JP3807333B2 - Melody search device and melody search program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a melody search apparatus and a computer-readable melody search program for searching for performance data related to a melody part from a plurality of groups of performance data each composed of performance data representing a series of musical note strings.
[0002]
[Prior art]
Performance data for automatic performance is a group of performance data representing a series of musical notes for each performance part such as melody, rhythm, chord accompaniment, and bass accompaniment, and multiple groups of performance data are associated with each track. . If you retrieve performance data of the melody part and hear the performance sound of the melody part only, arrange the melody part, or use the melody part as a reference for the arrangement of the chord accompaniment part or bass accompaniment part, It is necessary to extract only the performance data of the melody part from the performance data. Conventionally, the user has performed manual extraction of performance data of such a melody part.
[0003]
[Problems to be solved by the invention]
However, it takes time and effort to extract only performance data of an appropriate melody part from performance data including a large number of performance parts. Further, for beginners, it may be impossible to extract performance data of such a melody part.
[0004]
Summary of the Invention
The present invention has been made to cope with the above-described problems, and its purpose is to automatically perform performance data relating to a melody part from a plurality of groups of performance data each composed of performance data representing a series of note strings. It is an object of the present invention to provide a melody search device that can be searched for.
[0005]
To achieve the above object, the present invention Special In the melody search device for searching the performance data of the group corresponding to the melody part from the performance data of a plurality of groups each constituted by performance data representing a series of note strings, The average pitch of the note strings represented by the series of performance data is calculated for each group, and the calculated average pitch is equal to or greater than a predetermined value and longer than the average pitch of the entire song, or Performance data of a group that is relatively longer than the average pitch calculated by other groups is selected from melody part candidates. It is to be excluded.
[0011]
According to this feature of the present invention, the performance data of the group corresponding to the melody part is taken out from the performance data of the plurality of groups, the performance data of the group not corresponding to the melody part is excluded. Therefore, the performance data related to the melody part is automatically searched without the user's trouble, so that the performance sound of only the melody part can be heard, the melody part can be arranged, the melody part can be a chord accompaniment part or This is useful when you want to reference the arrangement of the bass accompaniment part.
[0014]
Another feature of the present invention is that In a melody search device for extracting performance data belonging to a melody part from a plurality of groups of performance data each composed of performance data representing a series of note strings, the entire song is used as a search section, and the plurality of groups. A first melody search means for searching for one group including performance data belonging to the melody part, and a single melody searched by the first melody search means. Based on group performance data, Within the same group A silent section detecting means for detecting a section in which no note exists as a silent section; By silent section detection means Detected silent section The performance data belonging to the melody part for which the silent section is to be searched from among the performance data belonging to the group excluding one group searched by the first melody search means from among the plurality of groups is searched. And a second melody search means for searching, wherein the performance data belonging to one group searched by the first melody search means is added with performance data belonging to the melody part searched by the second melody search means. The data is now the performance data of the melody part of the entire song There is.
[0016]
According to another aspect of the invention, multiple A few When performance data relating to a melody part is prepared over a loop, for example, even when a guitar solo part is added to a normal melody part, performance data of a melody part connected to the guitar solo part can be obtained.
[0017]
Further, the present invention relates to the above-described present invention. Special The symbol can also be understood as an invention relating to a melody search method and a computer program.
[0018]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a schematic block diagram of an electronic musical instrument incorporating a melody search device according to the present invention.
[0019]
This electronic musical instrument includes a performance operator group 11, a panel operator group 12, and a display 13. The performance operator group 11 is composed of, for example, a keyboard composed of a plurality of keys, and instructs the generation of musical sounds and the pitch of the generated musical sounds by these operations. The panel operator group 12 is disposed on each operation panel, and instructs the operation of the entire electronic musical instrument including designation of tone color, volume, etc. of generated musical tone, indication of display contents on the display 13, and the like. The panel operator group 12 includes operators such as a numeric keypad, a cursor movement key, and a mouse. The operations of the performance operator group 11 and the panel operator group 12 are detected by detection circuits 14 and 15 connected to the bus 20, respectively. The display 13 includes a CRT display, a liquid crystal display, and the like, and displays various information as characters, numbers, or images. The display 13 is controlled by a display circuit 16 connected to the bus 20.
[0020]
The electronic musical instrument also includes a tone generator circuit 17 connected to the bus 20. The tone generator circuit 17 has a plurality of tone signal forming channels for forming a tone signal in a time-sharing manner, and performance information (key-on signal, key-off signal, pitch data (pitch name) supplied via the bus 20 is provided. A musical tone signal is formed and output based on (data), timbre information, etc.). The tone generator circuit 17 forms not only musical tone signals having pitches such as piano sounds and guitar sounds, but also musical tone signals of rhythm instruments such as drums and cymbals. A sound system 18 including a D / A converter, an amplifier, and a speaker is connected to the tone generator circuit 17, and the system 18 generates a tone corresponding to a tone signal from the tone generator circuit 17.
[0021]
In addition, a CPU 21, ROM 22, RAM 23, timer 24 and external storage device 25 are also connected to the bus 20. The CPU 21, ROM 22, RAM 23, and timer 24 constitute a microcomputer main body, and execute various programs to control various operations of the electronic musical instrument.
[0022]
The external storage device 25 can read and write programs and data to and from a recording medium such as a hard disk HD incorporated in advance, a recording medium such as a flexible disk FD and a compact disk CD that are detachably assembled. Drive device. Various programs and various data are stored in the external recording medium. In the present embodiment, in particular, the melody formation program of FIG. 2 (including the melody search routine of FIG. 3) is stored, and a plurality of sets of automatic performance data corresponding to various songs are also stored. Various programs and some data are also stored in the ROM 22.
[0023]
Here, an example of the automatic performance data for one song will be described. As shown in FIG. 4A, the automatic performance data is composed of performance data for a plurality of tracks including the first to nth tracks. The automatic performance data for one track corresponds to, for example, a melody part, rhythm part, chord accompaniment part, bass accompaniment part, and the like. The performance data of each track is sandwiched between start codes and end codes representing the start and end of the performance data, respectively, and tone selection data, volume data, expression data, notes arranged according to the progress of the music (time passage) It consists of on-event data, note-off event data, duration data, and the like. Note that this track corresponds to a group constituted by a series of performance data according to the present invention.
[0024]
The timbre selection data is data for specifying the timbre of the musical tone signal at the start of the music or changing it in the middle of the tune, and as shown in FIG. MIDI channel) number data and tone name data are added. The sound source channel number data represents a sound source channel (MIDI channel) in the sound source circuit 17 to which the timbre name data is applied. The timbre name data represents the timbre of the musical tone signal formed in the sound source channel.
[0025]
The volume data is data for designating or changing the volume of the musical tone signal at the start of the song or in the middle of the song. As shown in FIG. 4C, the volume data as the identification code includes the sound source channel number data and It is configured with volume data added. The sound source channel number data represents a sound source channel in the sound source circuit 17 to which the volume data is applied. The volume data represents the volume level of the tone signal formed in the sound source channel. Note that the volume data represents a volume level that increases as the value increases. In the present embodiment, the volume data can take any value from “0” to “127”.
[0026]
The expression data is also data for designating or changing the volume of the musical tone signal at the start of the song or in the middle of the song. As shown in FIG. 4D, in the case of the volume data, the expression code is an identification code. The same tone generator channel number data and volume data are added. This expression data corresponds to information about an expression pedal in an electronic musical instrument, and is for dynamically changing the volume of a musical tone signal. On the other hand, the volume data corresponds to information of a volume operator in the electronic musical instrument, and controls the change of the volume of the tone signal statically.
[0027]
The note-on event data is data for instructing the start of generation of a musical sound signal. As shown in FIG. 4 (E), the note-on code as the identification code, sound source channel number data, note number data, and velocity data Is added. The sound source channel number data represents a sound source channel in the sound source circuit 17 to which the note number data and velocity data are applied. The note number data represents the pitch of the tone signal generated in the sound source channel. The velocity data corresponds to information on the strength of key touch in the electronic musical instrument, and represents the volume level of the tone signal specified by the note number data. The velocity data also represents a volume level that increases as the value increases, and in this embodiment, can take any value from “0” to “127”.
[0028]
The note-on event data is normally used for a musical tone signal having a pitch, but is also used for instructing generation of rhythm instrument sounds such as drums and cymbals. In this case, instrument name data representing the rhythm instrument name is used instead of the note number data.
[0029]
The note-off event data is data for instructing the end of generation of the musical tone signal. As shown in FIG. 4F, the note-off event data as the identification code is added to the note-off event data as shown in FIG. It is configured by adding channel number data and note number data. However, in this case as well, for rhythm instruments, instrument name data representing rhythm instrument names is used instead of the note number data.
[0030]
The duration data represents the time interval of appearance of the tone selection data, volume data, expression data, note-on event data, note-off event data, and the like described above during playback of performance data. It is configured by adding numerical data. The tick number data is defined as data representing the number of tempo clock signals with respect to a predetermined reference note length (data representing a kind of note length). In the present embodiment, the quarter note length is changed to the tick number “480”. Is defined. The tempo clock signal refers to a reference clock signal that is changed according to the tempo of the song.
[0031]
Such multi-track performance data is reproduced by an automatic performance reproduction program, and the above-described tone name data, volume data, note number data, velocity data and new data generated based on these data are stored in the song. The sound source circuit 17 is supplied in accordance with the progress. The tone generator circuit 17 is controlled according to the supplied data and generates a musical sound signal according to the data. However, since the present invention is not directly related to the reproduction of the performance data, further explanation is omitted. The sound source channel represented by the sound source channel number data in the performance data does not mean one tone signal forming channel for forming a tone signal in the tone generator circuit 17, but a plurality of tone sounds in the tone generator circuit 17. The musical tone signal forming channel group including the signal forming channel is shown.
[0032]
Returning to the description of FIG. A MIDI interface circuit 26 and a communication interface circuit 27 are also connected to the bus 20. The MIDI interface circuit 26 is connected to other MIDI-compatible devices 31 such as an automatic performance device (sequencer), a performance device such as a keyboard, other musical instruments, and a personal computer, and various MIDI data including automatic performance data from the same device 31. This is an interface circuit for receiving information or transmitting various MIDI information to the device 31. The communication interface circuit 27 enables communication with the outside via a communication network 32 such as the Internet.
[0033]
Next, the operation of the embodiment configured as described above will be described. First, by executing a program (not shown), the user operates the panel operator group 12 while viewing the screen of the display 13 to select performance data of a song for which performance data related to the melody part is to be extracted and store it in the RAM 23. Let In this case, as performance data, automatic performance data stored in advance in the external storage device 25, automatic performance data stored in a MIDI-compatible device 31 that can be input via the MIDI interface circuit 26, a communication interface circuit 27, and Automatic performance data available via the communication network 32 can be used. In some cases, performance data by performance using the performance operator group 11 and the panel operator group 12 in the electronic musical instrument is stored in the RAM 23 or the external storage device 25, and the stored performance data is used. You can also
[0034]
Next, the user activates a recording medium such as a hard disk HD, a flexible disk FD, and a compact disk CD in the external storage device 25 or a melody forming program stored in the ROM 22. If this melody formation program is not stored in the external storage device 25 or the ROM 22, the melody formation is performed from the MIDI compatible device 31 via the MIDI interface circuit 26 or from the outside via the communication interface circuit 27 and the communication network 32. It is also possible to receive a program.
[0035]
The execution of this melody formation program is started in step 100 of FIG. 2, and the CPU 21 executes initialization processing of the melody search section in step 102. In this initialization process, the entire tune, that is, all performance data sandwiched between start codes and end codes in each track in FIG. After this processing, in step 104, a melody search routine is executed.
[0036]
This melody search routine is shown in detail in FIG. After starting this execution, the CPU 21 executes various exclusion processes consisting of steps 202 to 212. The characteristic of the processing by this melody search routine is that the tracks storing the performance data having the characteristics of parts other than the melody are extracted in the various exclusion processes in steps 202 to 212, and the extracted tracks are sequentially excluded. The last remaining track is determined as the track in which the performance data of the melody part is stored. Hereinafter, the various exclusion processing algorithms in steps 202 to 212 will be sequentially described.
[0037]
a. Rhythm part exclusion process in step 202
In performance data relating to a piece of music including a plurality of parts, it is a general rule that the tone color of the generated tone signal is specified for each track. A track including performance data for specifying the tone color of a rhythm instrument such as a drum or a cymbal relates to a rhythm part. Focusing on these points, in this rhythm part exclusion process, a track related to the rhythm part is excluded from a plurality of tracks.
[0038]
Specifically, if the performance data is based on the GM standard, the rhythm instrument is assigned to the tone generator channel “10” of the tone generator circuit 17. The GM standard refers to a standard for MIDI sound source standardization. Therefore, in this rhythm part exclusion process, the CPU 21 confirms that the performance data is based on the GM standard (normally recorded in the performance data), and then sequentially reads the performance data of each track. It is sequentially checked whether the read performance data designates the sound source channel “10”. If the performance data designating the sound source channel “10” is included, it is determined that the track including the performance data relates to the rhythm part (see FIG. 4), and the track is selected from the melody part candidates. exclude. Alternatively, it may be determined that a track including performance data representing a rhythm instrument name relates to a rhythm part (see FIG. 4), and the track is excluded from melody part candidates. In this case, the performance data need not be based on the GM standard.
[0039]
b. Base part exclusion process in step 204
Bass part instruments such as electric bass and contrabass used for playing the bass part are not used for playing the melody. Focusing on this point, in this base part exclusion process, a track related to the base part is excluded from the remaining plurality of tracks after the exclusion process in step 202. Specifically, the CPU 21 sequentially reads the performance data of the remaining tracks, and checks whether the read performance data represents a bass part instrument name. If the performance data for designating the instrument name for the bass part is included, it is determined that the track including the performance data relates to the bass part (see FIG. 4), and the track is selected from the melody part candidates. exclude.
[0040]
c. Exclusion processing based on the note presence range ratio in step 206
If the track is related to the melody part, it is considered that notes exist in a section of a certain amount or more (for example, 60% or more of the whole song) in the whole song. Focusing on this point, in the exclusion process based on the note presence range ratio, tracks whose ratio of the section including the notes to the entire music is equal to or less than a predetermined value from the remaining plurality of tracks after the process of step 204 are as follows. The track is excluded from the melody part candidates because it is not related to the melody part.
[0041]
Specifically, the CPU 21 sequentially reads out the performance data of each track of the plurality of tracks remaining after the exclusion process in the step 204, and sets a section where a note exists for each track based on the read performance data. Look into. That is, a section from when one note-on event data appears until a note-off event corresponding to the note-on event data appears is a section where a note exists. As long as the section in which the next note exists continues to the section in which the one note exists, this continuous section is set as the section in which the note exists. On the other hand, when the next note interval starts after the end of the previous note interval, the interval from the end of the previous note interval to the start of the next note interval is defined as a non-existent interval.
[0042]
If the period from the end of the previous note interval to the start of the next note interval is short, it may be regarded as a continuous interval. For example, if the period from the end of the previous note interval to the start of the next note interval is equal to or shorter than the time corresponding to a 32nd note, it is considered as a continuous note interval, and the interval between synchronizations corresponds to a 32nd note If the time is exceeded, the section does not have a note.
[0043]
FIG. 5 shows the analysis result of the performance data of one track. In this case, sections A, B, and C are sections in which musical notes exist in the entire song. Then, the lengths of sections A, B, and C are added, and the added length is divided by the length of the entire song to calculate the note presence range ratio. For example, if the total length of the song is 16 bars and the sections A, B, and C are 4, 8, and 2 bars, respectively, the note presence range rate is 0.88 (= (4 + 8 + 2) / 16). Become. Actually, since the calculation of this section is calculated by the number of ticks, it is calculated by dividing the tick number “26,800” for 12 bars by the “30,720” for 16 bars. .
[0044]
After calculating such a note presence range rate for each track, the note presence range rate is less than or equal to a predetermined value (for example, 0.6) from among the plurality of tracks remaining after the exclusion process in steps 202 and 204. A track is excluded from melody part candidates.
[0045]
Further, as described above, instead of or in addition to excluding a track having a note existence range rate of a predetermined value or less from the melody part candidate, a track having a relatively low note existence range rate compared to other tracks is added to the melody part. You may make it exclude from a candidate. That is, one or a plurality of tracks may be excluded from the melody part candidates in order from the lowest note presence range rate.
[0046]
d. Exclusion processing based on average volume in step 208
If the track is related to the melody part, the average volume of the tone signal generated by the performance data of the track should be large. For example, it is considered that the average volume of the musical tone signal of the melody part is at least 80% of the maximum value (hereinafter referred to as the maximum average volume) of the average volumes for all tracks. Focusing on this point, in the exclusion process based on the average volume, the average volume designated by the performance data from the plurality of tracks remaining after the exclusion process in step 206 is a predetermined value or more than the maximum average volume. A small track is excluded from the melody part candidates as not relating to the melody part.
[0047]
Specifically, the CPU 21 sequentially reads out the performance data of each track of the plurality of tracks remaining after the exclusion process in the steps 202 to 206, and based on the read performance data, the average sound volume of the notes for each track. calculate. The volume of the notes belonging to one track is the product of the volume data in the volume data, the volume data in the expression data, and the velocity data in each note-on event data. It is determined by dividing the maximum value (in this embodiment, “127”) that can be taken by a squared value.
[0048]
Therefore, the CPU 21 stores the volume data in the immediately preceding volume data and the volume data in the expression data independently for each track. Each time note-on event data appears, a value obtained by multiplying the product of both stored volume data by the velocity data value in the note event data (= volume data value in volume data * volume in expression data) (Data value * velocity data value) is calculated, and the multiplication result is divided by the square value of the maximum value (for example, "127") that can be taken by the volume data to calculate volume data for one note.
[0049]
Next, the CPU 21 adds the volume data for one note related to all the notes included in one track, divides the result of addition by the number of notes included in the same track, and equals the one note related to one track. Calculate the average volume of. Such calculation of the average volume for one note is performed for each track. After calculating the average sound volume for one sound for each track, the track having the maximum calculated average sound volume is extracted from all the tracks remaining after the exclusion process in steps 202 to 206. The average volume of the extracted track is set as the maximum average volume. Then, tracks whose average volume is smaller than the maximum average volume by a predetermined value or more are excluded from melody part candidates. In this case, the volume smaller than the predetermined value may be a volume smaller than the maximum average volume by a predetermined ratio or a volume smaller than the absolute maximum by an absolute predetermined value.
[0050]
Further, instead of or in addition to comparing the average volume of each track with the maximum average volume as described above, each track may be excluded from the melody part candidates as follows. That is, tracks whose average volume is less than or equal to an absolute predetermined value are excluded from melody part candidates. Further, the average value of the average sound volume of all tracks is further calculated, and when the average sound volume of each track is lower than the average value by a predetermined value or more, the respective track is excluded from the melody part candidates.
[0051]
Furthermore, instead of or in addition to the exclusion condition relating to the average volume, a track whose average volume for one note is relatively smaller than other tracks may be excluded from the melody part candidates. That is, one or a plurality of tracks may be excluded from the melody part candidates in order from the lowest average volume for one note.
[0052]
e. Exclusion processing based on average sound length in step 210
If the track is related to the melody part, the average tone length of the musical sound signal generated by each performance data of the track should not be so long. For example, the average tone length of the musical tone signal of the melody part is considered to be less than the length corresponding to the half note. A part having a long average sound length is considered to be an accompaniment part such as a string sound or an accompaniment part of a pad. The pad is one of timbre names such as a synthesizer and is a relatively rounded continuous sound (sometimes called a white ball). Focusing on this point, in the exclusion process based on the average tone length, the average value of the tone lengths of the musical sound signal designated by the performance data is selected from the plurality of tracks remaining after the exclusion process in steps 202 to 208. A track having a predetermined length or longer is not related to the melody part, and the track is excluded from the melody part candidates.
[0053]
Specifically, the CPU 21 sequentially reads out the performance data of each track of the plurality of tracks remaining after the exclusion process in steps 202 to 208, and calculates the average note length for each track based on the read performance data. To calculate. The note length is approximately equal to the time length from when the note-on event data appears until the note-off event data corresponding to the note-on event data appears.
[0054]
Therefore, the CPU 21 calculates the time length from the occurrence timing of the note-on event data to the occurrence timing of the note-off event data for each note represented by the note-on event data. This time length is calculated by accumulating duration data, that is, the number of ticks located between the note-on event data and note-off event data of the corresponding note. Thereby, the note length of one note is calculated.
[0055]
Next, the CPU 21 adds the note lengths for all the notes included in one track, divides the addition result by the number of notes included in the same track, and calculates the average note length for one track. . Such calculation of the average sound length is performed for each track. After calculating the average sound length for each track in this way, tracks whose average sound length is a predetermined value or more are excluded from melody part candidates. In this case, as described above, since the sound length is represented by the number of ticks, the number of ticks is compared with a predetermined value. In the present embodiment, the note length of a quarter note corresponds to the number of ticks “480”, and therefore, for example, a track whose tic number representing the calculated average length is “960” or more is excluded from the melody part candidates. .
[0056]
Further, instead of or in addition to comparing the average sound length of each track with an absolute predetermined value as described above, each track may be excluded from melody part candidates as follows. That is, the average value of the average sound lengths of all the tracks is calculated, and when the average sound length of each track is longer than the average value by a predetermined value or more, the respective tracks are excluded from the melody part candidates. In this case as well, the average sound length longer than the average value by a predetermined value or longer is a sound length longer than the average value by a predetermined ratio or longer than the average value. There may be.
[0057]
Furthermore, instead of or in addition to the exclusion condition regarding the average sound length, a track whose average sound length is relatively longer than other tracks may be excluded from the melody part candidates. That is, one or a plurality of tracks are excluded from the melody part candidates in order of increasing average sound length.
[0058]
f. Exclusion processing based on duplication rate in step 212
If the track is related to the melody part, there are many single melodies, and multiple sounds are rarely pronounced simultaneously. Even if it is a single melody, in the case of tenuto performance, the preceding and following notes are slightly overlapped, so it cannot be said that there is no “overlap”, but it is considered that there is not much overlap. On the other hand, in an accompaniment part etc., there is a very high possibility that a plurality of sounds will be pronounced simultaneously such as chords. Focusing on this point, in the exclusion process based on the duplication rate, it is assumed that a track with a high duplication rate is not related to the melody part among the plurality of tracks remaining after the exclusion process of steps 202 to 210. Exclude from candidates. This overlap rate is defined as a value obtained by dividing the total length of notes appearing in one song by the total number of intervals in which notes exist.
[0059]
Specifically, the CPU 21 sequentially reads out the performance data of each track of the plurality of tracks remaining after the exclusion process in steps 202 to 210, and calculates the overlap rate for each track based on the read performance data. . First, the total length of the notes appearing in one song is calculated for one track. The total note length can be calculated anew, but the sum of note lengths for all notes included in one track calculated in the process of calculating the average note length in step 210 is stored. It is convenient to use this stored value.
[0060]
Next, the total of the section where a note exists is calculated. The total of the sections can be newly calculated, but the addition result of the section where the notes calculated in the calculation process of the note existence range ratio in the step 206 is stored, and this stored value is stored. It is convenient to use. Then, the duplication rate is calculated by dividing the total value of the note lengths of the notes appearing in the song by the total value of the section where the notes are present. After calculating the duplication rate for each track, tracks having the duplication rate of a predetermined value or more are excluded from the melody part candidates. For example, a value of about “2.0” can be adopted as the predetermined value.
[0061]
Further, instead of or in addition to comparing the overlap rate of each track with an absolute predetermined value as described above, each track may be excluded from the melody part candidates as follows. That is, the average value of the overlap ratios of all the tracks remaining after the exclusion process in the steps 202 to 210 is further calculated, and when the overlap ratio of each track is higher than the average value by a predetermined value or more, each track is melodies. Exclude from part candidates. In this case, the overlap rate higher than the average value by a predetermined value or more may be a overlap rate higher than the average value by a predetermined ratio or higher than the average value. Good.
[0062]
Further, instead of or in addition to the exclusion condition from the melody part candidate, a track having a relatively high overlap rate compared to other tracks may be excluded from the melody part candidate. That is, one or more tracks are excluded from the melody part candidates in descending order of overlap rate.
[0063]
Next, in step 212, the track corresponding to the melody part is determined. In this determination, if only one track remains in the exclusion process in steps 202 to 212, the same track is determined as a track corresponding to the melody part. On the other hand, when a plurality of tracks remain as melody part candidates, the track with the lowest overlap rate is determined as the track related to the melody part. In step 214, the execution of this melody search routine is terminated.
[0064]
In this way, the track corresponding to the melody part is searched, but the exclusion process in steps 202 to 212 described above may be performed in any order. Further, any one or more of the exclusion processes in steps 202 to 212 may be omitted. However, in step 212, in addition to the exclusion process, a process of finally determining one track is also performed. Therefore, when the order of the exclusion process conditions in steps 202 to 212 is changed, and a plurality of tracks remain after the exclusion process in the last step, the track with the lowest overlap rate is selected from the remaining plurality of tracks. Is determined as the track corresponding to the melody part. Further, one track most suitable for the melody part may be determined according to the exclusion process condition of the last step replaced. Furthermore, it is also possible to evaluate a combination of all or a part of the note presence range rate, average volume, average sound length, and overlap rate to determine one track most suitable for the melody part.
[0065]
Returning to the explanation of the melody formation program in FIG. After the execution of the melody search routine in step 104, it is checked in step 106 whether a silent section exists in the melody part. This is because the silent section check may store performance data relating to the melody part across a plurality of tracks.
[0066]
In checking the silent section, the CPU 21 sequentially reads the performance data of the track determined to be a track related to the melody part, and checks the silent section where the pronunciation is not instructed. Specifically, the interval from the time when the note-on event data appears until the time when the note-off event data appears is the interval in which the tone of the musical tone is instructed, so the other interval, that is, the note-on event data appears at all. A section from when the sound generation instruction by the previous or all note-on event data is canceled by the note-off event data until the next note-on event newly appears is detected. In this case, when the detected section is short (for example, when it is equal to or shorter than the half note length), there is a timing related to a rest, so that it is not detected as a silent section. If this silent section does not exist, “No” is determined in step 106, and the process proceeds to step 110.
[0067]
On the other hand, if the silent section exists, “Yes” is determined in step 106, and the process proceeds to step 108. In step 108, the detected silent section is set as a search section, and the process returns to step 104. In step 104, the above-described processing of the melody search routine of FIG. 3 is executed for all the tracks other than the track determined as the melody part for the set search section (silent section). Therefore, tracks other than the determined track are determined as tracks related to the melody part for this silent section.
[0068]
In the silent section check, if a plurality of silence sections are set, a plurality of silence sections are set as search sections in step 108, and the melody search routine in step 104 performs processing for the plurality of silence sections. Tracks related to the melody part are determined respectively. However, in each process in the melody search process routine, a plurality of silent sections are processed independently.
[0069]
After execution of the melody search routine in step 104, the CPU 21 checks the silent section again in step 106. This is because, even in the second melody search, there may be a silent section in the silent section by the first melody search. Then, the processes of steps 104 to 108 are repeatedly executed until there is no silent section over the entire song. As a result, if there is no silent section over the entire song, as described above, “Yes” is determined in step 106 and the process proceeds to step 110.
[0070]
The upper part of FIG. 6 shows a state of re-melody search in this silent section. In this case, the track i is determined as the track of the melody part by the first melody search. In this case, a silent section exists between each of melody A-1, melody A-2, and melody A-3 in track i. Also, the track j is determined as the track of the melody part for the previous silent section by the second melody search. In this case, there is a silent section between melody B-1 and melody B-2 in track j, and between melody A-2 in track i and melody B-3 in track j. For the silent section between the melody B-1 and the melody B-2, the track k is determined as the track of the melody part. For the silent section between the melody A-2 and the melody B-3, the track m is determined as the melody part track. The melody in these cases is represented as melody C-1 and melody D-1.
[0071]
In step 110, it is determined whether there are a plurality of melody sections, in other words, whether the performance data of a plurality of tracks has been determined as a melody. If a plurality of melody sections do not exist, “No” is determined in step 110, and the process proceeds to step 114. In step 114, the performance data of the track determined in the melody search routine is determined as data relating to the melody, and in step 116, the execution of the melody forming program is terminated.
[0072]
On the other hand, if there are a plurality of melody sections, “Yes” is determined in step 110 and the sections determined as melody in step 112 are connected. In the connection of each section, the performance data determined as the melody portion is extracted from the performance data in each track determined as the melody part, and the performance data is sequentially written in the RAM 23. Link. An example of this connection is shown in the lower part of FIG.
[0073]
Then, after the process of step 114, the execution of the melody forming program is terminated at step 116. The performance data representing the melody determined in this way may be stored together with performance data including a plurality of tracks read from the external storage device 25 or the like for melody extraction.
[0074]
As can be understood from the above operation description, the performance data of the group corresponding to the melody part is excluded from the performance data of the plurality of tracks by the execution of the melody search routine of FIG. Data is retrieved. Therefore, the performance data related to the melody part is automatically searched without the user's trouble, so that the performance sound of only the melody part can be heard, the melody part can be arranged, the melody part can be a chord accompaniment part or This is useful when you want to reference the arrangement of the bass accompaniment part.
[0075]
Also, in this melody search routine, multiple groups of performance data are selected from melody part candidates under a plurality of different conditions consisting of a rhythm instrument, bass part, note presence range rate, average volume, average note length, and note overlap rate. The performance data of the group remaining after the exclusion is determined as the performance data of the melody part. Therefore, the performance data of the group corresponding to the melody part is searched with high accuracy.
[0076]
Further, in the above embodiment, the silent section in the performance data searched for the melody search is checked by the processing in steps 104 to 114 in FIG. 2, and if there is a silent section, the melody for the silent section is targeted. The search is repeatedly executed, and the performance data corresponding to the new melody part is searched. The plurality of searched performance data are connected to form continuous melody part performance data. As a result, when performance data related to a melody part is prepared across multiple tracks, for example, even when a guitar solo part is added to a normal melody part, performance data of the melody part linked to this guitar solo part can be obtained. Can do.
[0077]
In the above-described embodiment, an example in which performance data of a plurality of parts is stored separately on a plurality of tracks has been described. However, the present invention can also be applied to automatic performance data in which performance data of a plurality of parts are mixed and stored in one track. In this case, each part may be distinguished by checking the sound source channel number data in each event data.
[0078]
In the above-described embodiment, an example in which the melody search device is applied to an electronic musical instrument has been described. However, the present invention can be applied to any electronic device as long as it is a variety of electronic devices that can be programmed such as a personal computer.
[0079]
Furthermore, in carrying out the present invention, the present invention is not limited to the above-described embodiment and its modifications, and various modifications can be made without departing from the object of the present invention.
[Brief description of the drawings]
FIG. 1 is an overall schematic block diagram of an electronic musical instrument to which a melody search device according to an embodiment of the present invention is applied.
FIG. 2 is a flowchart showing a melody forming program executed by the CPU of FIG.
FIG. 3 is a flowchart showing details of a melody search routine in FIG. 2;
FIG. 4 is a format diagram showing an example of performance data.
FIG. 5 is an explanatory diagram for explaining calculation of a note presence range rate;
FIG. 6 is an explanatory diagram for explaining a connection example of a plurality of melodies.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 11 ... Performance operator group, 12 ... Panel operator group, 13 ... Display, 17 ... Sound source circuit, 21 ... CPU, 22 ... ROM, 23 ... RAM, 25 ... External storage device.

Claims (4)

In a melody search device for searching performance data of a group corresponding to a melody part from performance data of a plurality of groups each constituted by performance data representing a series of note strings,
The average pitch of the note strings represented by the series of performance data is calculated for each group, and the calculated average pitch is equal to or greater than a predetermined value and longer than the average pitch of the entire song, or A melody search apparatus characterized in that performance data of a group that is relatively long with respect to an average tone length calculated by another group is excluded from melody part candidates. In a melody search apparatus for extracting performance data belonging to a melody part from a plurality of groups of performance data each constituted by performance data representing a series of note strings,
First melody search means for searching one group including performance data belonging to the melody part from the plurality of groups, with the entire song as a search section;
Silent section detecting means for detecting a section in which no note exists in the same group as a silent section based on the performance data of one group searched by the first melody searching means ;
The silent section detected by the silent section detected by the silent section detecting means is selected as a search target, and the silent section is selected from the performance data belonging to the group excluding one group searched by the first melody searching means among the plurality of groups. Second melody search means for searching performance data belonging to the melody part for which
The performance data obtained by adding the performance data belonging to the melody part searched by the second melody search means to the performance data belonging to one group searched by the first melody search means is used as performance data of the entire melody part. A melody search device characterized by that. In a computer-readable melody search program for causing a computer to search performance data of a group corresponding to a melody part from performance data of a plurality of groups each constituted by performance data representing a series of note strings,
In the computer,
The average pitch of the note sequence represented by a series of performance data is calculated for each group, and the calculated average pitch is equal to or greater than a predetermined value and longer than the average pitch of the entire song, Alternatively, a melody search program characterized in that performance data of a group relatively longer than the average pitch calculated by another group is excluded from melody part candidates. In a computer-readable melody search program for causing a computer to extract performance data belonging to a melody part from a plurality of groups of performance data each composed of performance data representing a series of note strings,
In the computer,
With the entire song as a search section, one group including performance data belonging to the melody part is searched from the plurality of groups,
Based on the retrieved performance data of one group, a section in which no note exists in the same group is detected as a silent section; and
Performance data belonging to a melody part whose search target is the silent section from among the performance data belonging to the group excluding the searched one group among the plurality of groups with the detected silent section as a search target Let's search for
Melody search characterized in that performance data obtained by adding performance data belonging to the retrieved melody part to performance data belonging to the retrieved group is used as performance data of the entire melody part. Program .

Images