JP6439239B2 - Performance data file search method, system, program, terminal device, and server device - Google Patents

Description

  The present invention relates to a technique for acquiring a performance data file corresponding to an audio song selected by a user.

  Nowadays, music is digitized and it is common to purchase CDs (compact discs) or download data from music sites on the Internet. The original audio data can also be created on a single computer depending on the genre. As music production and sales methods have been simplified and streamlined in this way, the life cycle of songs has become shorter and a large number of songs have been created. In addition, because of this change in the music production environment, music data can be created easily without going through a major music production company, so more and more songs are made from minor labels.

  Of these music pieces, there are songs that have songs, so that many songs are created and supplied as SMF (Standard MIDI File) data (for example, Patent Document 1). Some of them are live accompaniment or only audio data of chorus only, but for those with the function of scoring good and bad karaoke songs, a standard correct pitch is required Therefore, many have MIDI data. Although the number of audio accompaniments has increased, karaoke with SMF data is still more common because it is convenient for changing the tempo and changing the key.

JP 2012-150393 A

  On the other hand, electronic musical instruments have hit songs and classic songs as built-in song SMF data, and have a function of making a keyboard glow or lesson based on this data. Some models have more than 200 built-in songs, but most are classical songs, and only a few pop songs with a short life cycle are included. For this reason, when the user wants to play the music that each user listens to, it is often not included in the built-in music, and it is necessary to go to the music site and purchase a performance data file such as SMF data. However, some songs can be purchased, and not all songs can be purchased. As described above, even when the user wants to obtain a performance data file corresponding to a music piece desired by the user for use in an electronic musical instrument or the like, conventionally, it takes time to acquire the performance data file. There was a problem that there were many.

  An object of the present invention is to automatically obtain a performance data file corresponding to an audio song selected by a user from a performance data file database on a cloud server or the like.

In the performance data file search method in an example of the aspect, a first index search is performed to search a performance data file including a pitch data string representing the music, to which an index for identifying the music is assigned, from the performance data file database. And when the performance data file cannot be found by the first index search, the audio data extracted with the vocal corresponding to the song and the audio data without the vocal corresponding to the song are extracted. A first melody extraction is performed to extract a pitch data string representing the melody part of the song from the audio data including the melody part of the song, and the performance data file matching the extracted pitch data string is Search from the performance data file database based on one match criterion A first melody search is executed, and a performance data file corresponding to the song is acquired from the performance data file database based on the search result of at least one of the first index search and the first melody search. .

According to the present invention, it is possible to acquire performance data file corresponding to the O Dio songs from the cloud server.

It is a figure which shows the system configuration example of embodiment of this invention. It is a figure which shows the hardware structural example of a terminal device, and the functional structural example of the cloud server 102. FIG. It is a flowchart which shows the matching process of an audio music and SMF data. It is a flowchart which shows the process which adjusts SMF data to lesson data. It is a flowchart which shows a melody tone color change process. It is a flowchart which shows a genre detection process. It is a flowchart which shows the synchronization process of audio data and SMF data. It is a figure which shows the example of a data structure of an index (Index). It is a figure which shows the example of a data structure of the timbre group database for a melody timbre change.

  Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a diagram showing a system configuration example according to an embodiment of the present invention.

  The terminal device 101 is a computer having a communication function, such as a smartphone, a tablet terminal, or a personal computer. A USB (Universal Serial Bus) connector, for example, of the terminal device 101 and a MIDI (Musical Digital Interface) connector, for example, of the optical keyboard keyboard 103 can be connected by a connection cable. The terminal device 101 can communicate with the cloud server 102 as a server computer via the Internet from a wireless line, a broadband optical line, or a metal line.

  The terminal device 101 selects an audio song stored in the main body based on the user's operation, and sends a search request for SMF (Standard MIDI Format) data for karaoke corresponding to the audio song to the cloud server 102. Issued against. Search requests include those relating to audio data indexes (hereinafter also referred to as “Index”) and those relating to pitch data strings of melody data. The cloud server 102 executes a search process corresponding to the search request on the SMF database 223 for karaoke stored by itself, and extracts the corresponding SMF data. And the index of the SMF data for karaoke in the cloud server 102 are compared. When the cloud server 102 can extract the SMF data for karaoke, it transmits it to the terminal device 101.

  The terminal apparatus 101 generates optical key data by modifying the downloaded SMF data, such as a lesson function, and transmits it to the optical keyboard 103. Based on the optical key data, the user can lesson a favorite audio tune selected by the terminal device 101 on the optical keyboard keyboard 103.

  The conversion from the SMF data to the optical key data may be executed not by the terminal device 101 but by the optical keyboard keyboard 103.

  FIG. 2 is a diagram illustrating a hardware configuration example of the terminal device 101 illustrated in FIG. 1 and a functional configuration example of the cloud server 102.

  The terminal device 101 includes a CPU (Central Processing Unit) 201, a ROM (Read Only Memory) 202, a RAM (Random Access Memory) 203, an input unit 204, a display unit 205, a MIDI interface unit (hereinafter referred to as “MIDI I / F”). (Noted) 206 and the network interface unit 207 are connected to each other by a bus 208. The MIDI I / F 206 is connected to an external (or built-in) optical keyboard keyboard 103 (see FIG. 1).

  The CPU 201 performs overall control of the terminal device 101 in accordance with a control program stored in the ROM 202.

  The ROM 202 stores a control program as a normal communication device or computer, and a control program for generating lesson data shown in each flowchart described later.

  The RAM 203 stores various control data necessary for execution of the control program and audio song data, and temporarily stores SMF data acquired from the cloud server 102 or the like.

  The input unit 204 detects an input operation by a user using a touch panel display (in the case of a smartphone or tablet terminal), a keyboard or a mouse (in the case of a personal computer), or various switches, and notifies the CPU 201 of the detection result.

  The display unit 205 outputs data sent under the control of the CPU 201 to a liquid crystal display or the like. For example, the display unit 205 displays a screen for audio song selection.

  The MIDI I / F 206 supplies the optical key data in the MIDI format generated from the SMF data to the optical keyboard keyboard 103.

  The network interface unit 207 has a wireless communication function or a broadband communication function, for example. The network interface unit 207 accesses the SMF database 223 for karaoke in the cloud server 102 while communicating with the network interface unit 221 in the cloud server 102 and notifies the CPU 201 of the result.

  In FIG. 2, the cloud server 102 includes a karaoke SMF database 223. The cloud server 102 also has functions of a network interface unit 221 that communicates with the network interface unit 207 in the terminal device 101 and a database management unit 222 that controls access to the karaoke SMF database 223. The function of the database management unit 222 is realized by a CPU, a ROM, a RAM, an external storage device, and the like that constitute a normal computer, although not particularly illustrated. The CPU implements the function of the database management unit 222 by executing a database management program loaded into the RAM from the ROM or the external storage device. Further, the SMF database for karaoke 223 is stored in, for example, an external storage device.

  The operation of the terminal device 101 according to the present embodiment is performed by the CPU 101 reading out from the ROM 102 and sequentially executing a control program loaded with the function processing of each unit of FIG. 1 realized by the flowcharts of FIGS. Realized. The program may be provided via a portable recording medium (not shown), or may be acquired from the network via a network interface unit 207 (not shown).

  Hereinafter, the operation of the embodiment of the present invention will be described along a flowchart. In the description of each flowchart, each part in FIG. 2 is referred to as needed.

  FIG. 3 is a flowchart showing a matching process between an audio song and SMF data. In this process, the CPU 201 executes a terminal-side audio song and SMF data matching processing program stored in the ROM 202, and the database management unit 222 in the cloud server 102 performs a karaoke SMF database in accordance with a request from the terminal device 101. 223 is a process of executing a matching processing program for audio music on the cloud side that accesses the H.223 and SMF data. In FIG. 3, the steps in dark portions are processes executed by the database management unit 222 in the cloud server 102, and the other steps are processes executed by the CPU 201 in the terminal device 101.

  First, the user selects, from the input unit 204 of the terminal device 101, an audio song to be lessond using the optical keyboard 103 on the built-in music database or on a music providing site on the Internet. As a result, the audio tune selection information is notified from the input unit 204 to the CPU 201 (step S301).

  The CPU 201 acquires index information (hereinafter simply referred to as “index” or “Index”) of the audio song selected in step S301 from a music providing site on the Internet via the built-in music database or the network interface unit 207. To do. Then, the CPU 201 stores the index in a variable A_Index on the RAM 203. In the following description, when “A_Index” is described, it indicates a variable name and is also used in the meaning of the index value itself. The CPU 201 issues a first search request including A_Index and transmits it to the cloud server 102 via the network interface unit 207 (step S302 above). The first search request is received by the database management unit 222 via the network interface unit 221 in the cloud server 102.

  Upon receiving the first search request, the database management unit 222 accesses the song title index (Index) in the karaoke SMF database 223 and searches for an index that completely matches the A_Index notified by the terminal device 101. An index search is performed (step S303). FIG. 8 is a diagram showing a data configuration example of the song title index in the SMF database 223 for karaoke. An index that matches A_Index can be searched using the item “Song Title: Kana” in this data configuration example.

  The database management unit 222 determines whether an index that completely matches A_Index is found in the song title index (step S304).

  If the first index search is successful and the determination in step S304 is YES, the database management unit 222 proceeds to the process in step S316. Here, the database management unit 222 selects the SMF data of the main body corresponding to the found index from the SMF database 223 for karaoke. The database management unit 222 transmits the SMF data to the terminal device 101 via the network interface unit 221. As a result, the CPU 201 receives the SMF data via the network interface unit 207, stores it in the RAM 203, and ends the process of the flowchart of FIG.

  When the first index search fails and the determination in step S304 is NO and the database management unit 222 returns to that effect to the terminal device 101, the CPU 201 executes the following operation. The CPU 201 stores audio data with vocals corresponding to the audio song selected in step S301 and its karaoke version (instrument version) in the music database stored in the RAM 203 in the terminal device 101 or in particular an external storage device (not shown). It is determined whether or not there is a document version (step S305). In many pieces of music provided as audio data, so-called karaoke version data without vocals is often included as a set. In this embodiment, when the SMF data is not found on the cloud server 102, this karaoke version is used.

  When the karaoke version audio data is found and the determination in step S305 is YES, the CPU 201 combines the start time of the audio data time signal of the karaoke data without vocals from the time signal of the audio data with vocals. By subtracting above, the melody portion of the audio data is extracted as the first melody data. Melody extraction from audio data can be performed to some extent by center cut and vocal range collation, but it is not perfect, for example, it is difficult to accurately determine the difference between a chorus part and a melody part. However, in the case of a song with a karaoke version, both audio data with vocals and audio data without vocals are created from the same sound source, so the melody extraction of only the vocal part should be performed with high accuracy by the subtraction process described above. Can do. The CPU 201 executes the pitch analysis / extraction analysis on the first melody data extracted in this manner, whereby the first sound composed of the pitch sequence of the first melody data (melody part). A high data string can be extracted (step S306).

  After step S306, the CPU 201 issues a second search request including the first pitch data string extracted there, and transmits it to the cloud server 102 via the network interface unit 207 (step S307). Note that the CPU 201 also includes the A_Index acquired in step S302 in the above-described second search request for processing in step S313 to be described later in the cloud server 102. The second search request is received by the database management unit 222 via the network interface unit 221.

When receiving the second search request, the database management unit 222 performs the following operation. The database management unit 222 uses the first match criterion to match each pitch of the received note-on MIDI data sequence of the melody of each SMF data in the SMF database 223 for karaoke with the received first pitch data sequence. A first melody search is performed (step S308). In this case, for example, it is only necessary to compare about four measures of the beginning. Specifically, as shown in FIG. 8, a melody pitch data sequence from the start of the melody is prepared as data of “starting melody” in the music index, and this pitch data sequence and the reception are received. The first pitch data string is compared.
In the present embodiment, as shown in FIG. 8, both “absolute pitch” and “relative pitch” are prepared as data for “starting melody”. (Absolute pitch) uses pitch data represented by an absolute value, and (Relative pitch) uses a difference between the current pitch and the pitch data immediately before. In songs such as enka, in consideration of the key when singing, the original is often changed to a key such as +1 or -1 from the original, and a relative pitch is necessary for such a song.
The SMF data in the cloud server 102 has a data format determined by the company that provides the data. For example, in most SMF data, the melody is on 4ch and the bass is 2ch. For this reason, the pitches of the note-on MIDI data strings that are output from 4ch are compared, and if candidates are gathered, they are considered to the end. In addition, the karaoke SMF database 223 in the cloud server 102 often has a melody out-of-melody note-on MIDI data string in the database index. Then, song candidates can be extracted at high speed. Furthermore, the musical score may be delicate depending on how it is sung, and may differ depending on the SMF data producer. For this reason, if there is a timing and pitch of a sound with a long note value, the accuracy rate will be higher, and if the rhythm of the same tone is different or if the pitch of the hook is slightly different, etc. For example, a scoring method may be adopted in which, for example, 80% is met and the first match criterion is satisfied if they are almost matched.

  The database management unit 222 in the cloud server 102 determines whether or not SMF data that matches the first match criterion is found as a result of the first melody search (step S309).

  If the first melody search is successful and the determination in step S309 is YES, the database management unit 222 proceeds to step S316, selects and reads out the SMF data of the main body in the SMF database 223 for karaoke that matches the start. Then, the database management unit 222 notifies the terminal device 101 of the SMF data via the network interface unit 221. As a result, the CPU 201 receives the SMF data via the network interface unit 207, stores it in the RAM 203, and ends the process of the flowchart of FIG.

  If the first melody search fails and the determination in step S309 is NO, the database management unit 222 proceeds to the second index search process in step S312.

  In the terminal device 101, if the determination in step S305 described above is NO, the CPU 201 first has audio data with vocals corresponding to the audio song selected in step S301 not in the music database in the terminal device 101. You get it from a music provider on the Internet. Next, the CPU 201 extracts the melody portion as second melody data from the acquired audio data with vocals. Here, the karaoke version of the audio data cannot be used. Therefore, CPU201 extracts the 2nd melody data by acoustic analysis processing. Specifically, for example, the CPU 201 extracts data localized at the center from the audio data by a center cut technique, and is a sound at the center, not a drum, not a bass range, and a chord pressed simultaneously ( A sound that is not a chord part and falls within the melody range is extracted as second melody data. The CPU 201 executes the pitch analysis / extraction analysis on the second melody data extracted in this way, whereby the second sound composed of the pitch string of the second melody data (melody part). A high data string is extracted (step S310).

  After step S310, the CPU 201 issues a third search request including the second pitch data string extracted there and the A_Index acquired in step S302, and transmits the third search request to the cloud server 102 via the network interface unit 207 ( Step S311). The third search request is received by the database management unit 222 via the network interface unit 221.

  When the determination in step S309 is NO or when the third search request (step S311) is received from the terminal apparatus 101, the database management unit 222 performs the following operation. The database management unit 222 accesses the song name index (Index) (see FIG. 8) in the SMF database 223 for karaoke, and searches for an index that partially matches the A_Index notified by the terminal device 101, and performs a second index search. Implement (step S312). In this case, partial match means that if one character string matches part of the other character string, half-width and full-width characters, uppercase and lowercase letters, the presence or absence of special characters such as long marks and heart symbol used for song titles, etc. It includes the case of matching when not distinguished from each other and the case of matching according to a combination of both.

  The database management unit 222 determines whether an index that partially matches A_Index is found in the song title index (step S313).

  When the second index search is successful and the determination in step S313 is YES, the database management unit 222 selects the SMF data group of the main body corresponding to the found index candidate group from the SMF database 223 for karaoke. The database management unit 222 uses the first or second pitch data sequence received from the terminal device 101 in step S307 or S311 as the sound of the note-on MIDI data sequence from which each melody of the selected SMF data group starts. Compare with high (step S314). As a result, the database management unit 222 uses the pitch data string of the melody data received from the terminal device 101 and the second pitch data string of the note-on MIDI data string of the outgoing melody out of the SMF data with partially matching indexes. A second melody search is performed to search for songs that match on the match criteria (step S315). In this case, since the SMF data is narrowed down to some extent by the partial matching of the index, the second matching criterion is set lower than the matching degree of the first matching criterion in the first search in step S308. (The judgment condition may be relaxed).

  If the second melody search is also succeeded after the second index search, and the determination in step S315 is YES, the database management unit 222 has the pitch data string of the melody that has a partially matched index and the start melody. For example, the SMF data having the highest matching degree among the SMF data matched by the second matching criterion is selected from the SMF database 223 for karaoke and read. Then, the database management unit 222 notifies the terminal device 101 of the SMF data via the network interface unit 221 (step S316). As a result, the CPU 201 receives the SMF data via the network interface unit 207, stores it in the RAM 203, and ends the process of the flowchart of FIG.

  In the database management unit 222, when the partially matching index is not found and the second index search fails and the determination in step S313 is NO, or when the second index search is successful but the second melody search is performed. If the determination fails in step S315, the CPU 201 displays a comment without corresponding SMF data on the display unit 205 and ends the process (step S317). Many pieces of karaoke data are created and added every day. However, in the case of a newest song or a less famous song, it is not included in the SMF database 223 for karaoke of the cloud server 102 and there is no SMF data. Sometimes.

As described above, in the present embodiment, first, the first index is searched from the SMF database 223 for karaoke on the cloud server 102 for SMF data whose index completely matches the index (A_Index) of the song selected by the user. A search is performed.
If the melody-free audio data corresponding to the audio data of the selected song can be acquired when the first index search fails, the melody portion of the audio data is subtracted from the audio data. Are extracted as first melody data, and a first pitch data sequence consisting of a pitch sequence of the first melody data is extracted. Then, the SMF data that matches the first matching criteria in the first pitch data string and note-on MIDI data string (for example, four bars from the beginning) is searched from the SMF database 223 for karaoke. A first melody search is performed.
When the melody-free audio data cannot be acquired, the melody portion is extracted from the audio data as the second melody data by the sound analysis process such as center cut, and the second melody data pitch second row is formed. A pitch data string is extracted.
When the first melody search fails, or when the second melody data with low reliability that cannot be obtained without the melody is obtained, the SMF whose index (A_Index) and the index partially match are obtained. A second index search is performed in which data is searched from the SMF database 223 for karaoke on the cloud server 102. If the second index search is successful, the second match, which is looser than the first match criterion at the beginning of each pitch of the first or second pitch data string and the note-on MIDI data string, is further generated. A second melody search is performed in which SMF data matching the criteria is searched from the SMF database 223 for karaoke.
When the first index search, the first melody search, or the second melody search is successful, the performance data file corresponding to the song is acquired from the performance data file database.
Thus, according to the present embodiment, SMF data (performance data file) corresponding to the audio song selected by the user is automatically obtained from the SMF database 223 for karaoke (performance data file database) on the cloud server 102 or the like. Can be obtained.

  When SMF data for karaoke corresponding to the audio data selected by the user in the cloud server 102 can be acquired and sent to the terminal device 101, the SMF data for karaoke is a lesson for the optical keyboard 103 (FIGS. 1 and 2). Adjusted to data. In the SMF data for karaoke, the melody sound is generated only as a guide sound, so that it is not suitable for a melody lesson on the optical keyboard 103, for example. In the case of lesson data originally built in the optical keyboard 103 or the like, the melody sound is emphasized to some extent so that the user can easily do the lesson. However, as in the present embodiment, for example, when the latest or user-desired karaoke SMF data can be acquired from the cloud server 102 and used for lesson data, the karaoke SMF data Processing to adjust the melody sound so that it becomes a melody sound suitable for the lesson data is essential.

  FIG. 4 is a flowchart showing a process for adjusting SMF data to lesson data. In this process, the CPU 201 executes a processing program for adjusting the SMF data stored in the ROM 202 to lesson data.

  In this case, first, the CPU 201 specifies the melody data in the fourth channel out of the total 16 channels (channels) as the melody of the selected song (step S401).

  The CPU 201 adjusts the volume of the specified fourth channel (step S402). Since the SMF data for karaoke is for karaoke, the volume of the melody is set low as a guide melody. Therefore, the CPU 201 performs adjustment to increase the velocity of each MIDI data in the 4th channel to 1.4 times, for example. Note that the setting value of the magnification may be appropriately changed by the user from the input unit 204, for example, depending on the relationship between the sound source and the data.

  Next, the CPU 201 adjusts the volume of the fifth channel (step S403). Normally, the SMF data for karaoke contains a chorus part in the 5th channel. For example, when singing in a group, the 5th channel may have a melody part that covers a melody. Therefore, also for this part, the CPU 201 performs adjustment to increase the velocity of each MIDI data in the 5th channel to 1.2 times, for example. The setting value of the magnification may be appropriately changed by the user from the input unit 204, for example, depending on the relation between the sound source and the data, but may be set a little conservatively than the setting value of the melody.

  Next, the CPU 201 executes a melody tone color changing process (step S404). Here, an adjustment is performed to select an optimal melody tone color so as not to be obscured by a plurality of high-volume accompaniment sounds by a combination of a plurality of high-volume accompaniment tone colors. In the original karaoke data, the main purpose of the melody sound was to guide it, so a thin melody was selected, and an inconspicuous tone such as a continuous sound like a clarinet or a decaying sound like a vibraphone was selected. There are many cases. If this is done, the melody will not stand out and the accompaniment will be erased by simply raising the volume. Therefore, in the present embodiment, this process is executed in order to make the melody tone color stand out from the accompaniment tone color at the time of a lesson. When a lesson is performed by shining the keyboard with the optical keyboard 103, the melody sound is also generated, so the above processing is necessary. When the optical keyboard 103 was manufactured, the preset sound was originally set as such. However, when using SMF data for karaoke as in this embodiment, it is particularly difficult for beginners to change the tone. , Such measures are necessary. Details of this processing will be described later with reference to the flowcharts of FIGS.

  Next, the CPU 201 performs bar adjustment (step S405). Some SMF data for karaoke starts with the third beat or starts with the fourth beat, even if the song originally starts at the beginning of a bar. When such SMF data is used as lesson data, bar adjustment is required. Therefore, in step S405, the CPU 201 looks at the rhythm of melody notes, for example, up to 8 beats, and if the head of the measure in the SMF data to be the lesson data is not shifted or is a so-called “outact”, the correct beat is obtained. Judge whether it is. Then, based on the determination result, the CPU 201 scores the melody note rhythm using a strong / weak beat pattern that matches the time signature, and performs bar adjustment.

  Next, the CPU 201 adjusts the duration of the melody (step S406). For example, in the case where the keyboard is lit like the optical key of the optical keyboard 103, the duration of the keyboard will continue to shine unless there is an appropriate interval when the same sound continues. For this reason, when the same sound continues, the CPU 201 adjusts the duration, that is, the arrangement of the lengths of the musical sounds that are continuously generated so as to leave 10/480 (one beat), for example.

  Next, the CPU 201 executes melody quantization (step S407). If the melody sound is finer than necessary, the lesson will be difficult. Therefore, the CPU 201 quantizes the note-on MIDI data of the melody sound with a certain degree of roughness, for example, a 32nd note or a triplet (triplet).

  Next, the CPU 201 performs a shift adjustment at the time of playing the melody (step S408). If there is an accompaniment sound that is too fine at the time of the melody pronunciation during the lesson, an obligado shift occurs and it sounds like a dissonance. In step S407 described above, if there is a sound that is too early when the melody is generated, the sound is quantized. However, for tree chimes and glissando parts, it is necessary to take measures to avoid this handling from the timbre and the state before and after. Therefore, timbres such as tree chimes are removed from the timbre names and other sounds are adjusted.

  Next, the CPU 201 executes a musical style classification process using an SMF marker (step S409). If the song is too long, it will be tired of the lesson, so it is better to divide it into phrases of a certain length so that you can take lessons. This break is called a phrase marker. In this embodiment, when the phrase marker is described in the lesson data, the CPU 201 describes the phrase marker at a place where the marker is originally included in the SMF data. When creating SMF data for karaoke, there are many cases in which markers are put in accordance with musical formulas such as A melody, B melody, chorus and B chorus. Therefore, if a marker “A melody” is described in the SMF data, it is effective to regard it as a large phrase in the lesson.

  In step S409 described above, the time interval between markers extracted from the SMF data may be long. Therefore, when the phrase marker is described, if the marker interval is longer than, for example, a predetermined measure, the CPU 201 divides the marker interval into 1/2 or 1/4, creates a phrase marker that is easy to learn, and creates lesson data. (Step S410).

  Finally, the CPU 201 executes a lesson code creation process (step S411). Here, the CPU 201 extracts chords from the part that is the chord part of the accompaniment sound in the SMF data, determines the chord, and creates a chord part for the lesson.

  FIG. 5 is a flowchart showing the melody tone color changing process in step S404 of FIG.

  First, the CPU 201 executes a genre detection process for SMF data (step S501). Since the accompaniment organization is determined by the genre, the accompaniment sound can be easily understood by looking at the genre. If the accompaniment sound is known, the melody sound that stands out with respect to the accompaniment sound color at the lesson can be determined. Details of this processing will be described later using the flowchart of FIG.

  Next, the CPU 201 stores the genre detected in step S501 in a variable ToneG on the RAM 203 (step S502).

  Subsequently, the CPU 201 selects four parts having a high volume other than the bass and drums among the parts appearing together in the fourth channel where the melody is sounded with the SMF data. Then, the CPU 201 stores each part name in variables Part1, Part2, Part3, and Part4 on the RAM 203 (step S503).

  Next, the CPU 201 refers to the item corresponding to the genre indicated by ToneG on the timbre group database for changing the melody timbre stored in the ROM 202 or RAM 203 (step S504). FIG. 9 is a diagram showing a data configuration example of a timbre group database for changing a melody timbre. For each genre such as “Swing”, “Latin / European”, “Suffle / blues”, “Rock / Pops”, “Funk / DancePop”, and “Techno / Euro”, one column of the table in FIG. 10 is assigned. In the column, accompaniment tone color names that are frequently used as accompaniment parts of the genre are registered in the upper four rows, and melody tone color names that are frequently used as melody candidates (melody parts) are registered in the lower three rows. As the melody timbre name, a melody timbre name that stands out from the accompaniment tone color of the accompaniment part in the genre is registered.

  Subsequently, the CPU 201 displays the accompaniment parts Part1 to 4 in the current SMF data and the accompaniment parts registered in the genre item corresponding to ToneG on the timbre group database for changing the melody timbre illustrated in FIG. In comparison, it is determined whether there are two or more accompaniment part names (step S505). In the accompaniment part, Guita sees it as a tone group that includes both acoustic guitars and electric guitars. Also, the reason why the determination criterion is set to two or more parts is that Guitar is included in any genre, so that the genre cannot be identified with only one. In other words, we are looking at whether there is a match other than the guitar.

  If the determination in step S505 is YES, the CPU 201 refers to a melody candidate in the corresponding genre in the timbre group database (step S511). For example, when the genre is Swing and the accompaniment part of the SMF data is Guital, Piano, and Strings, the melody candidates Sax, Trumpet, and Organ illustrated in FIG. 9 are selected.

  If one melody candidate can be selected, the CPU 201 determines whether the melody candidate sound does not overlap with the accompaniment sounds Part1 to Part4 (step S512).

  If the melody candidate sound is the accompaniment sounds Part 1 to 4 and the determination in step S512 is NO, the CPU 201 selects the next melody candidate from the melody candidate items of the genre in the timbre group database (step S513). Then, the CPU 201 returns to the determination in step S512.

  If the melody candidate sound is not related to the accompaniment sounds Part1 to 4 and the determination in step S512 is YES, the melody candidate sound is determined as the melody sound for the lesson and the melody change is executed (step S514). Specifically, the 4ch melody tone color in the SMF data is rewritten to the determined melody tone. Thereafter, the CPU 201 ends the melody tone color changing process in step S404 of FIG. 4 shown in the flowchart of FIG.

  If the determination in step S505 is NO, the CPU 201 corrects and resets the value of the variable ToneG to the genre right next to the genre indicated by ToneG on the tone group database illustrated in FIG. 9 (step S506). .

  Subsequently, the CPU 201 determines whether or not the determination of the genre indicated by ToneG has been completed (step S507).

  If it is not completed and the determination in step S507 is NO, the CPU 201 returns to the determination in step S505.

  If the determination is made in step S507, the CPU 201 forcibly changes the fourth channel melody sound to a clarinet sound (step S508).

  The CPU 201 determines whether or not there is a clarinet sound in Part 1 to 4 (step S509).

  If the determination in step S509 is NO, the CPU 201 ends the melody tone color changing process in step S404 of FIG. 4 shown in the flowchart of FIG.

  If the determination in step S509 is YES, the CPU 201 forcibly changes the 4th channel melody sound to a harmonica sound (this sound does not overlap with any accompaniment sound), and is shown in the flowchart of FIG. The melody tone color changing process in step S404 is terminated.

FIG. 6 is a flowchart showing the genre detection process in step S501 of FIG. As described above, in order to change the melody tone for lessons, it is desirable that the accompaniment tone color does not overlap with the accompaniment tone color, but the accompaniment tone is determined to some extent by the music genre. The genre is analyzed.

  There are two easy-to-understand methods for identifying genres. Look at the bass or the drum. In the case of drums, there are standard sets, jazz sets, and room sets depending on the tone of the drum, but jazz sets are often used for jazz, and standard sets are used for various genres, but room sets are modern. I use it for laps. The most obvious is the bass tone. It is important to know what the bass tone is and whether the rhythm is jumping.

  In FIG. 6, the CPU 201 first determines whether or not the base is a wood base (wood bass) (step S601).

If the determination in step S601 is YES, the CPU 201 determines whether or not the rhythm is triplet (step S602). Whether or not it is a triple connection is, for example, drum performance data. For example, in the case of a 16th note system, if one beat is determined to be 480 clocks, for example, it is 0, 120, 240, 360 Thus, the rhythm is engraved with a clock count divided by 2 (there is a drum sound). On the other hand, in the case of a triple connection, the rhythm is engraved with a clock count divided by 3, such as 0, 160, and 320. Therefore, by determining the clock count of the rhythm, it is determined whether or not there is a triple connection.
Since the drum part has repeated striking in the intro and the skeleton of the basic rhythm is difficult to see, it is better to examine whether the rhythm where the melody comes out, particularly the part where the A melody marker is a triplet.

  If the determination in step S602 is YES, the CPU 201 determines the genre as a Swing timbre group (step S603), and if NO, the genre is determined as a Latin / European timbre group (step S604), and is shown in the flowchart of FIG. The genre detection process in step S501 in FIG.

  When the determination in step S601 is NO, the CPU 201 determines whether or not the base is an electret base (ElectBass) or a guitar (step S605).

  If the determination in step S605 is YES, the CPU 201 determines whether or not the rhythm is triple linked (step S606).

  If the determination in step S606 is YES, the CPU 201 determines the genre as a Shuffle / Blue tone group (step S607), and if NO, the genre is determined as a Rock / Pop tone group (step S608). The genre detection process of step S501 shown in FIG.

  When the determination in step S605 is NO, the CPU 201 determines whether the base is a synthesizer base (Synth Bass) (step S609).

  If the determination in step S609 is YES, the CPU 201 determines whether or not the rhythm is triple linked (step S610).

  If the determination in step S610 is YES, the CPU 201 determines the genre as a Funk / DancePop tone group (step S611), and if NO, the genre is determined as a Techno / Euro tone group (step S612). The genre detection process of step S501 shown in FIG.

  If the determination in step S609 is also NO, the CPU 201 determines the genre as another timbre group (step S613), and ends the genre detection process in step S501 of FIG. 5 shown in the flowchart of FIG.

  As described above, in this embodiment, when the cloud server 102 can acquire SMF data for karaoke corresponding to the audio data selected by the user and sent to the terminal device 101, the SMF data for karaoke is transmitted to the optical keyboard. It is possible to adjust to lesson data suitable for the lesson for the keyboard 103 (FIGS. 1 and 2). In particular, in this embodiment, by detecting the genre from the SMF data, the timbre group database is referred to determine a group of accompaniment sounds corresponding to the genre. It is possible to determine the melody sound from the melody candidates in the timbre group database so that the timbre is more attractive.

  In addition to this embodiment, an optimal melody sound may be set for each part such as A melody and B melody.

  Furthermore, in this embodiment, a timbre group database for genre estimation and melody timbre change is used. However, genres may be written in SMF data, audio data indexes, etc. (for example, as shown in FIG. 8). In this index, “popular (piano)” registered in the “comment” item), a timbre group database in which these genres and melody sounds are written may be prepared.

  The embodiment described above assumes that the SMF data is searched from the audio data, the matched SMF data is adjusted to lesson data, and the lesson is performed using the optical keyboard 103 or the like. On the other hand, by using only the melody of the matched SMF data and using the original audio data for accompaniment and the like, the realistic feeling of the performance lesson can be improved. For that purpose, it is necessary to synchronize audio data and SMF data.

  FIG. 7 is a flowchart showing a synchronization process between audio data and SMF data in the case where only the melody uses the SMF data and the accompaniment uses the original audio data as another embodiment of the present invention. The system configuration and hardware configuration of this embodiment are the same as the examples shown in FIGS.

  First, similarly to step S305 in FIG. 3, the CPU 201 determines whether or not there is audio data with vocals corresponding to the currently selected audio song and its karaoke version in the song database in the terminal device 101. Determination is made (step S701).

  If the determination in step S701 is YES, in the same manner as in step S306 in FIG. 3, the time signal of karaoke data without vocals is combined with the start time of both from the time signal of audio data with vocals. By subtracting, melody data is extracted (step S702).

  The CPU 201 matches the on-timing of the melody data of the SMF data with the on-timing of the melody data extracted in step S702 (step S703). Since the melody data is obtained by the difference between the audio data and the karaoke data, there is no sound until the melody data starts after the start of the audio data. Accordingly, since the start point of the melody data, that is, the on timing, can be known how many seconds after the start of the audio data, the on timing of the melody data in the SMF data is matched therewith.

  If the determination in step S701 is no, the CPU 201 searches for a characteristic part in the audio data (step S704). For example, it is a characteristic part in which the sound range is high or low at the beginning of the audio data, or the pan (sound image localization) is extreme.

  The CPU 201 determines whether there is a corresponding part as a result of the search in step S704 (step S705).

  If the determination in step S705 is yes, the CPU 201 determines whether or not the part of the SMF data matches the corresponding part of the audio data (step S706).

  If the determination in step S706 is YES, the CPU 201 synchronizes the SMF and the audio data by matching the corresponding part of the SMF data with the on timing of the corresponding part of the audio data.

  If the determination in step S705 or S706 is NO, the same processing as in step S310 in FIG. 3 is executed. That is, the CPU 201 extracts data localized at the center from the audio data by the center cut method, and the sound at the center is not a drum, is not a bass range, is not a chord (chord part) pressed simultaneously. Sounds that fall within the melody range are extracted as melody data. The CPU 201 extracts the pitch data string of the melody data (melody part) by executing pitch analysis / extraction analysis on the melody data extracted in this way (step S708).

  The CPU 201 determines whether or not the pitch data sequence of the melody data extracted in step S708 includes a portion that matches each pitch of the note-on MIDI data sequence of the SMF melody part (step S709).

  If the determination in step S709 is YES, the SMF data is put on the audio data on timing extracted in step S708 (step S710).

  After the processing of step S703, S707, or S710, the CPU 201 ends the processing of the flowchart of FIG. 7, adjusts the SMF data to lesson data by the processing of the flowchart illustrated in FIG. 4 to FIG. The melody data in the middle is transferred to the optical keyboard 103 as melody sound and audio data as accompaniment sound, and the user is given a lesson.

  If the determination in step S709 is NO, since the synchronization cannot be performed, a comment accompanied by SMF data is displayed on the display unit 205 instead of the audio accompaniment, and the processing of the flowchart of FIG. Subsequent processes such as adjustment of lesson data are as described above.

  As described above, the process of synchronizing the audio data and the SMF data is not limited to the creation of lesson data using the optical keyboard 103 or the like, and it is desired to perform with live accompaniment or to sing a song. In addition, it can be used to synchronize audio data with melody data.

  According to each of the above-described embodiments, it is possible to easily and easily operate by acquiring and adjusting SMF data in comparison with karaoke data stored in the cloud and karaoke data stored in the terminal device 101. Songs can be used as lesson data such as optical keys.

  In addition, by synchronizing the audio data and the SMF data, it is possible to use the optical key to give a lesson as a live sound as an accompaniment sound.

Each of the embodiments described above is characterized in that it can be adjusted for use as lesson data by specifying data by collating audio song data held by the user with the SMF index in the cloud data. Therefore, there are the following effects.
(1) A user can lesson a song he / she wants to play.
(2) Since it is adjusted so that it can be used for lessons, it can be used for lessons with optical keys.
(3) Phrase practice is possible because the phrase is divided using the SMF data marker.
(4) Since the genre analysis of the SMF data is performed and the melody sound is changed to match this, it can be played with the sound that matches the accompaniment.
(5) Since the SMF and the audio data can be synchronized, the accompaniment can be played with the audio data.

Regarding the above embodiments, the following additional notes are disclosed.
(Appendix 1)
Performing a first index search for searching a performance data file including a pitch data string representing the music to which an index corresponding to an index for identifying a music supplied from the outside is assigned;
First melody extraction for extracting a pitch data string representing a melody portion of the song from audio data including at least the melody portion of the song when the performance data file cannot be found by the first index search Run
Performing a first melody search for searching the performance data file including the extracted pitch data string from the performance data file database based on a first matching criterion;
Obtaining a performance data file corresponding to the song from the performance data file database based on a search result of at least one of the first index search and the first melody search;
The performance data file search method characterized by the above-mentioned.

(Appendix 2)
The supplementary note 1, wherein the performance data file to which the index that completely matches the index of the supplied music is retrieved from the performance data file database as the first index search. Performance data file search method.

(Appendix 3)
As the melody extraction, apart from the audio data, it is determined whether or not melody-free audio data that does not include the melody portion of the song can be obtained,
The performance according to claim 1 or 2, wherein when it is determined that the audio data without melody can be acquired, the melody portion of the audio data is extracted by subtracting the audio data without melody from the audio data. Data file search method.

(Appendix 4)
When the performance data file cannot be found by the first melody search, the performance data file group to which an index partially matching the index of the audio data is assigned is searched from the performance data file database. Perform an additional index search of
When the group of performance data files can be searched by the second index search, the extracted pitch data string matches with the second matching criterion with a judgment condition eased compared to the first matching criterion. 4. The performance data file search method according to any one of appendices 1 to 3, further comprising performing a second melody search for searching a performance data file including a pitch data string from the group of performance data files.

(Appendix 5)
5. The performance data file search method according to appendix 4, wherein the performance data file is acquired from the performance data file database when the performance data file can be searched by the second melody search.

(Appendix 6)
If it is determined that the melody-free audio data cannot be obtained, the melody part is extracted from the audio data by an acoustic analysis process,
4. The performance data file search method according to appendix 3, wherein a pitch data string representing the melody portion of the song is extracted from the melody portion of the extracted audio data.

(Appendix 7)
A second index search for searching the performance data file database for the group of performance data files to which an index that partially matches the index of the audio data is provided;
If the second index search is successful, the performance data includes the pitch data string that matches the extracted pitch data string with a second matching criterion that is more relaxed than the first matching criterion. The performance data file search method according to appendix 6, further comprising performing a second melody search for searching for a file from the group of performance data files.

(Appendix 8)
The performance data file search method according to appendix 7, wherein when the performance data file can be searched by the second melody search, the performance data file is acquired from the performance data file database.

(Appendix 9)
The performance data file is a standard MIDI file in MIDI format.
The performance data file search method according to any one of appendices 1 to 8, characterized in that:

(Appendix 10)
Index search means for searching a performance data file including a pitch data string, which is assigned an index for identifying a desired song and includes the pitch data string representing the song, from a performance data file database;
A melody extracting means for extracting a pitch data string representing a melody portion of the song from audio data including at least the melody portion of the song when the performance data file cannot be searched by the index search means;
Melody search means for searching the performance data file including the extracted pitch data string from the performance data file database based on a first match criterion;
An acquisition means for acquiring a performance data file corresponding to the song from the performance data file database based on a search result of at least one of the index search means and the melody search means;
A performance data file retrieval apparatus comprising:

(Appendix 11)
On the computer,
Performing a first index search that searches a performance data file database for a performance data file that is provided with an index corresponding to an index for identifying a desired music and includes a pitch data string representing the music;
If the performance data file cannot be searched by the index search, first melody extraction is performed to extract a pitch data string representing the melody portion of the song from audio data including at least the melody portion of the song. Steps,
Performing a melody search for searching the performance data file including the extracted pitch data string from the performance data file database based on a first matching criterion;
Obtaining a performance data file corresponding to the song from the performance data file database based on a search result of at least one of the first index search and the first melody search;
A program that executes

(Appendix 12)
Index search means for searching a performance data file including a pitch data string that is assigned an index for identifying a desired song and includes the pitch data string representing the song, and a received pitch data string Based on the search result of at least one of the melody search means, the index search means and the melody search means for searching the performance data file from the performance data file database based on a first match criterion, A terminal device connected to a server device having acquisition transmission means for acquiring and transmitting a corresponding performance data file from the performance data file database,
When the performance data file cannot be searched by the index search means, a pitch data string representing the melody portion of the song is extracted from audio data including at least the melody portion of the song and transmitted to the server device. Melody extraction and transmission means;
Receiving means for receiving a performance data file transmitted from the server device;
The terminal device characterized by having.

(Appendix 13)
A server device connected to a terminal device having melody extraction and transmission means for extracting and transmitting a pitch data string representing the melody portion of the song from audio data including at least the melody portion of the song,
Index search means for searching a performance data file including a pitch data string, which is assigned an index for identifying a desired song and includes the pitch data string representing the song, from a performance data file database;
Melody search means for searching the performance data file including the pitch data string transmitted from the terminal device from the performance data file database based on a first match criterion;
An acquisition means for acquiring a performance data file corresponding to the song from the performance data file database based on a search result of at least one of the index search means and the melody search means;
The server apparatus characterized by having.

101 Terminal Device 102 Cloud Server 103 Optical Keyboard Keyboard 201 CPU
202 ROM
203 RAM
204 Input unit 205 Display unit 206 MIDI interface unit (MIDI I / F)
207, 221 Network interface unit 208 Bus 222 Database management unit 223 Karaoke SMF database

Claims (11)

Performing a first index search that searches a performance data file database for a performance data file including a pitch data string representing the music, to which an index for identifying a song is assigned;
If the performance data file cannot be found by the first index search, the melody of the song extracted based on the audio data with vocal corresponding to the song and the audio data without vocal corresponding to the song Performing a first melody extraction for extracting a pitch data string representing a melody portion of the song from the audio data including the portion;
Performing a first melody search for searching the performance data file matching the extracted pitch data string from the performance data file database based on a first match criterion;
Obtaining a performance data file corresponding to the song from the performance data file database based on a search result of at least one of the first index search and the first melody search;
The performance data file search method characterized by the above-mentioned.   The first melody extraction subtracts audio data without vocal corresponding to the song from audio data with vocal corresponding to the song to extract audio data including the melody portion. 2. The performance data file search method according to 1. When the performance data file cannot be found by the first melody search, the performance data file group to which an index partially matching the index of the audio data is assigned is searched from the performance data file database. Perform an additional index search of
When the group of performance data files can be searched by the second index search, the extracted pitch data string matches with the second matching criterion with a judgment condition eased compared to the first matching criterion. 3. The performance data file search method according to claim 1, further comprising performing a second melody search for searching a performance data file including a pitch data string from the group of performance data files.   4. The performance data file search method according to claim 3, wherein when the performance data file can be searched by the second melody search, the performance data file is acquired from the performance data file database. In the first melody extraction, when audio data without vocal corresponding to the song cannot be acquired, a melody portion is extracted by audio analysis processing from the audio data with vocal corresponding to the song,
2. The performance data file search method according to claim 1, wherein a pitch data string representing the melody portion of the song is extracted from the melody portion of the extracted audio data. A second index search for searching the performance data file database for the group of performance data files to which an index that partially matches the index of the audio data is provided;
If the second index search is successful, the performance data includes the pitch data string that matches the extracted pitch data string with a second matching criterion that is more relaxed than the first matching criterion. 6. The performance data file search method according to claim 5, further comprising performing a second melody search for searching for a file from the group of performance data files.   7. The performance data file search method according to claim 6, wherein when the performance data file can be searched by the second melody search, the performance data file is acquired from the performance data file database. The performance data file is a standard MIDI file in MIDI format.
8. The performance data file search method according to claim 1, wherein the performance data file is searched. Index search means for searching a performance data file including a pitch data string representing the music, to which an index for identifying the music is assigned, from the performance data file database;
When the performance data file cannot be searched by the index search means, the melody portion of the song extracted based on the audio data with vocal corresponding to the song and the audio data without vocal corresponding to the song is obtained. Melody extraction means for extracting a pitch data string representing the melody part of the song from the audio data including,
Melody search means for searching the performance data file matching the extracted pitch data string from the performance data file database based on a first match criterion;
An acquisition means for acquiring a performance data file corresponding to the song from the performance data file database based on a search result of at least one of the index search means and the melody search means;
A performance data file retrieval apparatus comprising: On the computer,
Performing an index search that searches a performance data file database for a performance data file including a pitch data string representing the music, to which an index for identifying a song is assigned;
If the performance data file cannot be searched by the index search, the melody portion of the song extracted based on the audio data with vocal corresponding to the song and the audio data without vocal corresponding to the song is included. Performing a melody extraction for extracting a pitch data string representing a melody portion of the song from the audio data; and
Performing a melody search for searching for the performance data file that matches the extracted pitch data string from the performance data file database based on a first match criterion;
Obtaining a performance data file corresponding to the song from the performance data file database based on a search result of at least one of the index search and the melody search;
A program that executes A pitch data string representing the melody part of the song is extracted from the audio data including the melody part of the song extracted based on the audio data with vocal corresponding to the song and the audio data without vocal corresponding to the song. Melody extraction means to
Melody extraction and transmission means for transmitting the pitch data string extracted by the melody extraction means to a server device;
Receiving means for receiving a performance data file transmitted from the server device;
The terminal device characterized by having.