WO2006106825A1

WO2006106825A1 - Music composition search device, music composition search method, music composition search program, and information recording medium

Info

Publication number: WO2006106825A1
Application number: PCT/JP2006/306663
Authority: WO
Inventors: Yasuteru Kodama; Yasunori Suzuki; Takehiko Shioda; Satoshi Odagawa
Original assignee: Pioneer Corporation
Priority date: 2005-03-31
Filing date: 2006-03-30
Publication date: 2006-10-12
Also published as: JPWO2006106825A1; JP4459269B2

Abstract

There is provided a music composition search device capable of visually displaying characteristics of a search keyword itself and characteristics of a music composition searched by using the search keyword and enabling a user to change the search condition according to user’s preference. The music composition search device includes: music composition characteristic information database (3) for accumulating music composition characteristic information indicating the characteristics of lyrics; an input unit (9) for inputting a search word; a search word characteristic information database (4) for accumulating search word characteristic information indicating the characteristics of lyrics to be searched in accordance with a personal point indicated by the inputted search word; and a search processing unit (8) for generating comparison object information to be compared to the inputted search music composition characteristic information by converting the music composition characteristic information by using a conversion table T, comparing the inputted search music composition characteristic information to the generated comparison object information, and extracting a music composition corresponding to the music composition information corresponding to the comparison object information most similar to the inputted search music composition characteristic information. The search processing unit (8) modifies the conversion table T according to a modification instruction of the conversion table T.

Description

Song search apparatus, song search method, song search program, and information recording medium

[0001] The present application belongs to the technical field of a song search device, a song search method, a song search program, and an information recording medium, and more specifically, a song (singing) and a performance (prelude, accompaniment, interlude and follower), respectively. The same applies to the following.) A song search device and song search method for searching for one or a plurality of the songs from among a plurality of songs, and a song search program used for the song search and the song search program Belongs to the technical field of information recording media on which is recorded. Background art

[0002] In recent years, digital data for reproducing a large number of songs, such as in-vehicle navigation devices and home server devices, has been accumulated, and this favorite song has been selected and played. is there.

[0003] At this time, conventionally, as a first song search method, a part of a constituent word (phrase) included in a song to be reproduced is input as it is, and the constituent word is input. It has been common practice to search for and play songs that contain songs.

[0004] In addition, as a second song search method that reflects the user's subjectivity, each search keyword described in Patent Document 1 below (for example, "cheerful song", "song as a song") ) And each feature word (for example, “Brightness”, “Genki”), and a renewable sensitivity table is prepared, and the presence / absence of the feature related to each feature word is set to “1”. There was a search method in which a feature word list indicated by “0” was prepared, and when a user entered a desired search keyword, a search was made based on a plurality of sensibility tables and feature word lists that matched this.

[0005] On the other hand, as a display method of the sorted file group in a file management system that manages songs and general documents as files, a plurality of types of attached calo information are attached to each file and stored. In response to the instruction, the order of the files is determined (sorted) for one additional information, and the X coordinate of the file object corresponding to the file with the additional information is set to one additional information (for example, registration date and time). Associate with 'order' value, associate Y-coordinate with order / value of other additional information (eg match degree with search key), file There has been a display method as shown in Patent Document 2 below in which a plurality of file objects are displayed on one XY coordinate by associating the size of the object with a value of other additional information (for example, reference frequency).

Patent Document 1: JP 2003-132085 (FIGS. 1 to 3)

Patent Document 2: JP 2000-122770 (Fig. 2)

Disclosure of the invention

Problems to be solved by the invention

[0006] However, the conventional song search method and the search result display method described above only display how similar the search keyword used for the search itself is to the file itself. The concept of comparing the characteristics of the search keyword itself with the characteristics of a file such as a song searched by the search keyword has never been considered.

[0007] Therefore, the present application has been made in view of the above points, and one example of the problem is to visually display the characteristics of the search keyword itself input by the user and use the search keyword. The feature of the searched file can also be displayed visually, so that the similarity between the feature of the search keyword entered by the user and the feature of the extracted file can be grasped as an image and used. Provides a song search device and song search method that can change search conditions according to preference, a song search program used for the song search, and an information recording medium on which the song search program is recorded. There is to do. Means for solving the problem

[0008] In order to solve the above-mentioned problem, the invention according to claim 1 is characterized in that in the song search device for searching for one or a plurality of the songs from a plurality of songs, the characteristics of the lyrics included in the songs A song feature information storage means such as a song feature information database for storing the song feature information indicating at least each song in an identifiable manner, and a search word indicating the subject to be searched and a word indicating the subject A search term input means such as an input unit used to input the search term and any of the songs to be searched using the search term as appropriate for the subjectivity indicated by the input search term. A search word feature information database that stores search song feature information that at least shows the features of the included lyrics in an identifiable manner for each search word The search song feature information storage means such as the search song feature information to be compared with the input search song feature information which is the search song feature information corresponding to the input search word, and the stored song feature information A search processing unit or the like that compares the conversion means such as a search processing unit that is generated by conversion using conversion information such as a conversion table, the input search song feature information, and the generated comparison target information. And the song corresponding to the song feature information that is the generation source of the comparison target information that is most similar to the input search song feature information, based on the comparison result of the comparison means and the input search song feature information. An extraction unit such as a search processing unit that extracts the music corresponding to the searched search word, an instruction input unit such as an input unit used to input an instruction to change the conversion information, and the input Instructions Change means such as a search processing unit for changing the conversion information, and when the conversion information is changed, the conversion means uses the converted conversion information after the change to compare the music feature information. Configured to convert to information.

In order to solve the above problems, the invention according to claim 5 is a music feature information database or the like that stores music feature information that at least shows the characteristics of the lyrics included in the music in an identifiable manner for each of the songs. Included in any of the songs to be searched using the search word as being suitable for the subjectivity indicated by the search word comprising the song feature information storage means and the word indicating the song to be searched and indicating the subjectivity Search song feature information storage means such as a search word feature information database for storing the search song feature information indicating at least the feature of the lyrics to be identifiable for each search word, and a display means such as a display, A song search method executed in a song search device for searching for one or a plurality of the songs, the search word input step for inputting the search terms, and the input search terms A conversion step for generating comparison target information to be compared with input search song feature information, which is the search song feature information, by converting the stored song feature information using conversion information; and the input search A comparison step for comparing the song feature information with the generated comparison target information, and generation of the comparison target information most similar to the input search song feature information based on the comparison result in the comparison step. Extracting the music corresponding to the original music feature information as the music corresponding to the input search word; an instruction input step for inputting an instruction to change the conversion information; A change step of changing the conversion information based on an inputted instruction. In the conversion process, when the conversion information is changed, the conversion information after the change is used to change the conversion feature information. It is comprised so that conversion to the said comparison object information may be performed.

In order to solve the above-described problem, the invention according to claim 6 causes a computer to function as the music search device according to any one of claims 1 to 4.

[0011] In order to solve the above-described problem, the invention according to claim 7 is recorded with the program for searching for music according to claim 6 so as to be readable by the computer.

Brief Description of Drawings

FIG. 1 is a block diagram showing a schematic configuration of a music search device according to an embodiment.

FIG. 2 is a diagram illustrating a data structure of information stored in the song search device according to the embodiment; (a) is a diagram illustrating a data structure of song feature information; and (b) is a search word. It is a figure which illustrates the data structure of feature information.

FIG. 3 is a flowchart showing song search processing according to the embodiment.

FIG. 4 is a diagram illustrating a conversion table according to the embodiment.

FIG. 5 is a diagram showing a display example of a search word feature information figure and a song feature information figure.

FIG. 6 is a diagram illustrating the data structure of history information according to the embodiment, (a) is a diagram illustrating the data structure of match history information, and (b) is the data structure of non-match history information. It is a figure which shows an example.

FIG. 7 is a flowchart showing details of conversion table update processing.

FIG. 8 is a diagram (I) showing a specific example of conversion table update processing.

FIG. 9 is a diagram (II) showing a specific example of conversion table update processing.

FIG. 10 is a diagram showing another display example of a search word feature information figure and a song feature information figure. Explanation of symbols

[0013] 1 song input section

2 song database

3 Song feature information database

4 Search term feature information database

5 Sound feature information extraction unit 6 Component word information extraction unit

7 Song feature information generator

8 Search processing section

9 Input section

10 Song output section

11 History storage

12 display

20 song feature information

30 Search term feature information

40 Song feature information

50, 70 Search term feature information display graphic

51, 52, 53, 54, 55, 61, 62, 63, 64, 65, 71, 72, 73, 74, 75, 81, 82, 83, 84, 85

56, 66 keys

57, 67 arrows

60, 80 song feature information display figure

S song search device

T conversion table

TT correlation value

BEST MODE FOR CARRYING OUT THE INVENTION

Next, the best embodiment corresponding to the present application will be described with reference to the drawings. The embodiment described below is implemented when the present application is applied to a song search apparatus that accumulates a plurality of songs and searches for and outputs (plays back) any song according to a user's request. It is a form.

[0015] m body structure

First, the overall configuration and overall operation of the music search apparatus according to the embodiment will be described with reference to FIGS. 1 is a block diagram showing a schematic configuration of the song search device, and FIG. 2 is a diagram illustrating a data structure of information stored in the song search device. It is.

As shown in FIG. 1, a song search device S according to the embodiment includes a song input unit 1, a song database 2, a song feature information database 3 as song feature information storage means, and a search song feature information storage. Search word feature information database 4 as product means, sound feature information extraction section 5, constituent word information extraction section 6, song feature information generation section 7, conversion means, change means, comparison means, and extraction means A search processing unit 8, an input unit 9 as a search word input unit and an instruction input unit, a song output unit 10, a history storage unit 11, and a display 12 as a display unit.

At this time, the song database 2 stores and stores a plurality of songs as search objects to be searched by a song search process described later. Each song includes at least a song (singing) and a performance including a prelude, accompaniment, interlude and postlude.

Here, the input of the music to be stored in the music database 2 to the music database 2 is that the music information Ssg corresponding to the music is input to the music input unit 1 from the outside. In step 1, the music information Ssg is subjected to a format conversion process for storage in the music database 2, and the music information Ssg after the processing is input to the music database 2 and executed.

Next, song feature information corresponding to all songs stored in the song database 2 is stored in the song feature information database 3 so as to be identifiable for each song.

[0020] Here, the song feature information is stored in the song feature information database 3 corresponding to each song stored in the song database 2! It is the information that characterizes each song and performance.

Next, the music feature information will be specifically described with reference to FIG. 1 and FIG. 2 (a).

The above song feature information is obtained when a new song is input to song database 2 as song information Ssg.

Then, it is newly generated as corresponding to the song, and the generated song feature information is newly registered and stored in the song feature information database 3.

Here, when a new song is accumulated in the song database 2, the song information Ssg corresponding to the song is read from the song database 2 as shown in FIG. And output to the constituent word feature information extraction unit 6. [0023] Then, the sound feature information extraction unit 5 extracts a plurality of parameters indicating the acoustic features of the song from the song information Ssg and outputs the parameters to the song feature information generation unit 7 as the sound feature information Sav.

At this time, the plurality of parameters included in the sound feature information Sav include, for example, the speed of the song (BPM (Beat Per Minutes)), the maximum of the song, as shown in the right of FIG. Output level (maximum volume), average output level of the song (average volume), chords contained in the song, beat level of the song (ie, signal level (magnitude) of the beat component of the song), Key (key such as C major or A minor).

In parallel with this, the constituent word information extraction unit 6 extracts a song (lyrics) included in the song from the song information Ssg, and is preset in the extracted song. A search is made to determine whether or not a word (phrase; hereinafter simply referred to as a constituent word) is included, and the search result (whether it is a constituent word included in the song) is determined for each constituent. The constituent word information Swd shown for each word is generated and output to the song feature information generating unit 7.

[0026] At this time, the constituent word information Swd is included in the songs constituting the powerful songs such as "Ai", "Umi", "Omoi" or "Kibo", which are preset constituent words. For each song, if the constituent word is included in the song, the constituent word feature information Swd for that constituent word is “1”. On the other hand, if the constituent word is not included in the song, the value of the constituent word feature information Swd is “0” according to the constituent word. More specifically, for example, as shown in the left of FIG. 2 (a), in the song with the song number “0” in the song database 2, the constituent word “eye” is included in the song song. The words “Umi”, “Omoy” and “Kibo” are not included in the song. Similarly, for the song with the song number “1” in song data base 2, the constituent words “Ai”, “Umi” and “Kibo” are included in the song. However, the word “momoi” is included in the song song.

Thus, the song feature information generation unit 7 combines the sound feature information Sav and the component word feature information Swd for each corresponding song, and, for example, a plurality of songs corresponding to each song as shown in FIG. 2 (a). The song feature information Ssp composed of the feature information 20 is output to the song feature information database 3 and is registered and stored in the song feature information database 3. At this time, as shown in FIG. 2 (a), one piece of music feature information 20 is the sound of the music piece extracted by the sound feature information extraction unit 5. The characteristic feature and the constituent word information in the song in the song extracted by the constituent word information extracting unit 6 are paired for each song.

[0028] Further, the search word feature information database 4 is used for the search keyword input by the user (that is, the song that the user wants to listen to at that time is subjectively recorded in the song search process described later). Search term characteristic information corresponding to all of the search terms preset as search keywords (hereinafter simply referred to as “search terms”) is stored in an identifiable manner for each search term.

[0029] Here, the search term feature information is each of the search terms presented to the user as to be selected and input when the user searches for the song stored in the song database 2. Information that characterizes (not a search term itself!).

[0030] Next, the search term feature information will be specifically described with reference to FIG.

As shown in FIG. 2 (b), the search term feature information 30 for one search term feature information identification information (FIG. 2 (b)) for identifying each search term feature information 30 from other search feature information 30. ), The search term corresponding to the search term feature information 30 itself, and the search term corresponding to the search term feature information 30 should be searched and extracted from the song database 2. (In other words, search 'expected to be extracted' song feature information characterizing songs included in the song, and the search 'extracted to be extracted' And sound feature information including a plurality of parameters indicating the target features.

Here, the sound feature information constituting the search word feature information 30 specifically includes acoustic parameters similar to the parameters included in the sound feature information Sav.

[0032] Similarly, the song feature information constituting the search word feature information 30 includes a plurality of subjective characteristics that characterize the songs that are included in the song to be searched and extracted using the search word feature information 30. Each of the song feature information is searched using the search term corresponding to the search term feature information 30. It is a set of weighted according to the specific contents of each song included in the song to be extracted .

More specifically, in the example shown in FIG. 2 (b), in the search word feature information 30 corresponding to the search word “heart warming”, the song feature information named “heart warming” is the other song. The feature information is included with a weighting of “0.9”, and the song feature named “Encouraged” Collection information is included with a weighting of “0.3” for other song feature information, and “sad”, “t”, and song feature information for the song name is “0” for other song feature information. 1 is included with a weighting of “1!”, While “0!” Is assigned to the song feature information of “bright! /,”. On the other hand, in the search term feature information 30 corresponding to the search term “bright”, the song feature information “bright” is included with a weight of “0.7” with respect to other song feature information. The song feature information “heartwarming” is included with a weight of “0.2” with respect to other song feature information, and the song feature information “encouraged” is “0” with respect to other song feature information. .5 ”is included, while the weight of the song feature information“ Sadness / Samoshi ”is“ 0 ”. Further, in the search term feature information 30 corresponding to the search term “sad”, the song feature information “heartwarming” is included with a weight of “0.3” with respect to other song feature information. The song feature information “Sad” is included with a weight of “0.8” relative to other song feature information, while the song feature information “bright” and the song feature information “encouraged” The weights of "" are each "0". Furthermore, the search word feature information 30 ぉ which corresponds to other search words “encouraged” or “settled”, etc., but each song feature information “heartwarming”, “bright”, “sad” Each of 'Samish' and 'Encouraged' is included with a preset weight.

[0034] Note that the subjective concept that is the same as the subjective concept indicated by each song feature information is indicated by one of the search words (the search word itself becomes a word) as described later. This is to obtain a search result for each user according to their preference. Each song feature information itself is stored in the song database 2 and indicates the feature of the song (more specifically, the song included in the song), and is separate from the search term itself. Details will be described later.

[0035] Then, when searching for a song that the user desires subjectively using information stored in the song feature information database 3 and the search word feature information database 4, respectively, the input unit 9 first. When any one of the above search terms is input by the user, input information Sin indicating the input search term is output to the search processing unit 8.

Thereby, the search processing unit 8 uses the search term feature information data corresponding to the inputted search term as the search term feature information Swds based on the input information Sin as the search term feature information data. In addition to extracting from the base 4, a plurality of pieces of song feature information 20 corresponding to all songs stored in the song database 2 are extracted from the song feature information database 3 as song feature information Ssp. Then, the extracted search word feature information 30 and each song feature information 20 are respectively compared, and the song feature information 20 that is most similar to the search word feature information 30 indicates a song. The song identification information Srz is generated and output to the song database 2. In parallel with these, the search processing unit 8 is a figure indicating one search word feature information 30 from which the search word feature information database 4 is extracted as corresponding to the input search word (hereinafter simply referred to as a search word feature). And the figure showing the song feature information 20 that is most similar to the extracted search term feature information 30 (hereinafter simply called the song feature information figure). A display signal Sdp is generated and displayed on the display 12 for display on the display 12 as a figure of composition. A specific composition example of the search word feature information graphic and the song feature information graphic will be described in detail later.

Then, the song database 2 outputs the song indicated by the song identification information Srz to the song output unit 10 as song information Ssg.

[0038] Thereby, the music output unit 10 performs necessary output interface processing or the like on the output music information Ssg, and the processed music information Ssg is transmitted to an external amplifying unit (not shown) or broadcast transmission. Output to the part.

[0039] After the song information Ssg indicating one song is output from the song output unit 10, the song corresponding to the output song information Ssg is the user who first inputs the search word. Whether or not the force that was desired is input as evaluation information in the input unit 9 again, and the corresponding input information Sin is output to the search processing unit 8.

[0040] Thereby, the search processing unit 8 generates history information indicating the result of the past song search processing based on the evaluation information input as the input information Sin, and stores the history information as history information Sm. The data is temporarily stored in the unit 11 and read out as necessary, and the history management process described later is performed.

[0041] (II) Song search processing.

Next, the music search processing according to the embodiment executed using the music search device S having the above-described configuration will be specifically described with reference to FIGS. Figure 3 shows the song FIG. 4 is a flowchart illustrating song search processing in the search device, FIG. 4 is a diagram illustrating the contents of a conversion table as conversion information used in the song search processing, and FIG. 5 is a search word corresponding to the input search word. FIG. 6 is a diagram showing a display example of a music feature information graphic corresponding to a characteristic information graphic and a music extracted as appropriate, and FIG. 6 is a diagram illustrating history information used for history processing according to the embodiment; FIG. 7 is a flowchart showing the history management process using the history information, FIGS. 8 and 9 are diagrams illustrating databases used for the history management process, and FIG. 10 is an input search. It is a figure which shows the other example of a search word characteristic information figure corresponding to a word, and the other display example of the music feature information figure corresponding to the music extracted as it correspondingly.

[0042] As shown in FIG. 3, in the music search process of the embodiment that is mainly executed centering on the search processing unit 8, first, the user determines a desired subjective search word. Input in the input unit 9 (step S1), the search term feature information 30 corresponding to the input search term is extracted from the search term feature information database 4 and output to the search processing unit 8 (step S2). ). Then, the search processing unit 8 displays graphic information for displaying the search word feature information graphic corresponding to the extracted search word characteristic information 30 on the display 12 according to the contents of the search word characteristic information 30 at that time. And is temporarily stored in a memory (not shown) in the search processing unit 8 (step S15).

In parallel with the processing of steps S 1 and S 2, the constituent words constituting the songs included in all the songs are read from the song feature information database 3 for each song, and the search processing unit 8 (Step S3). Then, by the operation by the user using the input unit 9, the read constituent word is converted, and song characteristic information corresponding to the song included in each song in which the constituent word is included is generated. A conversion table is determined for (step S3-1).

Note that the conversion table is stored in a memory (not shown) in the search processing unit 8 as described later. A newly created (changed) conversion table is appropriately created based on the user's evaluation of the search word feature information figure and the song feature information figure displayed on the display 12 according to the song search result. Not shown in the figure above, it is stored in memory!

[0045] After that, the search processing unit 8 converts the read component word into the determined conversion tape. A process for converting the constituent words into song feature information corresponding to the songs included in each song is performed for each song (step S4).

[0046] Here, the conversion table will be described specifically with reference to FIG.

The song feature information generated by the processing in step S4 is the same as the song feature information weighted in the search word feature information 30 described above, and each song to which the song feature information should correspond. Each is a set of weights of the constituent words constituting the song included in the song according to the specific contents of the song feature information.

More specifically, in the example of the conversion table T shown in FIG. 4, in the song feature information 40 “heartwarming”, the constituent word “Kibo” is “0.4” with respect to other constituent words. It is included in the weighting, and the constituent word “umi” and the constituent word “momoi” are included with the weighting of “0.1” relative to other constituent words, while the constituent word “eye” is included. The weight is set to “0”, and the song feature information “warming up” is generated. On the other hand, in the song feature information 40 “bright”, the constituent word “momoi” is included with a weighting of “0.8” with respect to other constituent words, and “eye” A constituent word is included with a weight of “0.2” with respect to other constituent words, and a constituent word of “Umi” and a constituent word of “Kibo” are “0.1” with respect to other constituent words. ”Is included. Furthermore, in the song characteristic information 40 “sad”, the constituent word “Kibo” is included with a weighting of “0.7” relative to other constituent words. The constituent word “mi” is included with a weight of “0.2” relative to other constituent words, while the weighting of the constituent word “eye” and the constituent word “omoi” is “0”. Then, the song feature information 40 “sad, sad,” is generated. Finally, in the song feature information 40 “encouraged”, the constituent word “Kibo” is included with a weight of “0.8” relative to other constituent words, and the constituent word “Umi” is included. Is included with a weight of “0.4” for other constituent words, and the constituent word of “eye” is included with a weight of “0.5” for other constituent words. The weight of the constituent word “momoi” is set to “0”, and the song feature information 40 “encouraged” is generated.

[0048] In the process of step S4, corresponding song feature information 40 is generated from each constituent word for each song using the conversion table T illustrated in FIG. More specific For example, when the conversion table T illustrated in FIG. 4 is used, among the constituent words listed in the conversion table Τ, only the constituent words “Umi”, “Omoy”, and “Kibo” are included in a song. If it is included, the value of the song characteristic information 40, which is “warming up” for a certain song, is the “heart” of each of the constituent words “Umi”, “Omoy” and “Kibo”. It becomes “0.6” with the weights “0.1”, “0.1” and “0.4” in the song feature information 40 “warm”. Similarly, the value of the song feature information 40 “bright” for the song is the weight in the song feature information 40 “bright” for each of the constituent words “umi”, “moy”, and “kibo”. It becomes “1.0” with “0.1”, “0.8” and “0.1”. Similarly, each of the song characteristic information 40 listed in the conversion table き is added, and the respective weight values corresponding to the constituent words are added to determine the respective values in the song.

[0049] Further, in parallel with the processing of steps S1 and S2 and steps S3 to S4, only the sound feature information in each of the song feature information 20 corresponding to all songs from the song feature information database 3 is stored in each song. Each time it is read out and output to the search processing unit 8 (step S5).

[0050] Thus, the search processing unit 8 includes each song feature information (including the weight of each song feature information 40) included in the one search word feature information 30 extracted in step S2 and The song feature information 40 corresponding to each song converted in step S4 is compared for each song, and the sound feature information contained in the one search word feature information 30 and extracted in step S5. The sound feature information corresponding to each song is compared for each song, and the song feature information 40 corresponding to each song and the similarity between the sound feature information and the input search word are calculated for each song ( Step S6).

[0051] Based on the calculated similarity for each song, a reproduction list is created in which the songs to be output are arranged in descending order of the similarity (step S7).

[0052] Next, the search processing unit 8 selects a song feature information figure corresponding to the song feature information 20 corresponding to the song to be played first, for example, among the songs included in the created playlist. The graphic information to be displayed on the display 12 is constructed according to the contents of the song feature information 20. Then, the display signal Sdp including graphic information corresponding to the constructed music feature information graphic and graphic information corresponding to the search word characteristic information graphic constructed and stored by the processing of step S15 is obtained. Generated and output to display 12, The search word feature information figure and the song feature information figure are simultaneously displayed on the same screen (step S16).

[0053] Next, while the display is being executed, the above-mentioned song feature information is generated due to reasons such as the displayed search word feature information figure and the song feature information figure not being similar. Check whether the operation to change the conversion table T used in (Step S4) and generate song feature information again using another new conversion table T has been executed in the input unit 9. (Step S17).

[0054] When an operation for changing the conversion table T is performed on the input unit 9 (step S17; YES), the data is stored and stored in the memory, not shown so far. Display the conversion table T and let the user select it, or execute a display or the like on the display 12 to create a new conversion table T. As a result, the newly changed conversion table is displayed. T is stored in the memory (step S18), and the process proceeds to step S3-1. After this, a new conversion table is formally determined in the process of step S3-1, and is thereafter used for the process after step S4.

[0055] On the other hand, if the operation to change the conversion table T is not performed in the determination in step S17 (step S17; NO), it is indicated by the reproduction list (see step S7). The songs are also shown in the order of V and are extracted from the song database 2 and output via the song output unit 10 (step S8).

[0056] Here, the display forms of the search word feature information graphic and the song feature information graphic will be specifically described with reference to FIG.

As illustrated in FIG. 5, the search word feature information graphic 50 and the song feature information graphic 60 are displayed side by side in the display 12 at the same time.

At this time, more specifically, for example, as shown on the right side of FIG. 5, the music feature information graphic 60 is a graphic showing the tone of the music characterized by the music feature information 20 corresponding to the music feature information graphic 60. One of the song feature information generated using one of the parameters included in the sound feature information included in the song feature information 20 and the constituent word included in the song feature information 20, centering on the key figure 66 Characteristic figures 61 to 65 including preset ones are included. [0059] Of these, there are 24 types of key 66 (specifically, "C major,""Bminor,""Fmajor,""Dminor,""b-flat,""Gminor,"" `` E-flat major '', `` C minor '', `` B-flat major '', `` F minor '', `` B-flat major '', `` B-flat minor '', `` G-flat major '', `` E-flat minor '', `` B major '', `` 嬰(G minor), “E major”, “イ C minor”, “A major”, “F minor minor”, “D major”, “B minor”, “G major” and “E minor”)! / A figure that includes only one arrow 67 indicating one of the shifts (that is, one song is represented by one key). At this time, the color of the arrow 67 indicating each key and the display position and display direction of the arrow 67 (display position and display direction in a circle constituting the key figure 66) are determined in advance. In any one of the display position and the display direction, an arrow 67 force S indicating the key included in the music feature information 20 is displayed.

On the other hand, the feature figures 61 to 65 are obtained by converting the constituent words included in the song of the song characterized by the song feature information 20 using the conversion table T shown in FIG. In addition, one of the song feature information that characterizes the song, the deviation, and the parameters included in the song feature information 20 as sound feature information (excluding the parameters that indicate the key, see Fig. 2 (a)) Of the heels, the preset number (five in the case of FIG. 5) is set to the preset display position (in the case of FIG. 5, the positions of the respective vertices of the pentagon centered on the key figure 66). Is displayed. For example, the feature figure 61 corresponds to the song feature information “heartwarming” of the song, the feature figure 62 corresponds to the song parameter “BPM”, and the feature figure 63 corresponds to the song feature information “bright” of the song. Correspondingly, the feature graphic 64 corresponds to the parameter “maximum level” of the song, and the feature graphic 65 is displayed as corresponding to the parameter “average level” of the song. In each of the feature graphics 61 to 65, the song feature information or parameter values indicated as the feature graphics are obtained by using the maximum value and the minimum value that each of the song feature information and each parameter can take. It is expressed by a preset color as a normalized value.

[0061] Further, the conversion table T is stored in a memory (not shown) in the search processing unit 8 and can be arbitrarily selected by the user (see step S18). Based on the conversion table T selected by the user, the deviation of the song characteristic information obtained is displayed at the display position. [0062] Of the characteristic figures 61 to 65, the characteristic figure indicating the parameters included in the music characteristic information 20 as the sound characteristic information is displayed as the characteristic figure during the reproduction of the music. When the value of the specified parameter changes, the color or shape of the corresponding feature figure may be changed according to the change.

[0063] On the other hand, more specifically, for example, as shown on the left side of FIG. 5, in the search word feature information graphic 50, in the search word feature information graphic 50, first, the search word feature information graphic 50 corresponds to the search word feature information graphic 50. One or more strengths suitable for the song to be searched using the search terms characterized by the word feature information 30 One or more colors whose colors change according to their degree of suitability (hereinafter referred to as correlation values) Is displayed at the center of the search term characteristic figure 50 as a key figure 56 including an arrow 57. Further, with the key figure 56 as a center, from any of the power parameters included in the sound feature information included in the search word feature information 30 and any song feature information included in the search word feature information 30, Characteristic figures 51 to 55 including those previously set as corresponding to the characteristic figures 61 to 65 in the music characteristic information figure 60 are included.

[0064] Of these, the key figure 56 is a figure including one or more arrows 57 indicating any of the 24 key types described above in the music feature information figure 60. At this time, as in the case of the music feature graphic 60 described above, the display position and the display direction of the arrow 57 indicating each key are determined in advance, and the display position and the display direction are set to any of the determined display position and display direction. An arrow 57 indicating the key included in the search term characteristic information 30 is displayed. Then, the color of each arrow 57 in the diagram 56 is, for example, a negative value depending on the correlation value between each of the 24 types of key and the key of the song to be searched by the search word. Tones with correlation values are indicated by displaying blue arrows 57 in the corresponding preset display position and orientation, while tones with positive correlation values are preset with corresponding red arrows 57. It is indicated by displaying in the displayed display position and display direction. In addition, the absolute value of each correlation value is displayed darker as the absolute value is larger, depending on the color of the corresponding arrow 57 (red or blue).

[0065] On the other hand, the feature graphics 51 to 55 are included in the search word feature information 30 as sound feature information, as well as the song feature information included in the search word feature information 30. One of the parameters (excluding the parameter indicating the key, see Fig. 2 (b)) and the preset number (five in the case of Fig. 5) correspond to the above-mentioned music feature information figure 60. Are displayed at preset display positions (in the case of FIG. 5, the positions of the respective vertices of the pentagon centered on the above-mentioned key figure 56). For example, in the case of FIG. 5, the feature graphic 51 corresponding to the feature graphic 61 in the music feature information graphic 60 corresponds to the song characteristic information “heartwarming” included in the search word feature information 30 and the music characteristic information graphic The feature figure 52 corresponding to the feature figure 62 in 60 corresponds to the parameter “BPM” included in the search word feature information 30, and the feature figure 53 corresponding to the feature figure 63 in the song feature information figure 60 is the search word feature information 30 corresponds to the song feature information “bright”, and the feature figure 54 corresponding to the feature figure 64 in the song feature information figure 60 corresponds to the parameter “maximum level” contained in the search word feature information 30. The feature graphic 55 corresponding to the feature graphic 65 in the music feature information graphic 60 is displayed as corresponding to the parameter “average level” included in the search word feature information 30. Then, in each of the feature graphics 51 to 55, as in the above feature graphics 61 to 65, the song feature information or parameter values shown as the respective feature graphics are the same as each of the song feature information and each of the parameters. It is represented by a preset color as a value normalized using the maximum and minimum values that can be taken. In contrast, in the feature graphics 51 to 55, the weight value of each song feature information included in each search feature information 30 (see FIG. 2 (b) and its related explanation part) Due to the difference in gradation, the higher the weighting value, the more the background is not filled with gradation, and the smaller the weighting value, the more prominent the gradation is for the background.

[0066] When one song is output by the process in step S8, the user who has listened to the output song does the output song correspond to the search term input and determined in step S1? The evaluation result is input using the input unit 9 (step S9).

[0067] When the input evaluation result is an evaluation that the output song is suitable for the search word (step S9; match), the match history information described later is updated (step S9). 11) Proceed to step S12. On the other hand, if it is determined in step S9 that the output song is not suitable for the search term (step S9; non-match), it will be described later. The non-match history information to be updated is updated (step S10), and the process proceeds to step S12.

Here, the non-match history information updated in step S 10 and the match history information updated in step S 11 will be described more specifically with reference to FIG.

[0069] First, as the matching history information G, as shown in Fig. 6 (a), the song feature information 20 of the song that is evaluated as appropriate for the search word by the evaluation corresponding to the search word input by the user 20 Based on the constituent word information included in the song feature information 20, the method similar to the method described in step S4 above with reference to the conversion table T, which is valid at that time. Is included in the song feature information 40 (song feature information 40 corresponding to the song).

[0070] On the other hand, as the non-match history information NG, as shown in Fig. 6 (b), the song feature of the song evaluated by the above user's evaluation as not suitable for the search term input by the user is shown. Similar to the method described in step S4 above with reference to the conversion table T which is valid at that time based on the constituent word information included in the song feature information 20 instead of the information 20 The song feature information 40 (song feature information 40 corresponding to the song) generated by the above method is included as in the case of the matching history information G.

[0071] When the update of each history information is completed for the preset number of songs, the contents of the conversion table and the search word feature information 30 are updated based on the results (steps). S 12).

[0072] Here, in the case of the present embodiment, when updating the contents of the conversion table T among the updates, the conversion table T after the change is added when the conversion table T is changed in step S18. Will say processing. In addition, in the addition process, only the conversion table T is newly added, and other old! And conversion table T! / Are likely to be deleted by updating themselves.

[0073] Next, it is confirmed whether or not the output has been completed up to the last song in the playlist created in step S7 (step S13), and the output up to the last song is completed! / If not (Step S13; NO), return to Step S8 and output the next song in the playlist, and repeat Steps S9 to S12 described above for the next song. on the other hand, If it is determined in step S13 that the output up to the last song has been completed (step S13; YES), the series of song search processing ends.

Next, the update process of the conversion table T in each update process in step S12 will be described with reference to FIGS.

[0075] In the update process of the conversion table T, as shown in FIG. 7, whether or not the search term that is input at the timing to perform the update process is different from each song feature information 40. (Step S20), and if the subjectivity does not match (Step S20; NO), the conversion table T is not updated. Then, the process proceeds to processing for updating the next search term feature information 30.

[0076] On the other hand, when the subjectivity matches (step S20; YES), the process proceeds to the actual conversion table T update processing.

[0077] In the update process described below, one search term is used for a preset number of songs (40 songs in the case of FIGS. 8 and 9 (20 songs suitable and 20 songs not suitable)). (In the case shown in Fig. 8 and Fig. 9, it was evaluated that the contents of V, the constituent words included in the song evaluated as appropriate for the search word "heartwarming" and that it was not appropriate for the search word) The following describes how to update the conversion table T corresponding to the search term based on the contents of the! / And the constituent words included in the song. 8 and 9 show items necessary for the update process of the conversion table T in the history information shown in FIG. 6 and the correlation used for displaying the graphic 56 in the search word characteristic information graphic 50. Only items that are necessary for the calculation process described later are extracted.

[0078] In the actual update process of the conversion table T, first, for each piece of match history information, each song evaluated as V corresponding to the match history information (in FIG. Corresponding) For each word, add all “0” or “1” in the vertical direction in Fig. 8 for all songs (20 songs in total) and add it to the total number of songs (Fig. 8). In this case, the average value AA is obtained by dividing by “20”) (step S21). For example, in the case of the constituent word “eye” in the match history information, there are five songs that contain that constituent word (in other words, the value is “1” in FIG. 8). Is divided by “20” which is the total number of songs, the average value AA is “0.25” for the constituent word “eye”. Then, this average value calculation process is executed for all the constituent words. [0079] Next, in the same manner as above, for each non-matching history information, for each song composition that was evaluated as inappropriate, all songs were set to “0” in the vertical direction in FIG. Alternatively, all “1” s are added and divided by the total number of songs to obtain an average value DA (step S21). For example, in the case of the constituent word “Ai” in the non-match history information, there are 14 songs that contained that constituent word, and when this is divided by the total number of songs, the average value DA for the constituent word “Ai” is “0.70”. And this average value calculation process is performed about all the constituent words.

[0080] Here, for each constituent word, the larger the difference between the average value AA and the average value DA, the larger the V, the more the song feature information 40 corresponding to the current search word can be expressed by that constituent word. The sex will be higher.

[0081] However, it should be noted that the average values AA and DA calculated by the above-described processing are accumulated in the song database 2 and are evaluated as a result of evaluation for all the songs. It is not obtained. Therefore, the confidence margin of each calculated average value (statistically equivalent to what is called the “sample ratio”) is obtained, and the difference between the average value AA and the average value DA corresponding to each component word is obtained. Confirm whether (average value AA-average value DA) can be used as a weighting value in the history information of the constituent word. More specifically, assume that the reliability of the confidence limit range is 90%, and check using the trust limit range calculated by the following formula (1) (Step S22). That is, for each component word,

Confidence limit width = 2 X 1.65 X [(AAX (1− AA)) / number of songs] ¹ / ² … H) Calculate the corresponding confidence limit width.

[0082] Next, when the confidence limit width is calculated, it is used to determine whether the absolute value of the value obtained by subtracting the average value DA from the average value AA is equal to or greater than the calculated confidence limit width. (Step S23).

[0083] Then, when the absolute value of the average value AA is subtracted from the confidence limit width (step S23; YES), the difference is regarded as a reliable value. Adopted as a weighting value of the corresponding constituent word in the conversion table T (in the case of FIG. 8, the constituent word “heartwarming”) is registered (stored) in the conversion table T (step S 24). On the other hand, when the absolute value of the value obtained by subtracting the average value DA from the average value AA is less than the confidence limit range (step S23; NO), the difference is Value is unreliable Thus, the weight of the corresponding constituent word in the conversion table T is updated to “0” (step S25).

[0084] Note that, in the initial state of the information search device S, each history information stores a preset initial value, and the number of songs to be the history information is limited. As a result, old history information is overwritten with new history information. As a result, the subjectivity specific to each user is gradually reflected in the weighting value of the conversion table につれて as the evaluation of the output music progresses. Then, the newly created conversion table させ reflecting the overwritten history information is stored in a memory (not shown) in the search processing unit 8 each time, and the user converts it into song feature information (Step 3 in Fig. 3). The conversion table Τ used in (see S4) can be selected according to individual preference (see step S18 in Fig. 3).

Next, the calculation process of the correlation value used for displaying the arrow 57 in the graphic 56 in the search word feature information graphic 50 will be described with reference to FIG. 8 and FIG.

Specifically, the calculation process is calculated by the same process as the conversion table テーブル update process described above.

[0087] In the calculation process described below, one search term is used for a preset number of songs (40 songs in the case of Figs. 8 and 9 (20 songs that are not appropriate and 20 songs that are not appropriate)). (In the case shown in Figs. 8 and 9, the search word "heartwarming") and the key of the song evaluated as inappropriate for the search word. The case where the correlation value of the key corresponding to the search term is calculated will be described. As for the key corresponding to one song, even if the song is modulated in the middle, one key is associated with one song, so Fig. 8 and Fig. 9 (a There is only one key in a song with a value of “1” in the horizontal line.

[0088] In the actual correlation value calculation process, first, for each piece of matching history information, each piece of music that was evaluated as being appropriate (the song with the same storage address in Fig. 8 and Fig. 9 (a)). For all songs (20 songs each), add all “0” or “1” in the vertical direction and add it to the total number of songs (Fig. 8). In this case, the average value AA is obtained by dividing by “20” (see step S21 in Fig. 7). For example, in the case of C major in the match history information, there are 3 songs that include that key (in other words, the value is “1” in FIG. 8). When dividing by “20”, the average value AA for C major is “0.15”. Then, this average value calculation process is executed for all the keys.

[0089] Next, in the same manner as above, for each non-matching history information, all songs are evaluated for each of the songs evaluated as inappropriate! Then add “0” or “1” in the vertical direction in FIG. 9 (a) and divide it by the total number of songs to obtain the average value DA (see step S21 in FIG. 7). For example, in C major in the non-match history information, there are 4 songs that included that key, and when this is divided by the total number of songs, the average DA for C major is “0.20”. Become. Then, this average value calculation process is executed for all the keys.

[0090] Here, for each key, the greater the difference between the average value AA and the average value DA! /, The more the key is suitable for the current song key (in other words, the correlation value is larger). The possibility is high.

However, since there is a statistical problem similar to that in the case of updating the conversion table T in the correlation value calculation process, the confidence limit width of each average value calculated here is also calculated. It is determined whether the difference between the average value AA and the average value DA corresponding to each constituent word (average value AA—average value DA) can be used as a weighting value on the history information of the constituent word. Check. More specifically, the reliability of the confidence margin is assumed to be 90%, and confirmation is performed using the confidence margin calculated by the above formula (1) (see step S22 in Fig. 7).

[0092] Next, when the confidence limit width is calculated, it is used to determine whether the absolute value of the value obtained by subtracting the average value DA from the average value AA is equal to or greater than the calculated confidence limit width. (Refer to step S23 in Fig. 7).

[0093] When the absolute value of the average value AA is subtracted from the confidence limit width (see YES in step S23 in FIG. 7), this difference is a reliable value. If there is, it is newly adopted as the correlation value TT of that key (see step S24 in FIG. 7 and FIG. 9B). On the other hand, when the absolute value of the value obtained by subtracting the value of the average value DA from the average value AA is less than the confidence limit (see step S23; NO in Fig. 7), this difference is considered to be an unreliable value, and the correlation value Do not update (see step S25 in Figure 7).

[0094] Note that, in the initial state of the information search device S, as in the case of the update process of the conversion table T, each history information stores the initial value of the correlation value set in advance. In addition, since the number of pieces of music targeted for the history information is limited, the new history information is overwritten on the old history information as a result.

[0095] As described above, according to the operation of the song search device S of the embodiment, the search word feature information 30 corresponding to the input search word and the song features corresponding to each song that is accumulated! The information 20 and the song are searched to search for the song, so that it is possible to surely search for a suitable song by the input search word and to use the input search word directly compared to the case of searching for the song. Search for songs that match the user's subjectivity.

[0096] Since the search term feature information image 50 and the song feature information image 60 are displayed on the same screen of the display 12, the search term feature information 30 to be searched based on the search term input by the user. You can compare the content with the content of the song feature information 20 that indicates the song actually extracted by the search term, and use the similarity between the feature of the search term entered by the user and the feature of the extracted song as an image. I can grasp it.

[0097] Further, since the conversion table T used for generating the song feature information can be arbitrarily selected from those stored in a memory (not shown) in the search processing unit 8, the user can select If the song actually extracted by the search word is not like, the conversion table T can be arbitrarily selected from the past and changed.

[0098] Furthermore, since the search word feature information image 50 and the song feature information image 60 are displayed on the display 12 simultaneously and using images having the same composition, the comparison of both can be easily performed visually.

[0099] Furthermore, since the degree of weighting of the song feature information in the search word feature information 30 is displayed by the gradation change of the feature diagrams 51 to 55, the change in weighting can be easily recognized visually. be able to.

[0100] In addition, since the size of each parameter indicating the acoustic feature indicated by each of the song feature information 20 and the search word feature information 30 is displayed for each parameter in a visible manner, the details for each parameter are displayed. It is possible to visually recognize large and small relationships.

[0101] Further, since the size of each song feature information indicating the feature of the lyrics indicated by each of the song feature information 20 and the search word feature information 30 is displayed for each song feature information, It is possible to visually recognize a detailed magnitude relationship for each feature information. [0102] Furthermore, since the size of each parameter is expressed by the difference in the color of the graphic representing each parameter, the size of each parameter can be easily recognized visually.

[0103] The search word feature information 30 is updated based on the evaluation information, and the size of the feature graphics 51 to 55 indicating the weighting of the song feature information in the search word feature information 30 is the search word feature information 30. Therefore, it is possible to visually grasp the weighting of the song feature information in the search word feature information 30 and the transition of the search word feature information 30 based on the weighting.

In the above-described embodiment, the weighting difference of the song feature information in each search word is expressed by changing the gradation of the feature figures 51 to 55 in the search word feature information figure 50. In addition, for example, the length in the characteristic figures 71 to 75 (corresponding to the characteristic figures 51 to 55 in the search word characteristic information figure 50 shown in the left of FIG. 5) like the search word characteristic information figure 70 shown in the left of FIG. Can be used as an indication of the degree of weighting of the corresponding song feature information. Here, the right side of FIG. 10 is a music feature information figure 80 having a display form corresponding to the search word characteristic information figure 70, and the characteristic figure 81 to 85 (the characteristic figure in the music feature information figure 60 shown on the right side of FIG. 5). (Corresponding to 61 to 65) is also a triangle according to the search word feature information graphic 70. Further, in FIG. 10, the same members as those in each figure shown in FIG. 5 are indicated by the same member numbers.

[0105] In the case shown in FIG. 10, since the weighting size of the song feature information is displayed by the change in the size of the feature figures 71 to 75, the change in weighting can be easily recognized visually. Can do.

[0106] In the above-described embodiment, the user can change the search word feature information 30 corresponding to the displayed search word feature information graphic 50 or 70 by the operation of the input unit 9. You can also.

[0107] In this case, the user can change the contents of the search term characteristic information 30. Therefore, by changing the search word feature information 30 according to the user's preference, the user can be further changed. Music according to hobbies etc. can be extracted.

Further, by the operation in the input unit 9, the search word feature information 30 corresponding to the displayed search word feature information image 50 or 70 and the song feature information included in the search word feature information 30 are displayed. It is also possible to configure each weighting to be changed.

In this case, by changing the weighting of the search term feature information 30 and the song feature information according to the user's preference, it is possible to further extract the song according to the hobby etc. to the user. it can.

[0110] Furthermore, when the search word feature information graphic 50 or 70 and the song feature information graphic 60 or 80 are displayed at the same time, the search word characteristic information graphic 50 or 70 displayed is displayed. The search word feature information 30 corresponding to is replaced with the same content as the song feature information 20 corresponding to the song feature information image 60 or 80 displayed at that time in the search processing unit 8. Can be configured to be stored again in the search term feature information database 4

[0111] In this case, when the user likes the extracted song, the search term feature information 30 corresponding to the entered search term can be replaced with the song feature information 20 of the song, so that the search can be performed again. When a word is input, the extracted music is extracted again, and the music can be extracted according to the user's hobbies.

[0112] Furthermore, in the above-described embodiment, the power described in the case where the present application is applied to the music search apparatus S that stores and searches for a plurality of songs, other than this, still images or moving images are stored. It is also possible to apply the present application to an image search apparatus that searches for these according to the subjectivity of the user.

Furthermore, a program corresponding to the flowcharts shown in FIGS. 3 and 7 is recorded on an information recording medium such as a flexible disk, or the program is acquired and recorded via a network such as the Internet. The general-purpose microcomputer can be used as the search processing unit 8 according to the embodiment by reading and executing these using a general-purpose microcomputer.

Claims

The scope of the claims

[1] Intensity of multiple songs In a song search device that searches for one or more of the songs,

Song feature information accumulating means for accumulating song feature information indicating at least the features of the lyrics included in the song for each song;

A search word input means used for inputting a search word that is a search word indicating the music to be searched and is a word indicating subjectivity;

Search song feature information indicating at least the feature of the lyrics included in any of the songs to be searched using the search word as appropriate for the subjectivity indicated by the input search word is provided for each search word. Retrieving song feature information accumulating means for accumulating identifiable information, and comparison target information to be compared with input retrieval song feature information which is the searching song feature information corresponding to the inputted search word is accumulated! A conversion means for generating the song characteristic information by converting it using conversion information (conversion table);

Comparison means for comparing the input search song feature information and the generated comparison target information, respectively.

Based on the comparison result by the comparison means, the song corresponding to the song feature information that is the generation source of the comparison target information that is most similar to the input search song feature information corresponds to the inputted search word. Extracting means for extracting as the song;

An instruction input means used for inputting an instruction to change the conversion information; a changing means for changing the conversion information based on the input instruction;

With

When the conversion information is changed, the conversion means converts the music feature information into the comparison target information using the converted conversion information.

[2] In the music search device according to claim 1,

First display control means for displaying on the display means an input search song feature information image that is an image showing the contents of the input search song feature information;

Second display control means for causing the display means to display an extracted music feature information image that is an image showing the content of the music feature information corresponding to the extracted music; With

The first display control means and the second display control means receive the input search song feature information image and the extracted song feature information image on the display means when the instruction is input by the instruction input means. A music search apparatus characterized by displaying.

[3] In the song search device according to claim 2,

The first display control means and the second display control means simultaneously perform the input search song feature information image and the extracted song feature information image, and the input search song feature information image and the extracted song feature information image. Is displayed on the display means using an image having the same composition.

[4] In the music search device according to any one of claims 1 to 3,!

The song feature information and the search song feature information include a plurality of symbolic parameters that subjectively symbolize the contents of the lyrics constituting each song and correspond to different subjects as parameters indicating the characteristics of the lyrics. Included in each! /

The first display control means and the second display control means display the size of each symbolic parameter in a visible manner in each of the extracted music feature information image and the input search music feature information image. A song search device that is characterized.

[5] Song feature information accumulating means for accumulating at least each piece of song feature information indicating the characteristics of the lyrics included in the song;

It should be searched using the search word as corresponding to the subject indicated by the search word consisting of the word indicating the song and subjectivity to be searched! /, The lyrics included in any of the songs Search song feature information storage means for storing search song feature information indicating at least the feature of each search word for each of the search terms;

Display means;

With

A song search method that is executed in a song search device that searches for one or more songs from a plurality of songs,

A search term input step for inputting the search term;

Input search song feature information which is the search song feature information corresponding to the input search word The information to be compared to be compared with the information is accumulated! A conversion process for generating the song feature information by converting it using conversion information (conversion table);

A comparison step for comparing the input search song feature information and the generated comparison object information, respectively.

Based on the comparison result in the comparison step, the song corresponding to the song feature information that is the generation source of the comparison target information that is most similar to the input search song feature information is used as the inputted search word. An extraction step of extracting as the corresponding song;

An instruction input step of inputting an instruction to change the conversion information;

A changing step of changing the conversion information based on the input instruction;

Including

In the conversion step, when the conversion information is changed, the music characteristic information is converted into the comparison target information by using the changed conversion information.

[6] A program for searching for music, which causes a computer to function as the music search device according to any one of claims 1 to 4.

[7] The song search program according to claim 6 is recorded so as to be readable by the computer! / An information recording medium characterized by being able to speak.