Apparatus and method for automatic dissection of segmented audio signals

• US 7,962,330 B2
• Filed: 11/10/2004
• Issued: 06/14/2011
• Est. Priority Date: 11/12/2003
• Status: Expired due to Fees

• Current Assignee: Sony Deutschland GmbH (Sony Group Corp.)
• Original Assignee: Sony Deutschland GmbH (Sony Group Corp.)
• Inventors: Marasek, Krzysztof, Lam, Yin Hay, Tato, Raquel, Goronzy, Silke, Kompe, Ralf, Kemp, Thomas
• Primary Examiner(s): Wozniak; James S
• Assistant Examiner(s): Baker; Matthew H
• Application Number:
  US10/985,451
• Publication Number:
  US 20050160449A1
• Time in Patent Office: 2,407 Days
• Field of Search: 704240-250
• US Class Current: 704/201
• CPC Class Codes: G10L 15/00 Speech recognition G10L17/00 takes precedence
• CPC Class Codes: G10L 25/00 Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 muting semiconductor-based amplifiers when some special characteristics of a signal are sensed by a speech detector, e.g. sensing when no signal is present, H03G3/34

Abstract

An apparatus for automatic dissection of segmented audio signals, wherein at least one information signal for identifying programs included in said audio signals and for identifying contents included in said programs. Content detection device detects programs and contents belonging to the respective programs in the information signal. Program weighting device weights each program includes in the information signal based on the contents of the respective program detected by the content detection device. Program ranking device indentifies programmers of the same category and ranking said programs based on a weighting result for each program provided by the program weighting device.

Claims

1. An apparatus for automatic dissection, weighting and ranking of segmented audio signals, the apparatus comprising:

a content detection means that is configured to receive a segmented audio signal, to automatically detect programs in the segmented audio signal and to automatically segment the programs into contents, the contents being important events comprised in the programs;

a program weighting means for weighting each program comprised in the audio signal based on the contents of the respective program detected by the content detection means;

a viewing statistic database comprising raw audio signals previously selected by a user and/or acoustic features extracted from audio signals previously selected by a user, wherein the program weighting means uses information stored in the viewing statistic database for weighting each program;

a program ranking means for identifying programs of the same category and ranking the programs based on a weighting result for each program provided by the program weighting means; and

a cross category ranking means for identifying weighted programs which belong to different categories, normalizing the weighted programs and ranking the normalized and weighted programs based on a normalization result,

the normalization being performed by the cross category ranking means according to the following formula;

Normalized_Ranking_Score1(c,i)=function(W(1)I(1)/F(c,1)+W(2)I(2)/F(c,2)+ . . . +W(N)I(N)/F(c,N)), with

1, 2, . . . , N being an index identifying the contents of a respective program,

W(N) being a weighting vector comprising weighting factors for the contents 1 to N,

I(N) being a content vector comprising the distribution of contents 1-N in the respective program,

c being a category index,

i being a program index in category c, and

F(c,N) being a normalization factor corresponding to content N in category c, wherein the normalization factor F(c,N) is defined as the average duration or occurrence of content N for programs of category c.

2. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 1, wherein at least one information signal for identifying programs included in the audio signals and for identifying contents included in the programs is provided.

3. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 1, wherein:

the content detection means detects a distribution of the contents in each program; and

the program weighting means uses the distribution of the contents in each program detected by the content detection means for weighting the respective program.

4. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 1, wherein:

the content detection means detects for each content of a respective program a total duration and/or a number of occurrences and/or a duration normalized by a total duration of the respective program; and

the program weighting means uses the total duration of each content and/or the number of occurrences of each content and/or the duration of each content normalized by the total duration of the respective program detected by the content detection means for weighting the respective program.

5. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 1, wherein,

individual weighting factors for each content are provided,

wherein the program weighting means weights each content comprised in the respective program with the respective individual weighting factor for weighting the respective program.

6. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 5, wherein the weighting factors are generated heuristically.

7. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 5, wherein the weighting factors are defined manually by a user.

8. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 5, wherein:

the content detection means generates a content vector for identifying the contents comprised in each program; and

the weighting factors for each content are provided as a weighting vector,

wherein the program weighting means weights each program by multiplication of the content vector and the weighting vector.

9. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 1, wherein

the information stored in the viewing statistic database is analysed by speech recognition means to detect contents included in the information,

wherein the program weighting means weights each contents of a respective program detected by the content detection means based on the frequency of occurrence of the content in the information stored in the viewing statistic database.

10. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 1, wherein the apparatus for automatic dissection, weighting and ranking of segmented audio signals further comprises:

an user interface comprising a display for displaying information with respect to the ranked programs to an user.

11. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 10, wherein the user interface further comprises:

input means for allowing a manual input of weighting factors for the program weighting means by an user and/or requesting a certain type of category and/or a certain number of ranked programs to be displayed.

12. The apparatus for automatic dissection, weighting and ranking of segmented audio signals according to claim 1, wherein,

the audio signal is part of a video signal, the video signal being composed of at least an audio signal, a picture signal and an information signal,

wherein the information signal comprises information regarding the programs and respective contents comprised in the video signal.

13. A method for automatic dissection, weighting and ranking of segmented audio signals comprising:

supplying segmented audio signals to a detecting device;

automatically detecting by the detecting device programs in the segmented audio signals and automatically segmenting the programs into contents, the contents being important events comprised in the programs;

weighting each program comprised in the audio signal based on the detected contents of the respective program by using information stored in a viewing statistic database for weighting each program, wherein the information stored in the viewing statistic database comprises raw audio signals previously selected by a user and/or acoustic features extracted from audio signals previously selected by a user;

identifying programs of the same category;

ranking the programs by a program ranking device based on a weighting result for each program provided by weighting each program;

identifying weighted programs which belong to different categories;

normalizing the weighted programs; and

ranking the normalized and weighted programs,

the normalization being performed according to the following formula;

Normalized_Ranking_Score1(c,i)=function(W(1)I(1)/F(c,1)+W(2)I(2)/F(c,2)+ . . . +W(N)I(N)/F(c,N), with

1, 2, . . . , N being an index identifying the contents of a respective program,

W(N) being a weighting vector comprising weighting factors for the contents 1 to N,

I(N) being a content vector comprising the distribution of contents 1-N in the respective program,

c being a category index,

i being a program index in category c, and

F(c,N) being a normalization factor corresponding to content N in category c, wherein the normalization factor F(c,N) is defined as the average duration or occurrence of content N for programs of category c.

14. The method for automatic dissection, weighting and ranking of segmented audio signals according to claim 13, wherein the detecting programs and contents belonging to the programs in a segmented audio signal further comprises:

detecting a distribution of the contents in each program,

wherein the distribution of the contents in each program is used in the weighting each program.

15. The method for automatic dissection, weighting and ranking of segmented audio signals according to claim 13, wherein:

the detecting programs and contents belonging to the programs in a segmented audio signal comprises detection for each content of a respective program a total duration and/or a number of occurrences and/or a duration normalized by a total duration of the respective program; and

the total duration of each content and/or the number of occurrences of each content and/or the duration of each content normalized by the total duration of the respective program is used in the weighting each program.

16. The method for automatic dissection, weighting and ranking of segmented audio signals according to claim 13, wherein individual weighting factors for each content which are generated heuristically or defined manually by a user are used in the weighting each program.

17. The method for automatic dissection, weighting and ranking of segmented audio signals according to claim 13, wherein the method further comprises:

analyzing the information stored in the viewing statistic database by the speech recognition means to detect contents included in the information,

wherein the weighting each program comprises weighting each contents of a respective program based on the frequency of occurrence of the content in the information stored in the viewing statistic database.

18. The method for automatic dissection, weighting and ranking of segmented audio signals according to claim 13, wherein the method further comprises:

displaying information with respect to the ranked programs to an user.

19. The software product comprising a series of state elements which are adapted to be processed by a data processing means of a mobile terminal such that a method according to claim 13 may be executed thereon.

20. Apparatus for automatic dissection, weighting and ranking of segmented audio signals, the apparatus comprising:

a content detection device configured to receive a segmented audio signal and to automatically detect programs in the segmented audio signal and to automatically segment the programs into contents, the contents being important events comprised in the programs;

a program weighting device configured to weight each program comprised in the audio signal based on the contents of the respective program detected by the content detection device;

a viewing statistic database comprising raw audio signals previously selected by a user and/or acoustic features extracted from audio signals previously selected by a user, wherein the program weighting device uses information stored in the viewing statistic database for weighting each program;

a program ranking device configured to identifying programs of the same category and ranking the programs based on a weighting result for each program provided by the program weighting device; and

a cross category ranking device configured to identifying weighted programs which belong to different categories, normalizing the weighted programs and ranking the normalized and weighted programs based on a normalization result,

the normalization being performed by the cross category ranking device according to the following formula;

Normalized_Ranking_Score1(c,i)=function(W(1)I(1)/F(c,1)+W(2)I(2)/F(c,2)+ . . . +W(N)I(N)/F(c,N)), with

1, 2, . . . , N being an index identifying the contents of a respective program,

W(N) being a weighting vector comprising weighting factors for the contents 1 to N,

I(N) being a content vector comprising the distribution of contents 1-N in the respective program,

c being a category index,

i being a program index in category c, and

F(c,N) being a normalization factor corresponding to content N in category c, wherein the normalization factor F(c,N) is defined as the average duration or occurrence of content N for programs of category c.

21. An apparatus for automatic dissection, weighting and ranking of segmented audio signals, the apparatus comprising:

a content detection device that is configured to receive a segmented audio signal, to automatically detect programs in the segmented audio signal and to automatically segment the programs into contents, the contents being important events comprised in the programs, and to classify the contents;

a program weighting device for weighting and normalizing each program comprised in the audio signal based on the classification of the contents of the respective program detected by the content detection device; and

a program ranking device for identifying programs of a same category and ranking the programs based on a weighting result for each program provided by the program weighting device,

the program weighting device weighting and normalizing according to the following formula;

Normalized_Ranking_Score1(c,i)=function(W(1)I(1)/F(c,1)+W(2)I(2)/F(c,2)+ . . . +W(N)I(N)/F(c,N)), with

1, 2, . . . , N being an index identifying the contents of a respective program,

W(N) being a weighting vector comprising weighting factors for the contents 1 to N,

I(N) being a content vector comprising the distribution of contents 1-N in the respective program,

c being a category index,

i being a program index in category c, and

F(c,N) being a normalization factor corresponding to content N in category c, wherein the normalization factor F(c,N) is defined as the average duration or occurrence of content N for programs of category c.

22. A method for automatic dissection, weighting and ranking of segmented audio signals comprising:

supplying segmented audio signals to a detecting device;

automatically detecting by the detecting device programs in the segmented audio signal and automatically segmenting the programs into contents, the contents being important events comprised in the programs, and classifying the contents;

weighting and normalizing each program comprised in the audio signal based on the classification of the detected contents of the respective program; and

ranking the programs by a program ranking device based on a weighting result for each program provided by the step of weighting each program,

the weighting and normalizing performed according to the following formula;

Normalized_Ranking_Score1(c,i)=function(W(1)I(1)/F(c,1)+W(2)I(2)/F(c,2)+ . . . +W(N)I(N)/F(c,N)), with

1, 2, . . . , N being an index identifying the contents of a respective program,

W(N) being a weighting vector comprising weighting factors for the contents 1 to N,

I(N) being a content vector comprising the distribution of contents 1-N in the respective program,

c being a category index,

i being a program index in category c, and

F(c,N) being a normalization factor corresponding to content N in category c, wherein the normalization factor F(c,N) is defined as the average duration or occurrence of content N for programs of category c.