Interactive voice response system

• US 6,944,592 B1
• Filed: 09/25/2000
• Issued: 09/13/2005
• Est. Priority Date: 11/05/1999
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Pickering, John Brian
• Primary Examiner(s): Lerner, Martin
• Application Number:
  US09/670,046
• Time in Patent Office: 1,814 Days
• Field of Search: 704/231, 704/235, 704/251, 704/254, 704/257, 704/270, 704/275, 379/88.01, 379/88.04, 379/88.16
• US Class Current: 704/251
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: H04M 3/4936 Speech interaction details speech recognition per se G10L15/00

Abstract

An interactive voice response system which statistically analyses word usage in speech recognition results to select prompts for use in the interaction. A received voice signal is converted to text and calculating factors such as context and task word ratios and word rate. These factors are used to make dialogue decisions as to the whether to use expert, intermediate, or novice prompts depending on whether the factor falls below, inside or above a threshold range.

Claims

1. A method of managing dialogue in an interactive voice response system (IVR) comprising the steps of:

receiving a voice signal from a caller to the IVR;

converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text, wherein the caller type is based on the caller'"'"'s level of competence for using the IVR system; and

using the caller type to make a dialogue decision, wherein making the dialog decision comprises determining a type of automatic response to provide to the caller.

2. The method as in claim 1, wherein the using step further comprises the step of making the dialogue decision on which prompt to present next to the caller as a function of the estimated caller type.

3. The method as in claim 1, wherein one word type is made up of words relevant to the context of the dialogue.

4. The method as in claim 3, wherein another word type is made up of words relevant to the context of a task specific part of the dialogue.

5. A method of managing dialogue in an interactive voice response system (IVR) comprising the steps of:

receiving a voice signal from a caller to the IVR;

converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text, wherein the caller type is based on a word ratio; and

using the caller type to make a dialogue decision.

6. A method of managing dialogue in an interactive voice response system (IVR) comprising the steps of:

receiving a voice signal from a caller to the IVR;

converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text;

using the caller type to make a dialogue decision;

locating a word type flag in a dictionary for each word in the text;

incrementing a word type counter depending on the flag; and

estimating the caller type using the word type counter.

7. The method as in claim 6, further comprising the steps of:

identifying more than one word type from the word type flag; and

incrementing more than one word type counter.

8. The method as in claim 7, further comprising the step of identifying from each bit of the word type flag whether a word is a certain type, each bit of the word type flag being associated with one word type.

9. A method of managing dialogue in an interactive voice response system (IVR) comprising the steps of:

receiving a voice signal from a caller to the IVR;

converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text; and

using the caller type to make a dialogue decision whether to use expert or novice prompts to the caller, wherein the using step further comprises the step of making the dialogue decision on which prompt to present next to the caller as a function of the estimated caller type, wherein the dialogue decision whether to use expert or novice prompts is made depending on whether the caller type is above or below a threshold value.

10. A method of managing dialogue in an interactive voice response system (IVR) comprising the steps of:

receiving a voice signal from a caller to the IVR;

converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text; and

using the caller type to make a dialogue decision on which prompt to present next to the caller, wherein the using step further comprises the step of making the dialogue decision on which prompt to present next to the caller as a function of the estimated caller type, wherein a decision whether to use expert, intermediate, or novice prompts is made depending on whether the caller type falls below, inside or above a threshold range.

11. A method of managing dialogue in an interactive voice response system (IVR) comprising the steps of:

receiving a voice signal from a caller to the IVR;

converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text, wherein one word type is made up of words relevant to the context of the dialogue, wherein a number of words per second spoken by the caller is used together with the context type to give a further approximation as to a competence of the caller; and

using the caller type to make a dialogue decision.

12. The method as in claim 11, wherein the context ratio and the number of words per second are weighted and combined to give an overall factor having a threshold value or range to decide the competence of the caller.

13. A system for managing dialogue in an interactive voice response system (IVR) comprising:

the IVR receiving a voice signal from a caller;

an automatic speech recognition system (ASR) converting the voice signal to text;

a lexical analyzer estimating a caller type based on the number of words of a particular type within the text wherein the caller type is based on the caller'"'"'s level of competence for using the IVR system; and

a prompt generator using the caller type to make a dialogue decision wherein making the dialog decision comprises determining a type of automatic verbal response to provide to the caller.

14. The system as in claim 13, wherein the prompt generator makes the dialogue decision on which prompt to present next to the caller as a function of the estimated caller type.

15. The system as in claim 13, wherein one word type is made up of words relevant to the context of the dialogue.

16. The system as in claim 15, wherein another word type is made up of words relevant to the context of a task specific part of the dialogue.

17. A system for managing dialogue in an interactive voice response system (IVR) comprising:

the IVR receiving a voice signal from a caller;

an automatic speech recognition system (ASR) converting the voice signal to text;

a lexical analyzer estimating a caller type based on the number of words of a particular type within the text, wherein the caller type is based on a word ratio; and

a prompt generator using the caller type to make a dialogue decision.

18. A system for managing dialogue in an interactive voice response system (IVR) comprising:

the IVR receiving a voice signal from a caller;

an automatic speech recognition system (ASR) converting the voice signal to text;

a lexical analyzer estimating a caller type based on the number of words of a particular type within the text;

a prompt generator using the caller type to make a dialogue decision;

a search engine locating a word type flag in a dictionary for each word in the text;

account engine incrementing a word type counter depending on the flag; and

the lexical analyzer estimating the caller type using the word type counter.

19. The system as in claim 18, further comprising:

the search engine identifying more than one word type from the word type flag; and

the count engine incrementing more than one word type counter.

20. The system as in claim 19, further comprising the search engine identifying from each bit of the word type flag whether a word is a certain type, each bit of the word type flag being associated with one word type.

21. A system for managing dialogue in an interactive voice response system (IVR) comprising:

the IVR receiving a voice signal from a caller;

an automatic speech recognition system (ASR) converting the voice signal to text;

a lexical analyzer estimating a caller type based on the number of words of a particular type within the text; and

a prompt generator using the caller type to make a dialogue decision, wherein the prompt generator makes the dialogue decision on which prompt to present next to the caller as a function of estimated caller type, wherein the dialogue decision whether to use expert or novice prompts is made depending on whether the caller type is above or below a threshold value.

22. A system for managing dialogue in an interactive voice response system (IVR) comprising:

the IVR receiving a voice signal from a caller;

an automatic speech recognition system (ASR) converting the voice signal to text;

a lexical analyzer estimating a caller type based on the number of words of a particular type within the text, and

a prompt generator using the caller type to make a dialogue decision whether to use expert or novice prompts, wherein the prompt generator makes the dialogue decision on which prompt to present next to the caller as a function of the estimated caller type, wherein a decision whether to use expert, intermediate, or novice prompts is made depending on whether the caller type falls below, inside or above a threshold range.

23. A system for managing dialogue in an interactive voice response system (IVR) comprising:

the IVR receiving a voice signal from a caller;

an automatic speech recognition system (ASR) converting the voice signal to text;

a lexical analyzer estimating a caller type based on the number of words of a particular type within the text, wherein one word type is made up of words relevant to the context of the dialogue, wherein a number of words per second spoken by the caller is used together with the context type to give a further approximation as to a competence of the caller; and

a prompt generator using the caller type to make a dialogue decision.

24. The system as in claim 23, wherein the context ratio and the number of words per second are weighted and combined to give an overall factor having a threshold value or range to decide the competence of the caller.

25. A computer program product, stored on a computer-readable storage medium, for executing computer program instructions to carry out the steps of a method of managing dialogue in an interactive voice response system (IVR) comprising the program steps of:

in response to receipt of a voice signal from a caller to the IVR, converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text wherein the caller type is based on the caller'"'"'s level of competence for using the IVR system; and

using the caller type to make a dialogue decision wherein making the dialog decision comprises determining a type of automatic verbal response to provide to the caller.

26. The computer program product as in claim 25, wherein the using program step further comprises the program step of making the dialogue decision on which prompt to present next to the caller as a function of the estimated caller type.

27. The computer program product as in claim 25, wherein one word type is made up of words relevant to the context of the dialogue.

28. The computer program product as in claim 27, wherein another word type is made up of words relevant to the context of a task specific part of the dialogue.

29. A computer program product, stored on a computer-readable storage medium, for executing computer program instructions to carry out the steps of a method of managing dialogue in an interactive voice response system (IVR) comprising the program steps of:

in response to receipt of a voice signal from a caller to the IVR, converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text, wherein the caller type is based on a word ratio; and

using the caller type to make a dialogue decision.

30. A computer program product, stored on a computer-readable storage medium, for executing computer program instructions to carry out the steps of a method of managing dialogue in an interactive voice response system (IVR) comprising the program steps of:

in response to receipt of a voice signal from a caller to the IVR, converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text;

using the caller type to make a dialogue decision;

locating a word type flag in a dictionary for each word in the text;

incrementing a word type counter depending on the flag; and

estimating the caller type using the word type counter.

31. The computer program product as in claim 30, further comprising the program steps of:

identifying more than one word type from the word type flag; and

incrementing more than one word type counter.

32. The computer program product as in claim 31, further comprising the program step of identifying from each bit of the word type flag whether a word is a certain type, each bit of the word type flag being associated with one word type.

33. A computer program product, stored on a computer-readable storage medium, for executing computer program instructions to carry out the steps of a method of managing dialogue in an interactive voice response system (IVR) comprising the program steps of:

in response to receipt of a voice signal from a caller to the IVR, converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text; and

using the caller type to make a dialogue decision whether to use expert or novice prompts, wherein the using program step further comprises the program step of making the dialogue decision on which prompt to present next to the caller as a function of the estimated caller type, wherein the dialogue decision whether to use expert or novice prompts is made depending on whether the caller type is above or below a threshold value.

34. A computer program product, stored on a computer-readable storage medium, for executing computer program instructions to carry out the steps of a method of managing dialogue in an interactive voice response system (IVR) comprising the program steps of:

in response to receipt of a voice signal from a caller to the IVR, converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text; and

using the caller type to make a dialogue decision on which prompt to present next to the caller, wherein the using program step further comprises the program step of making the dialogue decision on which prompt to present next to the caller as a function of the estimated caller type, wherein a decision whether to use expert, intermediate, or novice prompts is made depending on whether the caller type falls below, inside or above a threshold range.

35. A computer program product, stored on a computer-readable storage medium, for executing computer program instructions to carry out the steps of a method of managing dialogue in an interactive voice response system (IVR) comprising the program steps of:

in response to receipt of a voice signal from a caller to the IVR, converting the voice signal to text;

estimating a caller type based on the number of words of a particular type within the text, wherein one word type is made up of words relevant to the context of the dialogue, wherein a number of words per second spoken by the caller is used together with the context type to give a further approximation as to a competence of the caller;

using the caller type to make a dialogue decision.

36. The computer program product as in claim 35, wherein the context ratio and the number of words per second are weighted and combined to give an overall factor having a threshold value or range to decide the competence of the caller.