Syntax-driven, operator assisted voice recognition system and methods

• US 7,136,814 B1
• Filed: 11/03/2000
• Issued: 11/14/2006
• Est. Priority Date: 11/03/2000
• Status: Expired due to Term

• Current Assignee: The Procter & Gamble Company
• Original Assignee: The Procter & Gamble Company
• Inventors: McConnell, Theodore Van Fossen
• Primary Examiner(s): CHAWAN, VIJAY B
• Application Number:
  US09/706,326
• Time in Patent Office: 2,202 Days
• Field of Search: 704270-275, 704/260, 704/235, 704/251, 704/246, 379/67.1, 379/88
• US Class Current: 704/246
• CPC Class Codes: G06Q 30/06 Buying, selling or leasing transactions
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

Methods and apparatus are described for effecting a computer transaction using speech as a primary input. The speech is captured using a speech recognition program. A context associated with the captured speech is determined. Where the context has been determined, the computer transaction is built based on the context and at least a portion of the captured speech. A representation of the computer transaction is presented to a human operator for verification. The computer transaction is effected upon verification by the human operator.

Claims

1. A method for effecting a computer transaction using speech as a primary input, comprising:

capturing the speech from a speaker using a speech recognition program;

determining a context associated with the captured speech;

where the context has been determined, building the computer transaction based on the context and at least a portion of the captured speech;

presenting a representation of the computer transaction to a human operator for verification, wherein said sneaker is distinct from said human operator; and

effecting the computer transaction upon verification by the human operator.

2. The method of claim 1 further comprising, where the context has not been determined, notifying the speaker that the computer transaction may not have been effected.

3. The method of claim 1 wherein capturing the speech comprises receiving the speech via a telephone network.

4. The method of claim 3 wherein capturing the speech further comprises receiving a call with an automatic call distribution system which routes the call to a particular service.

5. The method of claim 4 wherein at least one of a user identity and the particular service is identified using source information associated with the call.

6. The method of claim 1 further comprising verifying an identity of the speaker by comparing a first voice file corresponding to the captured speech with a previously generated voice file corresponding to the speaker.

7. The method of claim 6 further comprising notifying the speaker that the computer transaction may not have been effected where the identity of the speaker has not been verified.

8. The method of claim 1 wherein capturing the speech comprises receiving the speech via a wide area network.

9. The method of claim 8 wherein the wide area network comprises the World Wide Web.

10. The method of claim 1 wherein capturing the speech occurs without interruption by prompting.

11. The method of claim 1 wherein the speech comprises natural speech with flexible syntax.

12. The method of claim 1 wherein the speech recognition program is a speaker-independent speech recognition program.

13. The method of claim 1 further comprising preprocessing the speech to ameliorate effects of impairments impeding voice recognition.

14. The method of claim 13 wherein the impairments include at least one of low gain, poor signal-to-noise ratio, spurious noise, and ambient noise.

15. The method of claim 1 wherein determining the context comprises scanning a voice file associated with the captured speech for primary keywords.

16. The method of claim 15 wherein determining the context further comprises scanning the voice file for secondary key words with reference to the context established with reference to the primary keywords.

17. The method of claim 16 wherein the context corresponds to a limited keyword vocabulary, the secondary keywords being determined with reference to the limited keyword vocabulary.

18. The method of claim 16 wherein deterring the context further comprises scanning the voice file for non-keywords.

19. The method of claim 18 wherein the non-keywords are determined using a speaker-dependent voice recognition program.

20. The method of claim 1 wherein building the computer transaction comprises populating a predefined template according to the context and the at least a portion of the captured speech.

21. The method of claim 20 wherein building the computer transaction further comprises populating the predefined template according to at least one of an identity of the speaker and a transaction type.

22. The method of claim 21 wherein the identity of the speaker is determined from at least one of the speech and source information associated with transmission of the speech.

23. The method of claim 21 wherein the transaction type is determined from at least one of the speech and source information associated with transmission of the speech.

24. The method of claim 20 further comprising selecting the predefined template based on at least one of an identity of the speaker and a transaction type.

25. The method of claim 1 wherein presenting the representation of the computer transaction to the human operator comprises presenting a representation of a voice file associated with the captured speech to the human operator for comparison with the representation of the computer transaction.

26. The method of claim 25 wherein presenting the representation of the voice file to the human operator comprises presenting an audio playback of the captured speech associated with the voice file.

27. The method of claim 25 wherein presenting the representation of the voice file to the human operator comprises presenting a text representation of the captured speech associated with the voice file.

28. The method of claim 1 further comprising revising the computer transaction according to at least one instruction received from the human operator.

29. The method of claim 1 wherein effecting the computer transaction comprises transmitting the computer transaction to a target system for processing in response to at least one instruction received from the human operator.

30. The method of claim 1 wherein processing of the computer transaction subsequent to capturing of the speech occurs after a connection with the speaker is terminated.

31. A computer program product comprising a computer readable medium having computer program instructions stored therein for performing the method of claim 1.

32. A method for effecting a computer transaction using speech as a primary input, comprising:

capturing the speech from a speaker using a speech recognition program;

scanning a voice file associated with the captured speech for primary keywords;

determining a context using the primary keywords, the context comprising a limited keyword vocabulary;

scanning the voice file for secondary key words with reference to the limited keyword vocabulary;

building the computer transaction by populating a predefined template according to the context and the at least a portion of the captured speech;

presenting a representation of the computer transaction to a human operator,

wherein said human operator is distinct from said speaker;

presenting a representation of the voice file to the human operator for comparison with the representation of the computer transaction; and

transmitting the computer transaction to a target system for processing in response to at least one instruction received from the human operator.

33. A computer program product comprising a computer readable medium having computer program instructions stored therein for performing the method of claim 32.

34. A method for effecting a computer transaction using speech as a primary input, comprising:

capturing the speech in a communication from a speaker using a speech recognition program;

determining at least one of an identity of the speaker and a transaction type from at least one of source information associated with the communication and the speech;

notifying the speaker where at least one of the identity and the transaction type is not determined;

scanning a voice file associated with the captured speech for primary keywords;

determining a context using the primary keywords, the context comprising a limited keyword vocabulary;

scanning the voice file for secondary key words with reference to the limited keyword vocabulary;

building the computer transaction by populating a predefined template according to the context and the at least a portion of the captured speech;

presenting a representation of the computer transaction to a human operator,

wherein said human operator is distinct from said speaker;

presenting a representation of the voice file to the human operator for comparison with the representation of the computer transaction; and

transmitting the computer transaction to a target system for processing in response to at least one instruction received from the human operator.

35. A computer program product comprising a computer readable medium having computer program instructions stored therein for performing the method of claim 34.