Examiner(s): Harkcom, Gary V.
• Assistant Examiner(s): Knepper, David D.
• Application Number:
  US07/453,108
• Time in Patent Office: 572 Days
• Field of Search: 364/513.5, 381/41-43, 379/88, 379/89
• US Class Current: 704/275
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: H04M 2201/40 using speech recognition speech recognition per se G10L15/00
• CPC Class Codes: H04M 3/5166 in combination with interactive voice response systems or voice portals, e.g. as front-ends
• CPC Class Codes: H04M 3/527 Centralised call answering arrangements not requiring operator intervention

Abstract

Processing voice information is performed to complete a task with minimal human intervention. Information received by the automated system from the caller is fed into a speech recognition system, which attempts to recognize the information. The caller'"'"'s input also is recorded by a recording device. The task is completed if the information is reliably recognized. If the information is not reliably recognized, the speech recognition system optionally requests that the information be repeated. The repeated information is received by the system, and the speech recognition system again attempts to identify the information. The task is completed if the repeated information is reliably recognized. If the repeated information is again not reliably recognized by the speech recognition system, and usually without the caller knowing, the spoken information is transmitted to a human attendant who attempts to identify the information by a playback of the recorded information and possibly a visual display of the speech recognition system'"'"'s proposed solution to what the caller stated. If the caller'"'"'s information is recognized, the human attendant then inputs the information so that the system can complete the task. If the human attendant is also unable to recognize the information, the attendant directly requests further information from the caller until the desired information is received.

Claims

1. A method of processing verbal information received by an automated voice recognition system from a customer and completing a task on the basis of the information received, the method comprising the steps of:

prompting a verbal response from a human customer as an input source;

receiving information in the verbal response from the input source generated by the human customer, said information corresponding to a desired task to be completed;

recording the verbal response;

attempting to recognize said information through automation;

determining by automation if the information was reliably recognized;

transmitting the recorded verbal response of the human customer to a human attendant other than the human customer if it is determined that said information was not reliably recognized;

inputting correct information by the human attendant so as to enable the completion of the desired task through automation; and

completing the desired task with the automatically recognized information if it is determined that said information was reliably recognized or with the human attendant inputted information if it is determined that said information was not reliably recognized.

2. A method as defined by claim 1 and further comprising the step of:

prompting a repeated verbal response from the human input source.

3. An apparatus for processing verbal information from a customer to complete a task comprising:

means for prompting a verbal response from a human customer as an input sources;

a speech recognizer including means for recognizing information in the verbal response and means for determining if the information was reliably recognized;

recording means for recording the verbal response;

means for enabling information to be manually input into the speech recognition system; and

means for completing the task automatically if the information was reliably recognized by said speech recognizer, said means for completing the task completes the task with the recognized information, and including means for manually completing the task if the information was not reliably recognized by said speech recognizer, said means for completing the task completes the task with the information inputted by a human attendant other than the human customer after reviewing the recorded verbal response.

4. A method of processing verbal information received by an automated voice recognition system to complete a task, said method comprising the steps of:

prompting a verbal response from a human customer source, said verbal response including information for carrying out a specific task;

receiving said information in said verbal response from said human customer source;

recording said verbal response;

attempting to recognize said information via said voice recognition system;

determining if said information is either recognizable or unrecognizable by said voice recognition system; and

carrying out said specific task directly if said information is recognizable on the one hand, or on the other hand, transmitting said recorded verbal response to a human attendant other than the human customer if said information is unrecognizable, to allow said human attendant to provide the information corresponding to said specific task to enable completion of said specific task.

5. An apparatus for processing verbal information for completing a task comprising:

means for prompting a verbal response from a human customer as an input source;

speech recognizing means for attempting to recognize information in said verbal response and including means for determining if said information is recognizable or unrecognizable;

recording means for recording the verbal response;

means for completing said task;

means for enabling said information to be manually input by a human attendant other than the human customer to trigger said means for completing the task to complete the task if said information is unrecognizable, on the one hand, or on the other hand, for triggering directly said means for completing the task to complete the task if said information is recognizable.