System and method for prediction of musical preferences

• US 7,075,000 B2
• Filed: 06/29/2001
• Issued: 07/11/2006
• Est. Priority Date: 06/29/2000
• Status: Expired due to Term

• Current Assignee: Pandora Media LLC (Liberty Media Corporation)
• Original Assignee: MusicGenome.Com Incorporated
• Inventors: Lehmann, Daniel, Gang, Dan
• Primary Examiner(s): Fletcher, Marlon T.
• Assistant Examiner(s): Warren, David S.
• Application Number:
  US10/239,992
• Publication Number:
  US 20030089218A1
• Time in Patent Office: 1,838 Days
• Field of Search: 84600-609, 705/10
• US Class Current: 84/600
• CPC Class Codes: G06F 16/637 Administration of user profiles, e.g. generation, initialization, adaptation or distribution
• CPC Class Codes: G06F 16/639 using playlists
• CPC Class Codes: G06F 16/683 using metadata automatically derived from the content
• CPC Class Codes: G06F 16/686 using information manually generated, e.g. tags, keywords, comments, title or artist information, time, location or usage information, user ratings
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G10H 1/0008 Associated control or indicating means
• CPC Class Codes: G10H 1/0058 Transmission between separate instruments or between individual components of a musical system G10H1/0083 takes precedence
• CPC Class Codes: G10H 2240/061 MP3, i.e. MPEG-1 or MPEG-2 Audio Layer III, lossy audio compression
• CPC Class Codes: G10H 2240/081 Genre classification, i.e. descriptive metadata for classification or selection of musical pieces according to style analysis genre G10H2210/036
• CPC Class Codes: G10H 2240/305 Internet or TCP/IP protocol use for any electrophonic musical instrument data or musical parameter transmission purposes
• CPC Class Codes: G10H 2250/151 Fuzzy logic
• CPC Class Codes: G10H 2250/311 Neural networks for electrophonic musical instruments or musical processing, e.g. for musical recognition or control, automatic composition or improvisation musical analysis G10H2210/031; neural networks G06N3/02
• CPC Class Codes: G11B 27/105 of operating discs
• CPC Class Codes: G11B 27/28 by using information signals recorded by the same method as the main recording G11B27/22 takes precedence
• CPC Class Codes: G11B 27/34 Indicating arrangements indicating means incorporated in magazine or cassette G11B23/046 and G11B23/0875; indicating measured values in general G01D

Abstract

A system and a method for predicting the musical taste and/or preferences of the user The present invention receives, on one hand, ratings of a plurality of songs from the user and/or other information about the taste of the user, and on the other hand information about the songs in the catalog from which recommendations are to be given The method then combines both types of information in order to determine the musical preferences of the user. These preferences are then matched to at least one musical selection, which is predicted to be preferred by the user.

Claims

1. A method for predicting a preference of a user for a media item from a group of media items, each of said group of media items having a set of internal characteristics, the method comprising:

analyzing said group of media items to obtain respective quantifications of said set of internal characteristics

generating a test selection of media items from said group said test selection comprising a plurality of media items;

inviting the user to provide a respective rating for each of said selected media items in said test selection

obtaining each rating from said user;

analyzing said obtained ratings over said test selection to determine preferences of said user for said internal characteristics, and

predicting the preference of the user for at least one other media item in said group based on respective internal characteristics of said at least one other media items and said determined preferences of said test selection; and

providing to said user a recommendation for a media item for which said prediction is favorable.

2. The method of claim 1, wherein said test selection comprises a plurality of media items in which said internal characteristics differ.

3. The method of claim 1, wherein a matching process is a neural network, such that said plurality of internal characteristics of said media items forms a binary vector and said neural network learns to match said binary vector to internal characteristics of each item of said user-rated media selection.

4. The method of claim 1, wherein a matching process is a map method for creating a matrix of pseudo-distances between each pair of media items, for comparing each rated media item to at least one other media item in said catalog.

5. The method of claim 4, wherein said map method is a selection map method, for comparing all characteristics of each pair of media items.

6. The method of claim 4, wherein said map method is a feature map method, for comparing a selected group of features of each pair of media items.

7. The method of claim 4, wherein said map method is a fingerprint map method, for comparing a signature of each pair of media items.

8. The method of claim 1, wherein a matching process is a rating method for rating each media item of said catalog and directly comparing each rating of the user to said rating for said media item of said catalog.

9. The method of claim 1, wherein information about each media item in said catalog is stored in a feature base.

10. The method of claim 1, wherein each media item is also characterized by genre.

11. The method of any of claim 1, wherein said media items are musical items.

12. The method of claim 11, wherein said quantifying said internal characteristics comprises carrying out an analysis of musical features for a respective musical item by a trained musician.

13. The method of claim 11, wherein said quantifying said internal characteristics comprises rating said musical items plurality of users.

14. The method of claim 11, wherein said musical items are automatically analyzed according to a fingerprint of the audio signal.

15. The method of claim 11, wherein said musical items and rating information are presented to the user through a network.

16. The method of claim 15, wherein said network is the Internet.

17. The method of claim 11, wherein said musical items and rating information are presented to the user through a stand alone station.

18. The method of claim 15, further comprising:

purchasing at least one recommended musical item by the user.

19. The method of claim 1, wherein said media items feature printed material.

20. The method of claim 1, wherein said media items feature visual material.

21. The method of claim 20, wherein said visual material includes video data.

22. The method of claim 1, wherein said media items are rated by a plurality of users.

26. The method of claim 1, further comprising a first step of decomposing a collection of media items into a plurality of separate sub-catalogs.

23. A method for predicting a preference of a user for a media selection, the method comprising:

obtaining an analysis of at least a portion of a catalog of media items according to a plurality of internal media characteristics of each media item in said portion by a group of raters to obtain a quantification of respective internal characteristics of ones of said media items;

generating at least one test selection from said catalog, using said rating analysis, said test selection comprising a plurality of selected media items;

inviting a user to rate each of said selected media items;

obtaining a respective user rating for each of said selected media items;

matching an analysis of said user ratings over said test selection to obtain user preference data regarding said internal media characteristics;

with said user preference data predicting the preference of the user for at least one further of the media items of the catalog having a different quantification of said internal media characteristics; and

using said prediction, recommending at least one predicted media item to the user.

24. A method for predicting a preference of a user for a media selection, the method comprising:

automatically analyzing at least a portion of a catalog of media items according to plurality of internal media characteristic of each media item of said portion to obtain a quantification of respective internal characteristics of ones of said media items;

generating at least one test selection from said catalog, using said media internal characteristic analysis, said test selection comprising a plurality of selected media items;

inviting the user to rate each of said selected media items and obtaining a respective resulting user rating for each of said selected media items;

matching an analysis of said user rating over said test selection to obtain user preference data regarding said media items using said at least one internal characteristic;

with said user preference data predicting the preference of the user for said media items; and

using said prediction, recommending at least one other of said media items to the user.

25. The method of claim 24, wherein said media items feature audio data, and said characteristic is a fingerprint of said audio data as an audio signal.