Information retrieval system, information retrieval method, and information retrieval program

• US 8,712,779 B2
• Filed: 03/19/2008
• Issued: 04/29/2014
• Est. Priority Date: 03/19/2007
• Status: Active Grant

• Current Assignee: NEC Asia Pacific Pte Limited (NEC Corporation)
• Original Assignee: NEC Corporation
• Inventors: Nagatomo, Kentaro
• Primary Examiner(s): Godbold, Douglas
• Application Number:
  US12/530,765
• Publication Number:
  US 20100114571A1
• Time in Patent Office: 2,232 Days
• Field of Search: 704/1, 704/7, 704/9, 704/10, 704/270, 704/270.1
• US Class Current: 704/270.1
• CPC Class Codes: G06F 16/433 using audio data
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

An information retrieval system comprises: a speech input unit for inputting speech; an information storage unit for storing information with which speech information, of a length with which text degree of similarity is computable, is associated as a retrieval tag; an information selection unit for comparing a feature of each spoken content item extracted from each item of said speech information, with a feature of spoken content extracted from said input speech, to select information with which speech information similar to input speech is associated. The system further comprises an output unit for outputting information selected by said information selection unit, as information associated with input speech.

Claims

1. An information retrieval system comprising:

a speech input unit for inputting speech;

an information storage unit for storing information with which speech information, of a length with which text degree of similarity is computable, is associated as a retrieval tag;

an information selection unit for comparing a feature of each spoken content item extracted from each item of said speech information, with a feature of spoken content extracted from said input speech, to select information with which speech information similar to input speech is associated;

an output unit for outputting information selected by said information selection unit, as information associated with input speech;

a speech recognition unit for converting a speech signal into text data;

a degree of similarity computation unit for computing degree of similarity between two or more speech recognition results generated by said speech recognition unit; and

an information registration unit for registering a speech recognition result obtained by said speech recognition unit or a text feature vector generated from said speech recognition result, as said speech information associated with information that is to be retrieved, in said information storage unit;

wherein

speech recognition of said input speech is performed by said speech recognition unit;

degree of similarity between a speech recognition result stored in advance as speech information in said information storage unit, and a speech recognition result of said input speech, is computed by said degree of similarity computation unit; and

said information selection unit selects information with which speech information having a high degree of similarity to input speech is associated, based on degree of similarity computed by said degree of similarity computation unit.

2. The information retrieval system according to claim 1, wherein said information selection unit selects said speech information that has a word set similar to a word set included in said input speech.

3. The information retrieval system according to claim 1, further comprising:

a text feature vector generation unit for computing a text feature vector from a speech recognition result generated by said speech recognition unit wherein

said degree of similarity computation unit is further to compute degree of similarity between two or more text feature vectors computed by said text feature vector generation unit;

wherein,

after performing speech recognition of said input speech by said speech recognition unit, respective text feature vectors are computed by said text feature vector generation unit, from a speech recognition result of said input speech and a speech recognition result stored in advance as speech information in said information storage unit;

degree of similarity between each of said text feature vectors is computed by said degree of similarity computation unit; and

said information selection unit selects information with which speech information having a high degree of similarity to input speech is associated, based on degree of similarity computed by said degree of similarity computation unit.

4. The information retrieval system according to claim 3, wherein a speech recognition result stored in advance in said information storage unit is a speech recognition result generated by a speech recognition unit equivalent to said speech recognition unit.

5. The information retrieval system according to claim 1, further comprising:

a text feature vector generation unit for computing a text feature vector from a speech recognition result generated by said speech recognition unit wherein

said degree of similarity computation unit is further to compute degree of similarity between two or more text feature vectors computed by said text feature vector generation unit;

wherein,

after performing speech recognition of said input speech by said speech recognition unit, a text feature vector is computed by said text feature vector generation unit;

degree of similarity between a text feature vector stored in advance as speech information in said information storage unit and a text feature vector computed from said input speech is computed by said degree of similarity computation unit; and

said information selection unit selects information with which speech information having a high degree of similarity to input speech is associated, based on degree of similarity computed by said degree of similarity computation unit.

6. The information retrieval system according to claim 5, wherein a text feature vector stored in advance in said information storage unit is a text feature vector generated by a text feature vector generation unit equivalent to said text feature vector generation unit.

7. The information retrieval system according to claim 1, wherein, in said information storage unit, one or a plurality of different items of speech information can be associated with one item of information.

8. The information retrieval system according to claim 1, wherein, in said information storage unit, one or a plurality of different items of information can be associated with one item of speech information.

9. The information retrieval system according to claim 1, wherein said information registration unit for registering information to be retrieved and said speech information, in said information storage unit.

10. The information retrieval system according to claim 1, wherein said information registration unit receives, and registers in said information storage unit, input of speech information newly associated with information selected by said information selection unit.

11. The information retrieval system according to claim 1, wherein information stored in said information storage unit is text data.

12. The information retrieval system according to claim 1, wherein information stored in said information storage unit is any one selected from:

corresponding history text that is an abstract of content of a telephone call response;

a video of an explanation or catalog related to a specific product that is a topic of telephone call speech;

a URL of a website referred to in a telephone call; and

a video in which user behavior during a telephone call response is recorded.

13. The information retrieval system according to claim 12, wherein said information registration unit records, in said information storage unit, arbitrary information outside of information selected by said information selection unit, among information referred to by a telephone reception agent during a telephone call response, as information associated with content of conversation speech between a customer and said telephone reception agent.

14. The information retrieval system according to claim 1, wherein speech information stored in said information storage unit is conversation speech between a customer and a telephone reception agent.

15. An information retrieval system comprising:

a speech input unit for inputting speech;

an information storage unit for storing information with which speech information, of a length with which text degree of similarity is computable, is associated as a retrieval tag;

an information selection unit for comparing a feature of each spoken content item extracted from each item of said speech information, with a feature of spoken content extracted from said input speech, to select information with which speech information similar to input speech is associated;

an output unit for outputting information selected by said information selection unit, as information associated with input speech;

a speech recognition unit for converting a speech signal into text data;

a degree of similarity computation unit for computing degree of similarity between two or more speech recognition results generated by said speech recognition unit; and

an information registration unit for registering a speech recognition result obtained by said speech recognition unit or a text feature vector generated from said speech recognition result, as said speech information associated with information that is to be retrieved, in said information storage unit;

wherein,

speech recognition of each of a speech signal stored as said speech information in said information storage unit and said input speech is performed by said speech recognition unit, respectively;

degree of similarity between each of said speech recognition results is computed by said degree of similarity computation unit; and

said information selection unit selects information with which speech information having a high degree of similarity to input speech is associated, based on degree of similarity computed by said degree of similarity computation unit.

16. The information retrieval system according to claim 15, further comprising:

a text feature vector generation unit for computing a text feature vector from a speech recognition result generated by said speech recognition unit wherein

said degree of similarity computation unit is further to compute degree of similarity between two or more text feature vectors computed by said text feature vector generation unit;

wherein,

after performing speech recognition of each of said input speech and a speech signal stored as speech information in said information storage unit by said speech recognition unit, text feature vectors of each thereof are computed by said text feature vector generation unit, degree of similarity between each of said text feature vectors is computed by said degree of similarity computation unit; and

said information selection unit selects information with which speech information having a high degree of similarity to input speech is associated, based on degree of similarity computed by said degree of similarity computation unit.

17. The information retrieval system according to claim 16, wherein

said speech recognition unit divides said input speech into blocks of arbitrary size, and sequentially outputs a speech recognition result for each block;

and

for each speech recognition result output, said text feature vector generation unit generates a text feature vector for a speech recognition result of each of said blocks; and

in addition, said degree of similarity computation unit re-computes degree of similarity of said speech information and speech recognition results of all blocks that have been outputted; and

said information selection unit re-selects information with which speech information having a high degree of similarity to input speech is associated, based on degree of similarity that has been re-computed.

18. The information retrieval system according to claim 16, further comprising:

a buffer unit for holding a text feature vector generated by said text feature vector generation unit; and

a feature vector selection unit for selecting a text feature vector from said buffer unit according to a prescribed feature selection rule, to be inputted to said degree of similarity computation unit;

wherein

said degree of similarity computation unit uses a text feature vector selected by said feature vector selection unit to compute said respective degree of similarity.

19. The information retrieval system according to claim 15, wherein

said speech recognition unit divides said input speech into blocks of arbitrary size, and sequentially outputs a speech recognition result for each block;

and

for each speech recognition result output, said degree of similarity computation unit re-computes degree of similarity of said speech information and speech recognition results of all blocks that have been outputted; and

said information selection unit re-selects information with which speech information having a high degree of similarity to input speech is associated, based on degree of similarity that has been re-computed.

20. The information retrieval system according to claim 16, further comprising:

a buffer unit for holding a speech recognition result obtained by said speech recognition unit; and

a feature selection unit for selecting a speech recognition result from said buffer unit according to a prescribed feature selection rule, to be inputted to said degree of similarity computation unit;

wherein

said degree of similarity computation unit uses a speech recognition result selected by said feature selection unit to compute said respective degree of similarity.

21. The information retrieval system according to claim 20, wherein said feature selection unit or said feature vector selection unit changes said prescribed feature selection rule, based on feedback from said information selection unit.

22. The information retrieval system according to claim 20, wherein said information registration unit for registering a text feature vector or a speech recognition result held in said buffer unit, as said speech information associated with information that is to be retrieved, in said information storage unit.

23. The information retrieval system according to claim 20, wherein said information registration unit for registering a set of text feature vectors or speech recognition results selected by said feature selection unit, as said speech information associated with information that is to be retrieved, in said information storage unit.

24. An information retrieval method comprising:

inputting speech, by hardware including a processor;

storing information, by said hardware, with which speech information, of a length with which text degree of similarity is computable, is associated as a retrieval tag;

comparing, by said hardware, a feature of each spoken content item extracted from each item of said speech information, with a feature of spoken content extracted from said input speech, to select information with which speech information similar to input speech is associated;

outputting, by said hardware, information selected, as information associated with input speech;

converting, by said hardware, a speech signal into text data; and

computing, by said hardware, degree of similarity between two or more speech recognition results generated by converting said speech signal into said text data;

wherein

degree of similarity between a speech recognition result is stored in advance as speech information; and

information with which speech information having a high degree of similarity to input speech is associated is selected based on degree of similarity.

25. A non-transitory computer-readable data storage medium storing an information retrieval program for execution in an information retrieval system comprising:

a speech input unit for inputting speech, an information storage unit for storing information with which speech information, of a length with which text degree of similarity is computable, is associated as a retrieval tag; and

an output unit for outputting information selected by said information selection unit as information associated with input speech;

said information retrieval program executing in said information retrieval system, processing of comparing a feature of each spoken content item extracted from respective items of speech information stored by said information storage unit, with a feature of spoken content extracted from said input speech; and

selecting information with which speech information similar to said input speech is associated, to be passed to said output unit;

said information retrieval program converting a speech signal into text data; and

said information retrieval program computing degree of similarity between two or more speech recognition results generated by converting said speech signal into said text data;

wherein

degree of similarity between a speech recognition result is stored in advance as speech information; and

information with which speech information having a high degree of similarity to input speech is associated is selected based on degree of similarity.