JP5742472B2 - Data retrieval apparatus and program - Google Patents

Description

  The present invention relates to a technique for searching for musical sound waveform signals.

  There is a technology for storing pieces of sound material in a database and generating musical sounds by combining these sound materials. The sound material used when generating a musical sound is selected from a large number of sound materials registered in the database. For this reason, in order to make it easier for the user to select a desired sound material, a technique for performing a search process for narrowing down from these many sound materials to some sound materials has been developed. For example, Patent Literature 1 discloses a technique for searching for sound material similar to the search condition from the sound material registered in the database and displaying the search result by the user specifying the sound material as the search condition. Have been described.

JP 2010-271398 A

  A plurality of sound materials obtained as a result of the search are displayed in the order similar to the search condition. The user selects a desired sound material from the plurality of displayed search results. At this time, if the search results are displayed in a similar order close to the user's sense, the sound material can be easily selected, but it is not always displayed in such a similar order. Therefore, the user has to make trial and error by specifying various search conditions so that the search results are displayed in the order of similarity assumed by the user.

  The present invention has been made in view of the above circumstances, and an object thereof is to obtain a sound material close to a user's assumption as a search result.

In order to solve the above-described problem, the present invention uses a database in which musical sound data indicating a musical sound waveform signal and characteristic data indicating a plurality of types of characteristic values of the musical sound waveform signal are registered in association with each other. A data search device that performs a search based on a specified search condition, wherein the first search condition is condition determining means for determining a plurality of types of feature quantities, and the feature is similar to the first search condition. Algorithm determining means for determining the search algorithm as a second search condition including a method for calculating similarity from a plurality of search algorithms each indicating a search method for quantity data, and a search algorithm indicated by the second search condition using a feature identifying means for identifying and searching the feature data similar to the first search condition from the database, the first search condition, Calculation of the degree of similarity contained in the serial second search condition, and is displayed on the display means the search condition setting for selecting the set of weights of the feature amount, corresponding to the feature data identified by the feature identifying unit There is provided a data search apparatus comprising display control means for displaying information indicating musical tone data to be displayed on a display means as a search result.

  In another preferred embodiment, the musical sound specifying means for specifying the musical sound data selected by the user from the musical sound data corresponding to the information displayed on the display means, and the musical sound specified by the musical sound specifying means. Update means for updating the search algorithm determined as the second search condition based on the feature quantity relationship between the feature quantity data corresponding to the data and the first search condition, The feature specifying unit specifies a plurality of feature amount data similar to the first search condition, and the display control unit outputs musical sound data corresponding to the plurality of feature amount data specified by the feature specifying unit. Information to be displayed is displayed on the display means in a manner based on the similarity between the feature amount data and the first search condition.

  In another preferable aspect, the plurality of search algorithms include feature data similar to the first search condition based on a distance of the feature data for each type and a weight determined for each type. The search algorithm indicating a search method is included, and the updating means updates the search algorithm by changing the weight.

  In another preferred aspect, when the musical sound data is specified by the musical sound specifying means, the condition determining means is configured to perform the first determination based on feature data corresponding to the musical sound data specified by the musical sound specifying means. The algorithm determining means updates the search algorithm updated by the updating means as the second search condition, and the feature specifying means indicates the updated second search condition. The feature data similar to the updated first search condition is searched from the database and specified by a search method.

In another preferred aspect, the update means performs the update when the second search condition is the predetermined search algorithm.
In another preferable aspect, when the category indicating the classification of the plurality of types of feature amounts is designated by the user, the condition determining unit calculates the feature amounts of the respective types in the designated category. The feature specifying means searches and specifies the feature data similar to the first search condition from the database according to the designated category. To do.
In another preferred embodiment, the condition determining means determines the feature amounts of each of the plurality of types specified by a user as the first search condition, and the feature specifying means is the specified Further, the feature data similar to the first search condition is searched from the database and specified in accordance with the category into which each type of the feature value is classified.

According to the present invention, a computer is designated by a user using a database in which musical tone data indicating a musical sound waveform signal and feature data indicating a plurality of types of characteristic values of the musical sound waveform signal are registered in association with each other. A program for causing a computer to function as a device for performing a search based on a search condition, wherein the computer is used as a first search condition, condition determining means for determining a plurality of types of the feature quantities, and the first search condition Determination of the search algorithm as a second search condition including a calculation method of similarity and a setting of the weight of the feature amount from a plurality of search algorithms each indicating a search method of the feature amount data similar to And using the search algorithm indicated by the second search condition, the feature quantity data similar to the first search condition is stored in the data. A feature identifying means for identifying search from the base, the first search condition, the calculation method of the degree of similarity contained in the second search condition, and a search condition setting for selecting the set of weights of the feature amount There is provided a program for causing a display means to function as a display control means for displaying information indicating musical tone data corresponding to the feature amount data specified by the feature specifying means on the display means as a search result.

  According to the present invention, a sound material close to a user's assumption can be obtained as a search result.

It is a block diagram explaining the structure of the musical tone processing apparatus in embodiment of this invention. It is a figure explaining the example of sound material DB in the embodiment of the present invention. It is a figure explaining the content of the sound material represented by the specific data in the embodiment of the present invention. It is a figure explaining the example of the classification | category template in embodiment of this invention. It is a figure explaining the example of the setting template in embodiment of this invention. It is a block diagram explaining the structure of the data search function in embodiment of this invention. It is a figure explaining the example of the search condition setting display in embodiment of this invention. It is a figure explaining the example of the search result display in embodiment of this invention. It is a figure explaining the example of the sound material determination display in embodiment of this invention. It is a figure explaining the example of the search result display in the modification 4 of this invention.

<Embodiment>
[Overview]
The musical tone processing apparatus according to the embodiment of the present invention is an information processing apparatus such as a personal computer, a mobile phone, a PDA (Personal Digital Assistant), and a tablet terminal, and executes a specific application program on an OS (Operating System). , A device that realizes a function called DAW (Digital Audio Workstation). In the DAW realized in this musical tone processing apparatus, a function of performing control for generating musical tone using a sound material extracted as a part of a musical tone waveform signal is also realized. In addition, each function described below, such as a function for searching for sound material from a database, is also realized. Each of these functions is realized by executing a subroutine program during the execution of the application program for realizing the DAW.

[Configuration of Music Processing Device 10]
FIG. 1 is a block diagram illustrating the configuration of a musical tone processing apparatus 10 according to an embodiment of the present invention. The musical sound processing apparatus 10 includes a control unit 11, an operation unit 12, a display unit 13, an interface 14, a storage unit 15, and an acoustic processing unit 16. Each of these components is connected via a bus. In addition, the musical sound processing apparatus 10 includes a speaker 161 and a microphone 162 connected to the sound processing unit 16.

  The control unit 11 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), and the like. The control unit 11 implements various functions by executing various programs stored in the ROM or the storage unit 15. In this example, execution of the program by the control unit 11 includes execution of an application program that realizes DAW and execution of the above-described subroutine program. The subroutine program includes a reproduction program, an extraction program, and a search program stored in the storage unit 15, and is executed according to a user instruction.

The reproduction program is a program for realizing a reproduction function that reproduces sequence data that defines the sound content of a musical sound in the DAW, and performs a process of generating the musical sound. In this reproduction function, the data of each track in the sequence data described below is reproduced to synthesize a musical sound waveform signal, and audio data indicating this musical sound waveform signal is output to the acoustic processing unit 16 and output from the speaker 161. It is a function to let you.
The extraction program is a program for realizing a sound material extraction function for extracting sound material from various musical sound waveform signals such as a musical sound waveform signal synthesized by a reproduction function. Any known method can be used for extracting the sound material from the musical sound waveform signal. For example, an onset in which the volume change from the musical sound waveform signal changes by a certain level or more is detected, and the characteristic amount of the musical sound waveform signal is calculated in various sections within a predetermined time range from the onset. A section of a feature amount that satisfies the specified condition is extracted as a sound material. Moreover, you may use the method disclosed by Unexamined-Japanese-Patent No. 2010-191337.
The search program is a program for realizing a data search function for searching for sound material based on a search condition from the sound material DB stored in the storage unit 15. Details of the data search function will be described later.

The operation unit 12 has operation means such as an operation button that accepts an operation by a user, a keyboard, a mouse, and a touch panel, and outputs operation data indicating the contents of the accepted operation to the control unit 11. As a result, a user instruction is input to the musical tone processing apparatus 10.
The display unit 13 is a display device such as a liquid crystal display, and displays content on the display screen 131 according to the control of the control unit 11. The content displayed on the display screen 131 varies depending on the executed program in addition to the menu screen, the setting screen, and the like (see FIGS. 7, 8, and 9).

  The interface 14 has a function of transmitting and receiving various data by connecting to an external device and communicating by wire or wireless. The interface 14 is also provided with an AUX (Auxiliary) terminal to which audio data is input from an external device. The interface 14 outputs various data input from the external device to the control unit 11, and outputs various data to the external device under the control of the control unit 11. When an analog signal is input to the AUX terminal, A / D conversion (analog / digital conversion) is performed.

The microphone 162 outputs a musical sound waveform signal indicating the content of the input sound to the acoustic processing unit 16.
The acoustic processing unit 16 includes a signal processing circuit such as a DSP (Digital Signal Processor). In this example, the acoustic processing unit 16 performs A / D conversion on the musical sound waveform signal input from the microphone 162 and outputs it as audio data to the control unit 11. The acoustic processing unit 16 performs signal processing such as acoustic processing, D / A conversion (digital / analog conversion) processing, amplification processing, and the like set on the audio data output from the control unit 11. And output to the speaker 161 as a musical sound waveform signal.
The speaker 161 outputs the sound indicated by the musical sound waveform signal input from the acoustic processing unit 16.

  The storage unit 15 is a non-volatile memory such as a hard disk or a flash memory, and has a storage area for storing the various programs. The storage unit 15 also includes sequence data used when various programs are executed, a sound material database (hereinafter referred to as a sound material DB), a waveform database (hereinafter referred to as a waveform DB), a classification template, and a setting. Each has a storage area for storing a template.

  In the waveform DB, waveform data W1, W2,... Indicating musical tone waveform signals are registered. The musical sound waveform signals indicated by these waveform data are any of various contents such as music, specific instrument sounds, sound materials, etc., and their lengths vary from less than one second to several minutes or more. Note that the waveform data in this example includes data of a plurality of channels (for example, Lch and Rch). In the following description, each data such as waveform data indicating a musical sound waveform signal and audio data is composed of two channels of Lch and Rch, but may be composed of a larger number of channels. , May be composed of monaural (1ch).

  FIG. 2 is a diagram illustrating an example of the sound material DB in the embodiment of the present invention. In the sound material DB, information for specifying the contents of the sound material is registered. As shown in FIG. 2, the information for specifying the content of the sound material includes specification data for specifying the content of the sound waveform signal of the sound material, and feature data indicating the characteristics of the sound waveform signal of the sound material.

The specific data is composed of a combination of waveform designation information for designating one of the waveform data registered in the waveform DB and time designation information for designating a data range in the waveform data by time. In this example, the time designated by the time designation information is determined as the time from the beginning of the waveform data. About the sound material specified by specific data, the identifier (in this example, sn1, sn2, ...) for specifying each is attached | subjected. Hereinafter, when a specific sound material is indicated, it is referred to as a sound material sn1.
For specific data for which time designation information is not defined, the waveform data itself indicated by the waveform designation information represents the content of the musical sound of the sound material. For example, the musical sound indicated by the sound material sn4 is represented by the entire musical sound waveform signal indicated by the waveform data W5.

  FIG. 3 is a diagram for explaining the contents of the sound material represented by the specific data in the embodiment of the present invention. FIG. 3A is a diagram for explaining musical tone waveform signals of musical sounds indicated by the sound materials sn1, sn2, and sn3, and FIG. 3B is a diagram for explaining musical tone waveform signals of musical sounds indicated by the sound material sn4. is there. As shown in FIG. 2, the content of the sound material sn <b> 1 is specified such that the waveform designation information is the waveform data W <b> 1 and the time designation information is ts <b> 1 to te <b> 1. Therefore, as shown in FIG. 3A, the musical sound waveform signal corresponding to the sound material sn1 is a musical sound waveform signal at times ts1 to te1 among the musical sound waveform signals indicated by the waveform data W1. Similarly, the musical sound waveform signals respectively corresponding to the sound materials sn2 and sn3 are specified as a partial range of the musical sound waveform signal indicated by the waveform data W1. On the other hand, as shown in FIG. 3B, the musical sound waveform signal indicated by the sound material sn4 is specified as the whole musical sound waveform signal indicated by the waveform data W5 because the time designation information is not defined. Hereinafter, data representing musical sound waveform signals indicated by the sound materials sn1, sn2,... Will be referred to as musical sound data sn1, sn2,.

  Returning to FIG. 2, the description will be continued. The feature amount data indicates a plurality of types of feature amounts p1, p2,... Possessed by the tone waveform signal of each sound material. For example, for sound material, the intensity of each frequency (high range, mid range, low range), peak time of amplitude (time based on the beginning of the musical sound data), peak amplitude of amplitude, degree of cooperation, complexity This is a characteristic amount for the sound quality, and is a value obtained by analyzing the sound material data. For example, the feature quantity p1 indicates the intensity of the high frequency range of the sound material by its value. Hereinafter, the feature amount data is indicated as Pa, Pb,... By a combination of values of the feature amounts p1, p2,. In addition, the values of Pa feature values p1, p2,... Are indicated as p1a, p2a,. For example, for the sound material sn3, the feature amount data is Pc, and the combination of the feature amounts is p1c, p2c,. In this example, the value of each feature value is determined in the range of “0” to “1”.

FIG. 4 is a diagram illustrating an example of a classification template in the embodiment of the present invention. The classification template determines the classification criteria and the designated value as a representative value for each category for each type of feature for the categories classified by the features of the sound material, that is, the feature data. It is registered in.
This category is classified for each sound material with similar auditory characteristics. For example, a category classified as a sound with a clear attack and a strong edge (for example, edge sound), noise, etc. It is classified as a category (for example, texture sound) classified as a sound that can be heard. The categories to be classified are indicated as category C1, category C2,... According to their contents.

As the classification standard, a minimum value and a maximum value are determined for each feature amount. Therefore, the sound material is classified into a category in which each feature amount satisfies the classification criteria. For example, the sound material classified into the category C1 satisfies the feature quantity p1 in the range of “0.1” to “0.5” and the feature quantity p2 in the range of “0.0” to “0.2”. Will be. In the sound material DB, categories classified according to the sound material may be registered in advance.
As described above, the designated value is a value of a representative feature amount of the category. For example, for the category C2, as the specified values, the feature quantity p1 is “0.5”, and the feature quantity p2 is “0.5”. On the other hand, when a value is not determined like the specified value of the feature quantity p2 in the category C1, it is treated as having no representative value.

  FIG. 5 is a diagram illustrating an example of a setting template in the embodiment of the present invention. The setting template is a template used when the search program is executed, and each setting (weight setting A, weight setting) in which the importance for each type of feature amount is determined in the range of “0” to “1”. B, ...) are registered. In this example, there are a setting whose setting content is determined in advance and a weight setting (for example, a weight setting UB) whose setting content has been changed by the user.

  The weight setting is determined as, for example, “standard setting”, “mid-range strength priority”, “edge priority”, “noise priority”, and the type of feature quantity representing the feature has a relative importance. It is high. For example, in the case of “middle band strength priority”, the importance of the feature quantity indicating “middle band frequency intensity” is determined to be relatively higher than other feature quantities. In the case of “edge priority”, it is determined that the degree of importance is high for the type of feature amount that gives the impression of the edge sound. In the case of “standard setting”, the importance is the same for any kind of feature amount.

  Returning to FIG. 1, the description will be continued. The sequence data has a plurality of tracks that define pronunciation contents in time series. In this example, each type of track is assigned to any one of an audio track, a MIDI (Musical Instrument Digital Interface) track, and a sound material track.

  MIDI tracks are MIDI events such as note-on, note-off, note number, and velocity, and the processing timing of these events (for example, expressed by the number of bars, beats, ticks, etc. from the start of the track data) It is a track that defines the relationship. As described above, in this example, the MIDI track is defined in a commonly used MIDI format. However, the MIDI track is used to control a sound source that generates a musical sound waveform signal having a pronunciation content corresponding to an event. As long as the information track is defined, it may be defined in another format.

  The audio track is a track in which audio data and playback start timing of the data are defined. The audio data may be waveform data stored in the waveform DB, or may be data indicating a musical sound waveform signal input separately. The reproduction start timing is represented by the number of bars, the number of beats, and the number of ticks from the start of the track data, similarly to the above processing timing. Other information such as information indicating the playback volume of the audio data may be included.

The sound material track is a track in which the musical sound data and the reproduction start timing of the data are defined. The musical sound data is specified by the identifier of each sound material in the sound material DB. The reproduction start timing is represented by the number of bars, the number of beats, and the number of ticks from the start of the track data, similarly to the above processing timing. Note that the musical sound data may be specified not by the sound material identifier but by the sound material feature value data. In this case, in the playback function, feature amount data most similar to the feature amount data specified for the sound material track is specified from the sound material DB, and sound material data corresponding to the specified feature amount data should be played back. It may be specified as musical sound data.
The above is the description of the hardware configuration of the musical tone processing apparatus 10.

[Data search function configuration]
Next, a data search function realized when the control unit 11 of the musical sound processing apparatus 10 executes a search program will be described. Note that some or all of the components that implement the search function described below may be implemented by hardware.

  FIG. 6 is a block diagram illustrating the configuration of the data search function in the embodiment of the present invention. When the control unit 11 executes the search program, the data search unit 100 including the display control unit 110, the condition determination unit 120, the algorithm determination unit 130, the feature specification unit 140, the musical tone specification unit 150, and the update unit 160 is configured, and the data A search function is realized.

  The display control unit 110 causes the display screen 131 to display an image indicating the musical sound data instructed from the feature specifying unit 140 (data name, information indicating musical sound data such as an image corresponding to the corresponding feature amount data) (hereinafter, referred to as “music data”). Simply "display music data"). In addition, the display control unit 110 changes the display content on the display screen 131 in accordance with a user instruction input from the operation unit 12. The display contents are various contents related to the data search function (FIGS. 7, 8, and 9), and include, for example, a display for specifying a search condition. A specific example of the contents displayed on the display screen 131 will be described in an operation example described later.

  When a specific category is specified from the classification template according to an instruction from the user, the condition determining unit 120 determines the specified value of each type of feature amount in the category as the first search condition, and performs the first search Information indicating the condition is output to the feature specifying unit 140.

When a search algorithm indicating a search method used in the feature specifying unit 140 is specified by an instruction from the user, the algorithm determination unit 130 determines the search algorithm as a second search condition, and the second search condition Is output to the feature specifying unit 140 and the updating unit 160. In this example, the search algorithm includes a method of calculating the degree of similarity (hereinafter referred to as similarity) calculated by the feature specifying unit 140 and a weight setting selected from the setting template. In this example, the search algorithm also includes the upper limit value and the lower limit value (classification criterion) of each type of feature amount in the category specified by the user as described above.
Here, as an option of the calculation method of the similarity, for example, a calculation method using a method of calculating the distance of the n-dimensional vector or the similarity of the n-dimensional vector such as Euclidean distance calculation, Mahalanobis distance calculation, cosine similarity calculation, etc. If it is. This n-dimensional vector corresponds to the number of types of feature quantities to be compared in the similarity calculation.

  Thus, in this example, the user can specify a search algorithm by selecting a similarity calculation method, weight setting, and category. Then, the algorithm determination unit 130 determines the search algorithm designated by the user as the second search condition.

The feature specifying unit 140 searches and specifies the feature data similar to the first search condition determined as described above from the sound material DB using the search algorithm indicated by the second search condition. Specifically, the search is performed as follows.
The feature specifying unit 140 first calculates the distance between the feature amount data of each sound material registered in the sound material DB that satisfies the classification criteria included in the second search condition and the first search condition. Narrow down as target feature data. The feature specifying unit 140 calculates the similarity with the first search condition for each narrowed-down feature amount data in accordance with the similarity calculation method and the weight setting indicated by the second search condition. From this calculation result, the feature specifying unit 140 specifies feature amount data having a higher similarity (closer distance) than a predetermined value. Then, the feature specifying unit 140 outputs information indicating the tone data corresponding to the specified feature amount data to the tone specifying unit 150. In addition, the feature specifying unit 140 outputs information indicating the musical sound data corresponding to the specified feature data to the display control unit 110 in association with the similarity. As a result, the musical sound data is displayed on the display screen 131 as a search result in the order of similarity. Note that the display screen 131 may display a predetermined number of pieces of musical sound data from the one with higher similarity of the feature data.

Here, as an example of a search method when weight setting is applied to the similarity calculation method, a case where weight setting is applied to Euclidean distance calculation will be described.
First, the values of the feature quantities p1, p2,..., Pn determined as the first search condition are represented as p1x, p2x,. Further, the importance corresponding to the feature amounts p1, p2,..., Pn indicated by the weight setting determined as the second search condition is represented as A1, A2,. When weighting is applied to Euclidean distance calculation and importance is used as a weight for each type of feature amount, Euclid between feature amount data Py (p1y, p2y,..., Pny) and the first search condition. The distance dxy is expressed by the following formula (1).
dxy = (A1 × (p1x−p1y) ² + A2 × (p2x−p2y) ² +... + An × (p1x−p1y) ² ) ^1/2 (1)

Here, the type of feature quantity for which the designated value in the category is not determined as the first search condition is not used for the distance calculation. When the designated value is not determined as in p2 of category C1, for example, p2x = p2y or A2 = 0 in equation (1) may be handled.
It should be noted that even when other similarity calculation methods are used, it is the same that the weight setting is used as the weight for each type of feature amount. As a method of applying the weight setting to other calculation methods, it is only necessary to determine that the feature amount of a type having a larger weight has a greater influence on the change in the similarity (distance). That is, it is only necessary to determine that the feature amount of a type having a greater weight has a larger amount of decrease in similarity (amount of increase in distance) when the difference between the values becomes larger.

  The musical tone specifying unit 150 specifies musical tone data selected by the user as desired musical tone data from the musical tone data displayed as a search result on the display screen 131, and outputs information indicating the specified musical tone data to the updating unit 160. . In this example, the musical sound data specified in this way is used to specify the identifier of the sound material used when creating the sound material track of the sequence data.

The update unit 160 determines the importance in weight setting in the search algorithm indicated by the second search condition, and the relationship between the feature amount corresponding to the musical sound data specified by the musical tone specifying unit 150 and the feature amount of the first search condition Update based on. The update unit 160 registers the weight setting with the updated importance in the setting template. At this time, the update unit 160 may overwrite the original weight setting and register it, or may newly register it as another weight setting. Which one is to be used may be determined in advance according to a user instruction.
In the case of registration by overwriting, the update may be performed only when a predetermined weight setting is designated by the user. In this way, since the weight setting in the setting template changes according to the selection result of the musical sound data by the user, even if the user specifies the same weight setting as the second search condition, the search result changes. Can be.

An example of the importance update mode in the update unit 160 will be described. The updating unit 160 compares the first search condition and the feature amount corresponding to the specified musical sound data for each type, and calculates a difference for each type. Then, the update unit 160 updates the importance corresponding to each type of feature amount so that the feature amount having a smaller difference has a higher importance level. By updating in this way, the user's sense is reflected in the weight setting. Therefore, when the next search is performed, even if the user specifies the same search condition, the display order of the musical sound data displayed as the search result However, compared with the previous display order, it will be arranged in a similar order that suits the user's sense.
This completes the description of the data search function. With the above functional configurations, the data search apparatus of the present invention is realized in the musical tone processing apparatus 10.

[Operation example]
Next, an operation example when the above-described search program is executed will be described with reference to a display example on the display screen 131.
First, the user inputs an instruction to execute a search program to the musical sound processing apparatus 10 when the user wants to select a sound material used for creating sequence data from the sound material DB on the DAW. In this way, the display screen 131 displays the content shown in FIG.

  FIG. 7 is a diagram illustrating an example of search condition setting display according to the embodiment of the present invention. A menu area MA is provided at the top of the display screen 131, and a registration area WA is provided at the bottom. The menu area MA is an area provided for performing various instructions such as start of execution of the search program, storage of data, and stop of execution of the search program. The registration area WA is an area for registering musical sound data selected as a result of the search, and has musical sound registration areas WA1, WA2,..., WA7 for registering musical sound data in the area. The cursor Cs2 is used to select in which area the selected musical sound data is registered.

The category area CA is an area for displaying categories registered in the classification template. In this example, some of the categories C1, C2,... C7, and C8 are displayed. This area can be scrolled up and down, and C8 and after are displayed by scrolling. Here, the cursor Cs1 indicates a state where the category C2 is selected.
Also, among the second search conditions, a selection box SB1 for designating a similarity calculation method (such as the Euclidean distance calculation described above) and a selection box SB2 for designating a weight setting from the setting templates are displayed. Has been. Here, the Euclidean distance calculation is selected as the similarity calculation method, and the weight setting B in the setting template is selected as the weight setting. For example, the user changes the positions of the cursors Cs1 and Cs2 and the contents of the selection boxes SB1 and SB2 using the pointer pt, and operates the decision button b1 to perform the first search with the selected contents. A condition and a second search condition are designated. When the user specifies the first search condition and the second search condition, the condition determination unit 120 determines the first search condition and the algorithm determination unit 130 determines the second search condition. Processing in the feature specifying unit 140 is started. Then, the display screen 131 transitions to the search result display shown in FIG.

  FIG. 8 is a diagram for explaining an example of a search result display in the embodiment of the present invention. The search result area SA is an area where musical sound data (sn5, sn3, sn1,...) Corresponding to the feature amount data specified by the feature specifying unit 140 is displayed as a search result, and is similar to the category area. It is possible to scroll up and down. The musical sound data displayed here corresponds to the feature data searched and specified from the sound material DB by the feature specifying unit 140 according to the first search condition and the second search condition, as described above. In this example, the higher the similarity (the closer the distance), the higher the display. A broken line cursor Cm indicates a category (see FIG. 7) designated by the user in the search condition setting display.

  Here, when the similarity calculation method and the weight setting selected by the user in the search condition setting display are different from those shown in FIG. 7, the similarity for each musical tone data is also different, so that the musical tone data in the search result area SA is different. The display order of changes. For example, when the weight setting indicates “standard setting”, any kind of feature amount is handled in the same manner and the similarity is calculated. On the other hand, when the weight setting indicates “middle band strength priority”, among the feature quantities in the first search condition, the feature quantity data that is close to the middle band feature quantity has high similarity. Therefore, the display order of the musical sound data is displayed differently from the case of “standard setting”. Thus, since the user can designate the second search condition, the display screen 131 displays a search result close to the user's assumption.

Here, when the musical sound data selected by the cursor Cs1 is changed by moving the cursor Cs1 up and down by the user, the control unit 11 sends the musical sound waveform signal of the sound material indicated by the musical sound data via the acoustic processing unit 16. To the speaker 161. For example, when the musical sound data selected by the cursor Cs1 changes from the musical sound data sn3 to the musical sound data sn5, the musical sound corresponding to the musical sound data sn5 is generated from the speaker 161. Thereby, the user can listen to the content of the sound material corresponding to the musical sound data selected by the cursor Cs1.
When the user determines the desired sound material while confirming the pronunciation accompanying the movement of the cursor Cs1, the user moves the cursor Cs1 to the musical sound data corresponding to the desired sound material and operates the decision button b1. The musical sound data selected by the user is specified by the musical sound specifying unit 150. Then, the display screen 131 transitions to the sound material determination display shown in FIG.

  FIG. 9 is a diagram for explaining an example of the sound material determination display in the embodiment of the present invention. When the musical sound data (musical data sn11 in this example) is specified by the musical sound specifying unit 150, information indicating the specified musical data (“11 indicating the musical data sn11” is stored in the musical sound registration area WA1 selected by the cursor Cs2. ]) Is displayed. As described above, the musical sound data registered in the musical sound registration areas WA1, WA2,... Are used to specify the sound material identifier when creating the sound material track.

  The updating unit 160 compares the feature amount indicated by the feature amount data corresponding to the musical sound data sn11 with the feature amount of the category C2 specified by the user for each type, and the weight specified as the second search condition. Update the importance of setting B and register it in the setting template. In this example, it is assumed that the weight setting UB is newly registered. As shown in FIG. 5, the weight setting UB has an importance level different from the weight setting B (the importance level of the feature quantity p2 increases, the feature quantity p3 The importance has decreased).

  When the above-described search is performed again, if the weight setting UB is selected in the selection box SB2 and other search conditions are the same, the user's feeling is reflected compared to the case of the weight setting B. In the search result display, the musical sound data sn11 is displayed at a higher rank as having a higher degree of similarity than the case shown in FIG. For example, in the case where the weight setting B indicates “middle band strength priority”, the user determines from the result of listening to the pronunciation that the middle band intensity is similar among the characteristic quantities of the first search condition. The musical sound data sn11 felt as being the musical sound data corresponding to the data is selected. Therefore, when the user selects “middle band strength priority” again as the weight setting and the other conditions are the same, the musical sound data sn11 is displayed at a higher level. You will feel that you have gained.

<Modification>
As mentioned above, although embodiment of this invention was described, this invention can be implemented in various aspects as follows.
[Modification 1]
In the embodiment described above, the first search condition is determined by the category specified by the classification template being specified by the user. However, the first search condition is determined by the user specifying each type of feature amount individually. May be. In addition, the designated content may be registered in the classification template.
Similarly, regarding the weight setting in the second search condition, the importance level for each type of feature amount may be individually designated by the user. Then, the specified content may be registered in the setting template.

[Modification 2]
In the above-described embodiment, the content specified by the user as the second search condition includes both the similarity calculation method and the weight setting. However, either one is not specified by the user in advance. It may be decided.

[Modification 3]
In the above-described embodiment, the setting template is registered with the weight setting, but may be registered by combining the similarity calculation method and the weight setting. In this case, the importance level update by the update unit 160 may be performed separately for each type of similarity calculation method.
For example, when the second search condition is determined by a combination of the Euclidean distance calculation and the weight setting B and the importance is updated, in the embodiment, only the calculation method is changed to the Mahalanobis distance calculation in the second search condition. Even so, since the weight setting is not changed, the search algorithm is determined by the updated importance. On the other hand, in this modification, the calculation method and weight setting are combined and registered in the setting template. Therefore, if the calculation method is changed, the specified weight setting is also changed to the setting that has not been updated. Is done. In this way, it is possible to separately manage the update of the importance for each calculation method.

[Modification 4]
In the above-described embodiment, the search may be further performed from the state where the search result is displayed on the display screen 131. Processing in this case will be described.
When the display screen 131 is the search result display shown in FIG. 8, a re-search is instructed by the user in a state where any musical sound data is selected by the cursor Cs <b> 1 (the re-search button b <b> 1 shown in FIG. 10 operates ), The musical sound specifying unit 150 specifies the musical sound data, and the updating unit 160 updates the importance, as in the embodiment. On the other hand, in this case, unlike the embodiment, information indicating the specified musical tone data is not displayed in the musical tone registration area WA1.

  Subsequently, the condition determining unit 120 updates the first search condition to the content of the feature amount data corresponding to the musical sound data specified by the musical sound specifying unit 150. Further, the algorithm determination unit 130 updates the weight setting of the second search condition so that the content updated by the update unit 160 is obtained. Then, the feature specifying unit 140 searches and specifies the feature data similar to the updated first search condition from the sound material DB using the updated second search condition. As a result, the display screen 131 transitions to the search result display shown in FIG. Here, it is assumed that the cursor Cs1 has selected the musical sound data sn11.

  FIG. 10 is a diagram for explaining an example of a search result display in Modification 4 of the present invention. The musical tone data displayed in the search result area SA in the state shown in FIG. 8 slides to the left as it is and is displayed in the area where the category was displayed. On the other hand, musical tone data as a new search result is displayed in the search result area SA. That is, in this search result area SA, musical tone data having feature data similar to the musical tone data sn11 is displayed as a search result. The importance level updated by the update unit 160 is used for the weight setting of the second search condition used for this search.

In this search result display, when the re-search button b2 is operated again, a re-search is performed using the first search condition and the second search condition updated as described above. On the other hand, when the determination button b1 is operated, the musical sound data specified by the musical sound specifying unit 150 (the cursor Cs1 when the determination button b1 is operated) is entered in the musical sound registration area WA1 selected by the cursor Cs2 as in the embodiment. Information indicating the musical sound data selected by (1) is displayed.
Note that the importance level may not be updated when the re-search is performed. In this case, the second search condition is not changed, and only the first search condition is changed and the search is performed again.

[Modification 5]
In the embodiment described above, the display order of the musical sound data displayed in the search result area SA represents the similarity order, but the similarity order may be expressed in a display mode other than the order. For example, the display size and density may be displayed, or the similarities may be displayed together, and the display mode may be changed according to the similarities.

[Modification 6]
In the above-described embodiment, an identification unit for identifying a user may be provided, and a setting template may be stored for each user. And the setting template corresponding to the user identified by the identification part should just be used for each process in embodiment. When information for identifying a user is input to the musical sound processing apparatus 10 by operating the operation unit 12 or the like, the identification unit may identify the user with reference to this information.
In this way, since the importance level is updated corresponding to each user, even if one musical sound processing apparatus 10 is shared by a plurality of users, the senses of other users are reflected in the search results. Can be prevented.

[Modification 7]
In the embodiment described above, the update unit 160 that updates the importance level exists, but it does not necessarily have to exist. In this case, the musical sound data displayed as the search result may be only the most similar musical sound data.

[Modification 8]
In the above-described embodiment, the musical sound data registered in the musical sound registration areas WA1, WA2,... Is used for creating a sound material track, but may be used for other purposes. For example, it may be used in a musical instrument or a sound source that generates sound using the musical sound data. When used in a musical instrument, the pitch may be changed, or musical tone data having different pitches may be stored in the storage unit 15 in advance. The musical tone data having only different pitches may not be searched for in the feature specifying unit 140. As described above, the musical sound processing apparatus 10 constituting the data search unit 100 of the present invention can be applied to musical instruments, sound sources, etc. in addition to the information processing apparatus.

[Modification 9]
Each program in the above-described embodiment is provided in a state stored in a computer-readable recording medium such as a magnetic recording medium (magnetic tape, magnetic disk, etc.), an optical recording medium (optical disk, etc.), a magneto-optical recording medium, or a semiconductor memory. Can do. In addition, the musical sound processing apparatus 10 may download each program via a network.

DESCRIPTION OF SYMBOLS 10 ... Music processing apparatus, 11 ... Control part, 12 ... Operation part, 13 ... Display part, 131 ... Display screen, 14 ... Interface, 15 ... Memory | storage part, 16 ... Sound processing part, 161 ... Speaker, 162 ... Microphone, 100 ... Data search unit, 110 ... Display control unit, 120 ... Condition determination unit, 130 ... Algorithm determination unit, 140 ... Feature identification unit, 150 ... Musical sound identification unit, 160 ... Update unit

Claims (8)

Data to be searched based on a search condition designated by the user using a database in which musical sound data indicating a musical sound waveform signal and feature data indicating a plurality of types of characteristic values of the musical sound waveform signal are registered in association with each other A search device,
Condition determining means for determining a plurality of types of feature quantities as a first search condition;
The search as a second search condition including a similarity calculation method and setting of the weight of the feature quantity from a plurality of search algorithms each indicating a search method of the feature quantity data similar to the first search condition An algorithm determining means for determining an algorithm;
Using a search algorithm indicated by the second search condition, feature specifying means for searching and specifying the feature amount data similar to the first search condition from the database;
Displaying the first search condition, the calculation method of the similarity included in the second search condition, and the search condition setting for selecting the setting of the weight of the feature amount on a display unit,
A data search apparatus comprising: display control means for displaying information indicating musical tone data corresponding to the feature data specified by the feature specifying means on the display means as a search result. A musical sound specifying means for specifying the musical sound data selected by the user from the musical sound data corresponding to the information displayed on the display means;
The search algorithm determined as the second search condition is updated based on the feature quantity relationship between the feature quantity data corresponding to the musical tone data specified by the musical tone specifying means and the first search condition. And further updating means,
The feature specifying means specifies a plurality of the feature amount data similar to the first search condition,
The display control unit is configured to display information indicating musical tone data corresponding to the plurality of feature amount data specified by the feature specifying unit based on the similarity between the feature amount data and the first search condition. The data search device according to claim 1, wherein the data is displayed on the display unit. In the plurality of search algorithms, the search algorithm showing a method of searching for feature amount data similar to the first search condition based on a distance of the feature amount for each type and a weight determined for each type. Contains
The data search device according to claim 2, wherein the update unit updates the search algorithm by changing the weight. When the musical sound data is specified by the musical sound specifying means,
The condition determining means updates the first search condition based on feature amount data corresponding to the musical sound data specified by the musical sound specifying means,
The algorithm determination means updates the search algorithm updated by the update means as the second search condition,
The feature specifying means searches and specifies the feature amount data similar to the updated first search condition by the search method indicated by the updated second search condition. The data search device according to claim 2 or 3. The data search apparatus according to any one of claims 2 to 4, wherein the update means performs the update when the second search condition is the predetermined search algorithm. When the category indicating the classification of the plurality of types of feature values is specified by the user, the condition determining unit determines the feature values of each type in the specified category as the first search condition,
The feature specifying means searches and specifies the feature amount data similar to the first search condition from the database according to the designated category.
6. A data search apparatus according to claim 1, wherein the data search apparatus is a data search apparatus. The condition determining means determines the feature amount of each of the plurality of types specified by a user as the first search condition,
The feature specifying unit searches and specifies the feature amount data similar to the first search condition from the database according to a category in which the designated feature amounts of the respective types are classified.
6. A data search apparatus according to claim 1, wherein the data search apparatus is a data search apparatus. The computer,
An apparatus for performing a search based on a search condition designated by a user using a database in which musical sound data indicating a musical sound waveform signal and feature data indicating a plurality of types of characteristic values of the musical sound waveform signal are registered in association with each other Is a program for functioning as
The computer,
Condition determining means for determining a plurality of types of feature quantities as a first search condition;
The search as a second search condition including a similarity calculation method and setting of the weight of the feature quantity from a plurality of search algorithms each indicating a search method of the feature quantity data similar to the first search condition An algorithm determining means for determining an algorithm;
Using a search algorithm indicated by the second search condition, feature specifying means for searching and specifying the feature amount data similar to the first search condition from the database;
Displaying the first search condition, the calculation method of the similarity included in the second search condition, and the search condition setting for selecting the setting of the weight of the feature amount on the display means,
A program for causing information indicating musical tone data corresponding to feature amount data specified by the feature specifying means to function as a display control means for displaying on a display means as a search result.

Images