JP3948464B2 - Performance data generating apparatus and recording medium - Google Patents

Description

  The present invention relates to an apparatus and a recording medium for generating performance data.

  2. Description of the Related Art Conventionally, when creating performance data with a facial expression in an automatic performance data creation device such as an automatic music composition device, first, the striking point position (pronunciation timing) and the sound length are determined, and then the staccato / tenuto etc. Facial expressions were given by the length of the sound and pitch changes such as slur and shackle.

  In addition, as a facial expression to be added to the performance data, “slur” that smoothly connects the pitch between two notes, and “sharsh” that begins to pronounce at a slightly lower pitch than the original pitch and gradually changes to the original pitch ”And the like are accompanied by a pitch change.

  However, the conventional technique has a drawback that the determination of the hit point and the sound length is different from the application of the facial expression, and the processing efficiency is poor. In addition, there is a disadvantage that it is not possible to reapply only the facial expression while keeping the positions of the hit points and the basic pitch constituting the basic performance data.

  Furthermore, in the conventional technology, if the speed of the pitch change is slow, depending on the length of the note (in the case of a short note), the time when the pitch is stable after the end of the pitch change becomes extremely short. Or the sound of the performance may become unnatural. In addition, if the speed of the pitch change is simply increased, the pitch changes quickly even for a long note, and a slow pitch change may not be obtained.

  Conventional melody generation is performed by determining a musical range, key, time signature, etc. as music conditions for generation, and specifying various musical instruments (timbres). However, in an actual musical instrument, the range of sound that can be generated, the jumping pitch width of the sound, the jumping frequency, and the expression method differ depending on the musical instrument.

  For example, in the case of a piano, the sound range is wide (A1 to C7) and the pitch is 88 pitches, and as a performance expression, legato (a little bit overlaps the previous note when playing the next note) or staccato (cuts the sound short) ) Etc. On the other hand, in the case of a human voice, the range is generally narrow, roughly one octave and a half, and as a performance expression, slurs (sounds are continued as they are when moving to the next sound, and only the pitch is changed) or staccato Can be expressed. On the other hand, there is a characteristic that the slur cannot be expressed with the piano, and it is difficult to repeat the jump of a very intense sound with the human voice.

  In the conventional melody generation technology, it has been impossible to generate a melody that takes advantage of the characteristics (or restrictions) of such an instrument. For example, when a melody is generated in a wide range such as playing on a piano, the melody could not be sung. In addition, when a melody is generated in a narrow range for singing, when the melody is played on a piano, the range is narrow and lacking in interest.

  In view of such circumstances, an object of the present invention is to provide a performance data generating apparatus capable of generating performance data having a melody suitable for a musical instrument to be sounded (played), and a recording medium therefor.

  According to the main feature of the present invention, music condition designating means for designating musical conditions independent of instruments, including at least key, note resolution, time signature, and song style, for the performance data to be generated, and an instrument to be played with the performance data Musical instrument designating means, parameter obtaining means for obtaining performance data generation parameters depending on the musical instrument designated by the musical instrument designating means, including at least a range, a jump distance, a jump frequency, and an expression method; music Parameter changing means for changing at least the jump distance, jump frequency, and expression method included in the performance data generation parameters acquired by the parameter acquiring means based on the musical composition included in the music conditions specified by the condition specifying means; Music conditions specified by the music condition specifying means and performance data changed by the parameter changing means A performance data generation device comprising performance data generation means for generating performance data based on the composition parameters, and the performance data to be generated includes at least key, note resolution, time signature and musical style. A music condition designating step for designating music conditions independent of an instrument, an instrument designating step for designating an instrument to be played with the performance data, and a performance data generating parameter depending on the instrument designated in the instrument designating step. A parameter acquisition step that acquires at least a range, a jump distance, a jump frequency, and an expression method, and a performance acquired in the parameter acquisition step based on the music style included in the music conditions specified in the music condition specification step Parameters that change at least the jump distance, jump frequency, and expression method included in the data generation parameters A program for causing a computer to execute a procedure comprising a further step and a performance data generation step for generating performance data based on the music condition specified in the music condition specification step and the performance data generation parameter changed in the parameter change step A computer-readable recording medium (claim 2) is provided for generating performance data in which the program is recorded.

In the present invention, the parameters such as the pitch, jump distance (jump pitch width), jump frequency, and expression method are highly dependent on the musical instrument. , Prepared for each type of instrument. When generating performance data, specify musical conditions that do not depend on the instrument, such as key, note resolution, and time signature, and specify the type of musical instrument to be played, and parameters for generating performance data will be extracted for the specified instrument. Then, at least the jump distance, the jump frequency, and the expression method included in the performance data generation parameters of the musical instrument thus taken out are changed to data suitable for the designated music condition. Then, performance data is generated based on the designated music condition and the changed performance data generation parameter. Therefore, according to the present invention, performance data having a melody suitable for the designated musical instrument can be generated.

[Various features]
In order to deal with the drawbacks and inconveniences described in the previous section of the “Prior Art” section, the following (1) to (3) are configured, so that the basic performance data remains unchanged and different expressions are given. In addition, it is possible to efficiently determine the length of the performance data and to add facial expressions. Furthermore, natural pitch changes such as “slur” and “shrimp” can be applied to the basic performance data. Can be added.

(1) Means for acquiring basic performance data, means for setting facial expression data for adding a facial expression corresponding to the acquired basic performance data so as to be resettable, acquired basic performance data and set A facial expression comprising means for storing the corresponding facial expression data and means for creating performance data with a facial expression based on the stored basic performance data and facial expression data (facial expression information) is provided. A device for creating performance data or a step of acquiring basic performance data, a step of setting facial expression data for imparting a facial expression corresponding to the acquired basic performance data to be resettable, and an acquired basic A step of storing the performance data and the set facial expression data correspondingly, and a facial expression is assigned based on the stored basic performance data and facial expression data (facial expression information). Computer readable recording medium for creating performance data expression that records a program to execute a procedure comprising a step of creating a performance data to the computer is granted. Here, the expression data is for giving an expression accompanied by a change in sound length or an expression accompanied by a pitch change.

(2) Regarding basic performance data having at least hit point information, means for setting a hit point to which a facial expression accompanied by a change in sound length is set out of a plurality of hit points represented by the hit point information, and a hit point for which facial expression is set Determines a predetermined ratio of the interval between the hit point and the next hit point as the sound length, and for a hit point for which no facial expression is set, means for determining the interval between the hit point and the next hit point as the sound length, and the determined sound length A performance data creation device with a facial expression provided with a means for creating performance data with a facial expression accompanied by a change in pitch, or at least basic performance data having spot information. Among the plurality of hit points represented by the step of setting a hit point to which a facial expression accompanied by a change in sound length is set, and for the hit point for which a facial expression is set, A predetermined ratio of the interval between the hit points is determined as the sound length, and for a hit point for which no facial expression is set, the step of determining the interval between the hit point and the next hit point as the sound length, and the sound length based on the determined sound length A computer-readable recording medium for creating performance data with a facial expression recording a program for causing a computer to execute a procedure comprising creating performance data with a facial expression accompanied by a change.

(3) Means for acquiring basic performance data, means for setting pitch change facial expression data for giving a facial expression accompanied by pitch change corresponding to the acquired basic performance data, and set pitch change facial expression data Means for determining whether or not a predetermined pitch stable section can be secured based on a predetermined transition time, and determining that a predetermined pitch stable section cannot be secured. If it is, the performance data with the expression that accompanies the pitch change to the basic performance data is created based on the means for shortening the predetermined transition time, the predetermined transition time or the shortened transition time. A performance data creation device to which a facial expression is given, or a step of acquiring basic performance data, and corresponding to the acquired basic performance data A step of setting pitch change facial expression data for giving a facial expression accompanied by a change in pitch, and a predetermined expression based on a predetermined transition time when a facial expression corresponding to the set pitch change facial expression data is given to the basic performance data A step of determining whether or not a predetermined pitch stable section can be secured, a step of shortening a predetermined transition time when it is determined that a predetermined pitch stable section cannot be secured, and a predetermined transition time or Based on the shortened transition time, a facial expression recording a program for causing the computer to execute a procedure including the step of creating performance data in which a facial expression accompanied by a pitch change is added to the basic performance data is provided. A computer-readable recording medium for creating performance data.

  That is, according to the performance data creation device or the storage medium of (1) above, the facial expression data for giving a facial expression accompanied by a change in pitch or a facial expression accompanied by a pitch change is reproduced corresponding to the basic performance data representing the basic note. Since it is configured to create performance data with facial expressions on the basis of basic performance data and facial expression data (facial expression information), it is possible to set the hit point position (spot timing), basic pitch, etc. It is possible to reapply only the facial expression accompanied by the change in pitch or the facial expression accompanied by the pitch change, with the basic performance data as it is. Therefore, "facial expression data" can be effectively added to a given note among a number of basic notes, giving various different facial expressions due to, for example, sound length changes due to staccato, pitch changes due to slurs, shackles, etc. Can do. Further, according to a specific example, after the hitting point timing is first determined, and then the pitch (note number) is generated, the sound length determination with facial expression is performed, and the facial expression with pitch change is set. Such a performance data generation procedure matches the natural sense of the user.

  According to the performance data creation device or the storage medium of (2) above, the basic performance data includes at least dot information, and among the plurality of dots represented by the dot information of the basic performance data, facial expression with a change in sound length is set. For a given hit point, a predetermined ratio of the interval between the hit point and the next hit point is determined as the sound length, and for a hit point for which no facial expression is set, the interval between the hit point and the next hit point is determined as the sound length. Since the performance data to which a facial expression accompanied by a change in sound length is added is created, the determined note and the assigned facial expression can be changed according to the predetermined ratio, and the sound length can be determined. Since the facial expression is added at the same time, the performance data can be created efficiently. Specifically, when facial expression with a change in sound length is set, for example, if a predetermined ratio for determining the sound length of a hit point for which facial expression is set is 70%, If the interval is an eighth note length, it becomes a staccato eighth note, and if this interval is a dotted eighth note length, it becomes a tenuto eighth note and a 16th rest. Note that the rest does not have to be stored in the performance data, and can be regarded as a rest if there is a rest period in which no sound is produced even if only the notes are stored. If the above-mentioned predetermined ratio is different, the determined note and the expression to be given also change. For example, if the predetermined ratio is 50%, if the interval is an eighth note, tenuto 16th note and 16th rest become. Also, the hit points that do not give a facial expression are always tenuto-like notes. As described above, since the sound length determination and the facial expression are performed simultaneously, performance data can be created efficiently.

  According to the performance data creating apparatus or storage medium of (3) above, the basic performance data is further added with pitch information, and a pitch for giving a facial expression accompanied by a pitch change corresponding to the basic performance data. When changing facial expression data is set, it is determined whether a predetermined pitch stable section can be secured based on a predetermined transition time, and the basic performance is determined based on an appropriate transition time obtained according to the determination result. Performance data in which a facial expression with a pitch change is added to the data is created. In other words, when facial expression with a pitch change is set, the speed of the pitch change is changed according to whether or not the time of the stable state of the pitch is secured for a predetermined time or longer. However, a natural performance sound with a stable part of the pitch can be obtained, and even a long note can provide a relaxed pitch change.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings. The following embodiments are merely examples, and various modifications can be made without departing from the spirit of the present invention.

[Hardware configuration]
FIG. 1 shows a block diagram of a hardware configuration of a performance data creation apparatus according to an embodiment of the present invention. In this example, the system includes a central processing unit (CPU) 1, a read only memory (ROM) 2, a random access memory (RAM) 3, first and second detection circuits 4 and 5, a display circuit 6, and a sound source circuit 7. , An effect circuit 8, an external storage device 9 and the like. These devices 1 to 9 are connected to each other via a bus 10 and constitute a performance data creation system for performing performance data creation processing.

  The CPU 1 that controls the entire system includes a timer 11 that is used to generate a tempo clock, an interrupt clock, and the like, and performs various controls according to a predetermined program. The ROM 2 stores predetermined control programs for controlling the performance data creation system. These control programs include various processes including the automatic melody generation process according to the present invention as well as basic performance information processing. Various tables such as programs, formats of original melody data and reproduction melody data, and various data can be included. The RAM 3 stores data and parameters necessary for these processes, and is used as a work area for temporarily storing various data being processed such as various registers and flags, original melody data, and reproduction melody data.

  The first detection circuit 4 is connected to a performance operation device 12 having a performance operator such as a keyboard, and the operation switch device 13 connected to the second detection circuit 5 is for specifying an automatic melody generation processing mode or the like. There are controls for setting various modes, parameters, and operations. These controls include switches on the panel and character / number keys for setting various modes and inputting data, as well as a mouse, etc. Includes pointing devices. The display circuit 6 includes a display 14 and various indicators, and these displays 14 and indicators can be juxtaposed on various operators on the operation panel of the switch device 13. In addition, various settings screens and various operation buttons are displayed on the display 14, and various modes and data values can be set and displayed by using the above-described pointing device, character / number keys, and the like. it can.

  The sound system 15 connected to the effect circuit 8 composed of a DSP or the like constitutes a musical sound output unit together with the sound source circuit 7 and the effect circuit 8, and reproduces the melody data for reproduction created during the automatic melody generation process according to the present invention. Musical sounds can be emitted based on various performance data.

  The external storage device 9 includes a hard disk drive (HDD), a compact disk read only memory (CD-ROM) drive, a floppy disk drive (FDD), a magneto-optical (MO) disk drive, a digital multipurpose disk (DVD) drive, etc. It is possible to store various control programs and various data, and is also used to construct various databases for melody generation. Therefore, programs and various data necessary for performance data processing can be read from the external storage device 9 into the RAM 3 as well as using the ROM 2, and the processing results can be externally stored via the RAM 3 as necessary. It can also be recorded in the device 9.

  In this example, a MIDI interface (I / F) 16 is connected to the bus 10, and the system can communicate with other MIDI devices 17. Further, a communication interface 18 is also connected to the bus 10, and a control program and various data can be stored in the external storage device 9 from the server computer 20 via the communication network 19.

[Melody data format]
2 and 3 show data formats of original melody data and reproduction melody data used in one embodiment of the present invention. As shown in FIG. 2 (1), the original melody data for one song is obtained from the first bar data MS1 in time series, in the order of the second bar, the third bar, the fourth bar,. Measure data MS2, MS3, MS4,... Follow and end with end data EDo. FIG. 2 (2) illustrates the data format for the first bar of FIG. 2 (1), and the other bars have the same format.

  As shown in FIG. 2B, the original melody data is composed of basic note data NT and facial expression data EX for each predetermined note (note). Further, the basic note data NT is composed of dot data BP indicating the presence / absence of a note (note) N and note number data NN indicating a note number, and the facial expression data EX is a staccato indicating the presence / absence of staccato, slur, and screaming. It consists of data ST, slur data SL, and shackle data SY. FIG. 2 (2) shows an example of a data format in the case where the minimum note is an eighth note, and each row represents data of an eighth note. That is, the original melody data is represented by a data set of a total of 8 rows per measure, and is stored for each storage area corresponding to each row in a predetermined area in the RAM 3. When the minimum note is a 16th note or a triple eighth note, the data is for every 16th note or triple eighth note, and the storage area is 16 or 12 per measure. .

  The reproduction melody data is data that is created by a reproduction melody data creation process, which will be described later, using the original melody data described above, and is stored in a predetermined area in the RAM 3. For example, as shown in FIG. Data TM1, TM2, TM3,... And event data EV1, EV2, EV3,... Are alternately and sequentially arranged in time series, and end with end data EDp.

  The event data EV1, EV2, EV3,... Includes note bend event data representing pitch bend in addition to note event data composed of note on event data NON representing note on and note off event data NOF representing note off. Timing data TM1, TM2, TM3,... Represents the time interval between the event represented by the previous event data and the event represented by the next event data, and the first timing data TM1 represents the elapsed time from the beginning of the music. In the example of FIG. 3, the first event data EV1 is note-on event data NON, and the second and third event data EV2 and EV3 are pitch bend event data.

[Automatic melody generation processing]
4 to 6 are flowcharts showing an automatic melody generation process according to an embodiment of the present invention. This process flow is a switch or display 14 of the operation switch device 13 for designating an automatic melody generation process mode. Start by operating the control buttons on the screen. In the first step M1 of this processing flow, various melody generation data are supplied. Such melody generation data is data necessary for melody generation, such as time signature, number of measures, passage configuration, range, key, chord progression, tempo, timbre, “feeling parameter”, etc. “Cheerful”, “Simple”, “Easy”, “Refreshing”, “Ukiuki” and other types of music, “Ikiiki degree 1” to “Ikiiki degree 5” graded “Ikiiki degree”, etc. It is a parameter for setting various feelings.

  As a method of supplying the data for generating melody, there is a method in which a user designates a value for each data type by using an operation button of the operation switch device 13 or an operation button on the display 14 screen. For the chord progression, a number of typical passage configurations and typical chord progressions may be stored, and the user may select one of them. As another data supply method, a large number of sets of values for a plurality of data types are stored as templates, and the user selects one of them, or from the external devices 17 and 20 and the external storage device 9. A method of supplying each data can be considered.

  In the next step M2, the hit point data BP for the entire song is generated. That is, one piece of hit point data BP representing the position of a note (note) is generated based on the time signature, the number of measures, the passage configuration, the feeling parameters, and the like. As an example of the generation of such hit point data BP, a lot of one-measure hit point patterns are stored for each time signature in the minimum note, and the hit point pattern (matching the time signature and feeling based on the time signature and feeling parameters ( The minimum note is selected according to the feeling) is selected for each measure, and this is connected by the number of measures to obtain the hit data for one song. At this time, for a passage that is the same as or similar to the preceding passage in the passage configuration, the dot data is also the same as or similar to the preceding passage. Then, each dot data BP is written into the dot storage area of the original melody data in the RAM 3. The hit point data BP may be generated by a method other than this.

  In the third step M3, note number data NN is generated and assigned to each dot (BP) to create basic note data NT. That is, note number data NN to be given to each hit point is generated based on the tone range, key, and chord progression, and given to each hit point. As an example of generation of such note number data NN, for the important hit points in each hit point data BP (for example, a hit point in a strong beat or a hit point in the vicinity when there is no hit point in a strong beat). , One of the sounds in the range of chords at the location of the hit point is randomly selected and assigned to the hit point, and the tone in the range is adjusted with respect to the remaining hit points. Are randomly selected from the notes of the scale and the notes of the available note scale of the chord and given to the hit points. At this time, only a sound satisfying the music rule may be selected with reference to a predetermined music rule. Further, the range of the vertical movement of the pitch may be regulated according to the feeling parameter. In this case, regarding the passage that is the same or similar to the preceding passage in the passage configuration, the note number data NN is also the same or similar to the preceding passage. Then, each note number data NN is written into the note number storage area of the original melody data. Note number data NN may be generated by other methods.

  When the basic note data NT composed of the dot data BP and the note number NN is created as described above, in the next fourth to sixth steps M4 to M6, the basic note represented by the basic note data NT is generated. Then, facial expression data EX is given. First, in the fourth step M4, the stat cart data ST is generated as the facial expression data EX and given to each hit point, that is, each basic note. That is, staccato data ST (whether or not staccato is indicated by “1” or “0”) indicating whether or not to add staccato to each note is generated based on the passage structure, feeling parameters, and the like. Give to each note.

  As an example of the generation of the staccato data ST in step M4, the frequency of staccato notes (appearance rate) is determined according to the stages “liveness degree 1” to “liveness degree 5” of the “liveness degree”, and according to this frequency Thus, it is possible to adopt a method of determining whether or not the staccato is randomly applied to each note (details will be described later). At this time, regarding the passage that is the same or similar to the preceding passage in the passage configuration, the staccato data ST is also the same or similar to the preceding passage. Then, each staccato data ST is written to the staccato storage area of the original melody data. Note that the staccato data ST may be generated by other methods.

  In the fifth step M5, slur data SL is generated as the facial expression data EX and given to each hit point, that is, each basic note. That is, based on the passage structure, timbre, staccato data ST, etc., the slur data SL (whether or not slur is indicated by “1” or “0”) indicating whether or not to add slur to each note is generated. Give to notes. As an example of the generation of such slur data SL, the frequency (appearance rate) of slur notes is determined according to the tone color, and for each note whose stacking data ST is “0”, according to the determined frequency. A method of determining whether to give randomly or not can be taken (details will be described later). At this time, regarding the passage that is the same or similar to the preceding passage in the passage configuration, the slur data SL is also the same or similar to the preceding passage. Then, each slur data SL is written into the original melody data staccato storage area. Note that the slur data SL may be generated by other methods.

  In the sixth step M6, the scooping data SY is generated as the facial expression data EX and given to each dot, that is, each basic note. In other words, based on the passage structure, tone color, staccato data ST, etc., the screaming data SY (whether or not screaming is indicated by “1” or “0”) indicating whether or not to squeal each note is generated. To grant. As an example of the generation of such scramble data SY, the frequency (frequency of appearance) of squeak notes is determined according to the tone color, and for each note whose staccato data ST is “1”, according to the determined frequency. A method of determining whether or not to give to the rantam can be taken (details will be described later). At this time, for a passage that is the same as or similar to the preceding passage in the passage configuration, the shackle data SY is also the same as or similar to the preceding passage. Then, each of the scouring data SY is written into the squeezing storage area of the original merotidake. Note that the scooping data SY may be generated by other methods.

  In the next step M7, reproduction melody data is created based on the original melody data composed of the data BP, NN, ST, SL, and SY obtained by the processing in steps M2 to M6. In this step M7, first, data representing the gate time GT of each reproduction note, that is, the period from note-on to note-off, is created based on the staccato data ST. Next, two notes to be slurred are combined based on the slur data SL, and pitch bend event data PBL for connecting the two notes with a smooth pitch change is created. Then, pitte bend data PBY that gradually changes from the pitch slightly below the note N to be squeezed to the pitch of the note based on the squeeze data SY is created. A detailed procedure for creating the gate time GT and the pitch bend event data PBL and PBY will be described later.

  In step M7, volume data may be created as necessary. For example, the end of the last long note of a predetermined section such as a passage or a phrase may be decremented. . The velocity for each note data NT may be constant, or may be varied randomly or according to a predetermined rule. For example, the velocity is determined according to the note number (NN) of each hit point (note N) (the higher the pitch, the higher the pitch), and the velocity is determined according to the average note number of a predetermined section such as a passage or phrase. A method of determining the velocity according to the hit point position in the measure (the strong beat is large and the weak beat is small) can be used.

  In the subsequent step M8, the created reproduction melody data is evaluated. In order to perform this evaluation, for example, an operation element of the operation switch device 13 or an operation element button on the screen of the display 14 is operated to reproduce the sound emitted from the sound system 12 via the sound source circuit 7 and the effect circuit 8. A musical tone based on the melody data for the player is auditioned, and the musical score data is displayed on the display 14 screen in the form of a musical score or the like for the melody data for reproduction. In step M8, the user evaluates the quality of the melody data by, for example, listening to or displaying the created reproduction melody data in this way, and the process proceeds to step M9.

  In step M9, as a result of the evaluation in the previous step M8, it is determined whether the created reproduction melody data is “OK” (good) or “NG” (bad), and if “OK” (= YES), step Proceeding to M10, in the case of “NG” (= NO), proceeding to Step M11, the processing of the facial expression data regeneration processing routine consisting of Steps M11 to M18 is executed. In order to determine the quality, for example, an “OK” / “NG” switch or button is provided as an operation element on the screen of the operation switch device 13 or the display 14 so that the user can select one. Alternatively, a “regenerate” switch or button may be provided, and “NG” may be determined when this switch is operated. Alternatively, an “output (playback, recording, output to outside, etc.)” switch or button may be provided, and when this is operated, “OK” may be determined.

  When the process proceeds to step M10, the melody data for generation determined as “OK” is reproduced from the sound system 15, recorded in the external storage device 9, or output to the external devices 17 and 19, etc. After performing the output operation, this automatic melody generation process is ended.

  On the other hand, if the created reproduction melody data is determined as “NG” in step M9 and the process proceeds to step M11, the reproduction melody data is recreated. At this time, various melody generation data are newly supplied. It is determined whether to correct or use the previous melody generation data as it is. This is because there are some changes when creating a melody again even if the data for generating various melody is the same as before, but if you create a new melody with various data for generating melody, the change will be larger. is there. This determination can be performed by the same operation as the operation example in step M9.

  If it is determined in step M11 that the various melody generation data is supplied again, new various melody generation data is supplied in step M12, and the process proceeds to step M13. If not, the process immediately proceeds to step M13. Note that the various melody generation newly supplied in step M12 may be all types of data described in the description of step M1 or only a part of the data.

  Steps M13 to M18 after step M13 include determination steps M13, M15, and M17 for determining whether or not to regenerate staccato data ST, slur data SL, and scoop data SY that have already been generated as facial expression data EX. When the facial expression data ST, SL, and SY are regenerated, the facial expression data regeneration steps M14, M16, and M18 are executed to regenerate each data to a required value.

  Whether or not to regenerate each data ST, SL, and SY in each of the determination steps M13, M15, and M17 may specify which data the user regenerates, or newly supplied melody generation data. It may be automatically specified according to the type. For example, when the timbre data is supplied again in step M12, the slur data SL and the scooping data SY are generated again, the staccato data ST is not generated again, and the “liveness” data is supplied again. Can adopt a method of regenerating the staccato data ST and not regenerating the slur data SL or the squeak data SY.

  In each facial expression data regeneration step M14, M16, M18, staccato data ST, slur data SL, and squeak data SY are regenerated and given to each dot, that is, each basic note, and stored in the corresponding storage area of the original melody data. Write. In this way, it is possible to regenerate only the facial expression data EX without changing the basic note data NT.

  After regenerating the facial expression data EX through these steps M13 to M18, the process returns to step M7 to regenerate reproduction melody data based on the regenerated facial expression data EX. The regenerated melody data for reproduction is evaluated in step M8, and it is determined whether or not “OK” is set in step M9. In the case of “OK”, the process proceeds to step M10 and the output process is performed to end the automatic melody generation process. Otherwise, the facial expression data regeneration process routine in steps M11 to 18 is performed again, and step M9 is performed. This regeneration process is repeated until “OK” is determined.

[Generation of staccato data]
In steps M4 and M14 of the automatic melody generation process (FIGS. 4 to 6), in order to determine whether or not to add staccato for each striking point (note), according to the data for generating melody such as “liveness”. The frequency of staccato notes is determined. In the case of using the “ikiki degree” for this determination, for example, as shown in FIG. 7, a “staccato frequency table” in which the staccato frequency in the entire melody to be generated is set in advance according to the “ikiki degree” data value. It is prepared in the ROM 2 or the external storage device 9. Then, the staccato frequency in the entire melody can be obtained by referring to this staccato frequency table in accordance with “live degree 1” to “live degree 5” supplied in steps M1 and M12.

  That is, in each note, staccato is given at a rate indicated by the staccato frequency corresponding to the “liveness degree” of the staccato frequency table, and the staccato data ST of the note to which the staccato is given is set to “1”. As an example of staccato assignment, first, random numbers of values “0” to “9” are generated for each note. When the staccato frequency is 40%, staccato is assigned if the random number value is in the range of “0” to “3”, and not assigned if the random number is in the range of “4” to “9”. On the other hand, when the staccato frequency is 60%, staccato is assigned if the random number is a value “0” to “5”, and is not assigned if the random number is a value “6” to “9”. When the staccato frequency is 0%, no staccato is given to all notes. Note that the staccato data ST may be generated by other methods.

[Generation of slur data]
In steps M5 and M16, the slur frequency is determined according to melody generation data such as timbre in order to determine whether to add or not add slur for each hit point (note). When a timbre is used for this determination, for example, as shown in FIG. 8, the slur frequency of all notes that are not staccato in the entire melody to be generated is set to a “slur frequency preset according to the timbre. A “table” is prepared in the ROM 2 or the external storage device 9. Then, by referring to this slur frequency table according to the tone color supplied in steps M1 and M12, the slur frequency in the non-staccato sound in the melody can be obtained.

  That is, in each note that is not a staccato, slur is given at a rate indicated by the slur frequency corresponding to the tone color type of the slur frequency table, and the slur data SL of the note with the slur is set to “1”. As an example of adding slur, first, random numbers of values “0” to “9” are generated for each note that is not a staccato. When the slur frequency is 20%, a slur is assigned if the random value is in the range of values “0” to “1”, and is not assigned if the value is “2” to “9”. On the other hand, when the slur frequency is 40%, a slur is assigned if the random number is a value “0” to “3”, and is not assigned if the random number is a value “4” to “9”. When the slur frequency is 0%, no slur is given to all non-staccato notes and no slur is given to notes that are staccato. Note that the slur data SL may be generated by other methods.

[Generation of scouring data]
In steps M6 and M18, in order to determine whether to give or not to squeak each note (note), the frequency of squealing is determined according to melody generation data such as timbre. When a timbre is used for this determination, for example, as shown in FIG. 9, the frequency of squealing in all the sounds that are staccato among the generated melody is set in advance corresponding to the timbre. The “scrubbing frequency table” is prepared in the ROM 2 or the external storage device 9. Then, by referring to this squeezing frequency table according to the tone color supplied in steps M1 and M12, the squeezing frequency in the staccato sound in the melody can be obtained.

  That is, in each note that is a staccato, squealing is given at a rate indicated by the squealing frequency corresponding to the tone type in the screaming frequency table, and the screaming data SY of the note to which screaming is given is “1”. As an example of scrambling, first, random numbers of values “0” to “9” are generated for each note that is a staccato. When the scrambling frequency is 30%, scrambling is assigned if the random value is in the range of values “0” to “2”, and not given if the random number is in the range of “3” to “9”. On the other hand, if the scrambling frequency is 50%, scrambling is assigned if the random number is a value “0” to “4”, and not given if the random number is a value “5” to “9”. In addition, when the squealing frequency is 0%, squealing is not given to all notes that are staccato and screaming is not given to notes that are not staccato. Note that the scooping data SY may be generated by other methods.

[Effects of facial expression data]
Next, with reference to FIG. 10 showing the “pitch-time” characteristics of a plurality of basic notes N1, N2, and N3, effects of facial expression data EX such as staccato, slur, and sneeze will be described. In FIG. 10, note-on timings t1, t2, t3,... Of each note N1, N2, N3 represent the hit point start timing indicated by the corresponding hit point data BP of the basic note data NT, and the pitch values P1, P2 , P3 represents a pitch represented by each corresponding note number NN of the basic note data NT. Each small circle of the mesh pattern represents the pitch position at the time of note-on of each of the notes N1, N2, and N3 represented by the basic note data NT, and each band represents the “pitch-time characteristic of each note by the expression data EX. "" Represents how it is changed.

  FIG. 10 (1) shows the effect of the expression data EX by the staccato data ST. When the staccato data ST is “1” and the staccato is applied, a predetermined proportion of all the basic note intervals until the timing of the next hitting point is obtained. It becomes gate time GT. Here, the first and third notes N1 and N3 have the staccato data ST of “1” and the staccato ratios of 50% and 75%, respectively, while the staccato data ST of the second note N2 is “0”. is there. Accordingly, as shown in FIG. 10 (1), the gate times GT1 and GT3 of the first and third notes N1 and N3 are all basic note intervals t1 to t1 from the hit point timings t1 and t3 to the next hit point timings t2 and t4. It becomes 50% and 75% of t2, t3 to t4, respectively. On the other hand, the second notebook N2 is not staccato.

  FIG. 10B shows the effect of the facial expression data EX by the slur data SL. When the slur data SL is “1” and a slur is applied, it is combined with the next note, and the pitch between both notes is smoothly connected. It is. Here, the slur data SL is “1” and the slur is applied to the second note N2, and as shown in FIG. 10 (2), the second note N2 is combined with the next third note N3, and the pitch P2, P2 between both notes N2, N3. P3 is smoothly connected.

  FIG. 10 (3) shows the effect of the facial expression data EX by the squealing data SY. When the squealing data SY is “1” and the squealing is applied, a predetermined value corresponding to a preset “shearing depth” is shown. Sound generation starts from a pitch lower by the pitch Py, and the pitch gradually changes to the pitch P1 of the note. Here, the first note N1 is screamed when the squealing data SY is “1”, and as shown in FIG. 10 (3), the first note N1 begins to sound from a pitch lower by the squealing depth Py, and gradually becomes the first note. It changes to the pitch P1 of N1.

[Gate time processing based on staccato]
FIGS. 11 to 12 show flowcharts of the gate time generation process based on the staccato data ST in the reproduction melody data generation process in step M7 (FIG. 4) of the automatic melody generation process of FIGS. When a new dot, that is, a basic note Ni is selected in the first step T1, [hereinafter, “i” (i = 1, 2,...) Is used as a subscript symbol indicating the note number from the beginning of the song. In step T2, whether or not the staccato data STi of the hit point (note Ni) is “1” is checked. If “1”, the process proceeds to step T3, and if not, the process proceeds to step T4. In step T3, it is checked whether the hit point (note Ni) is the last hit point (final note) of the phrase or the passage. If it is the last hit point, the process proceeds to step T4. Otherwise, the process proceeds to step T5. When the process proceeds to step T4, the interval between the start timing ti of the hit point (note Ni) and the start timing ti + 1 of the next hit point (note Ni + 1) is set as the gate time GTi of the note Ni.

  On the other hand, when the process proceeds to step T5, processing for applying staccato is sequentially performed through steps T6 to T9. First, in step T5, the first staccato ratio is obtained by referring to the staccato ratio table based on the data representing the musical taste such as “slowly” / “simple” /... In the next step T6, the second staccato ratio is obtained by referring to the staccato ratio table based on the data representing the “degree of excitement”. Further, in step T7, the first and second staccato ratios are synthesized.

  FIG. 13 shows an example of a staccato ratio table used in steps T5 and T6. In the first staccato ratio table of FIG. 13 (1), for example, the first staccato ratio is set corresponding to each musical taste such as “Carefully”, “Simple”, “Easy”, “Sawayaka”, and “Ukiuki”. The second staccato ratio table of FIG. 13 (2) corresponds to, for example, the second staccato corresponding to each “liveness degree” divided into “liveness degree 1” to “liveness degree 5”. The ratio is set. In addition, the staccato ratio table is not limited to the fixed correspondence of the staccato ratio in this way, but the range of staccato ratio is determined according to “loosely” / “simple” /.../ “sparkling” or “degree of excitement”. It may be determined (for example, “40 to 60%” when “liveness” = 3), and may be determined randomly within this model.

  For the synthesis of the first and second staccato ratios in step T7, for example, an average value of the first and second staccato ratios obtained in steps T5 and T6 is taken (a simple average may be used, Any method may be employed, such as a weighted average according to the degree of importance of “liveness” or a method of multiplying both staccato ratios.

  The staccato ratio synthesized in this way is corrected in step T8 according to the tempo of the song (step M1: FIG. 4). For example, a predetermined tempo (for example, 90) is used as a reference (= no correction), and if the tempo is earlier than this, the staccato ratio is increased, and if the tempo is slower, the correction is performed according to such a tempo. be able to. Further, as a correction method, a correction method based on a calculation formula, a correction value table corresponding to the tempo, and a correction method using this table can be considered.

  In step T9, the corrected staccato ratio is multiplied by the interval between the start timings ti and ti + 1 of the corresponding hit point (note Ni) and the next hit point (note Ni + 1), and the multiplication result is used as the gate time GTi of the note Ni. After the gate time acquisition process at step T4 or step T9, the process proceeds to step T10. Note that the gate time GT obtained in steps T4 and T9 may be quantized with a predetermined note (for example, a 16th note or a 32nd note).

  In step T10, the hit point start timing ti is set as the timing data (TM) of the note-on event data NONi corresponding to the hit point (note Ni), and the note-on event data NONi and the timing data (TM) are reproduced in the melody data for reproduction in the RAM 3. In the next step T11, the timing data (TM) of the note-off event data NOFi corresponding to the hit point (note Ni) is determined based on the gate time GTi, and the note-off data NOFi and the timing data are reproduced. Then, the process proceeds to step T12.

  In step T12, it is determined whether the hit point (note Ni) is the final hit point (final note). If the hit point is the final hit point, the gate time generation process is terminated. If not, the process returns to step T1 again. Processing for acquiring the gate time in steps T1 to T11 is performed, and this gate time acquisition processing is repeated until it is determined in step T12 that the processing has been performed up to the final hit point.

[Pitch bend event data creation based on slur and note combination processing]
FIG. 14 shows a flowchart of pitch bend event data creation and note combination processing based on slur in the reproduction melody data creation processing in step M7 (FIG. 4) of the automatic melody generation processing of FIGS. When a new hit point, that is, a note Ni is selected in the first step L1, it is checked in step L2 whether the slur data SLi of the hit point (note Ni) is “1”. Sequentially proceed to steps L3 to L5, otherwise proceed to step L6.

  In step L3, the pitch bend event gradually changes from the vicinity of the timing of the note-off event data NOFi corresponding to the hit point (note Ni) stored in the reproduction melody data memory in the RAM 3 toward the pitch Pi + 1 of the next note Ni + 1. Data PBLi is created and inserted. Details of creation of pitch bend event data PBL that gradually changes based on such slur data SL will be described later.

  The next steps L4 and L5 are steps for joining two notes Ni and Ni + 1. First, in step L4, the timing of note-off event data NOFi corresponding to the hit point (note Ni) is corrected to the timing of note-off event data NOF + 1 corresponding to the next hit point (note Ni + 1), and then step L5. The note-on event data NONi + 1 and note-off event data NOFi + 1 corresponding to the next hit point (note Ni + 1) and their timing data are erased, and the process proceeds to step L6.

  In step L6, it is determined whether the hit point (note Ni) is the final hit point (final note). If the hit point is the final hit point, the pitch bend event data creation and note combination processing is terminated. Returning to step L1, the processes of steps L1 to L5 are performed, and these processes are repeated until it is determined in step L6 that the process has been completed up to the final hit point.

[Procedure for creating pitch bend event data in slurs]
FIG. 15 is a diagram for explaining a method of creating pitch bend event data PBLi that gradually changes in accordance with the slur data SLi in step L3 described above, and FIGS. 16 to 17 show such a method in step L3. It is a flowchart showing the process which produces the pitch bend event data PBLi in a slur based.

  When the pitch bend event data creation process in the slur starts, first, in the first step LB1, the pitch difference Pdi between the notes Ni and Ni + 1 from the pitch Pi of the note Ni and the pitch Pi + 1 of the next note Ni + 1 [unit: eg: cents] (Cent)] is calculated, and then in step LB2, a standard transition time [unit: for example: millisecond (m.sec)] corresponding to the pitch difference is calculated. In order to change the transition time according to the number of pitches in this way, for example, the transition time when the pitch difference is 100 cents and the transition time when the pitch difference is 1200 cents are preset and stored in the ROM 2 or the like. A method of calculating the standard transition time corresponding to the pitch difference Pdi by linearly interpolating between the transition times with the cent value of the pitch difference Pdi can be employed.

  In the next step LB3, the standard transition time calculated in step LB2 is corrected according to the tempo of the music (step M1: FIG. 4), and the process proceeds to step LB4. In order to correct the transition time according to the tempo, for example, the slur transition time correction value (for example, 100 to 150%) when the tempo value is “60” and the tempo value is “120”. A slur transition time correction value (for example, 66 to 100%) is preset and stored in the ROM 2 or the like, and linear interpolation is performed between both slur transition time correction values with the standard transition time calculated in step LB2. Thus, it is possible to employ a method of calculating a corrected standard transition time corresponding to the calculated standard transition time. It should be noted that the transition time value corrected in accordance with the tempo in step LB3 also represents the standard transition time at the tempo.

In step LB4, the number of transition clocks corresponding to the standard transition time corrected in step LB3 is calculated, the process proceeds to step LB5, and the number of transition clocks is corrected in steps LB5 to LB9. The number of transition clocks can be calculated by the following equation (1) from the relationship (in this embodiment) that one clock is 1/48 of a quarter note. As seen from the number of transition clocks, the number of clocks is used instead of time units such as milliseconds to represent the time transition, but the number of clocks also substantially represents “time”. You can say:

  Of steps LB5 to LB9, steps LB5 and LB8 are based on the standard transition time when the pitch transition as shown by the stepped broken line in FIG. 15 is performed (in FIG. 15, for convenience of explanation). This is a processing block for determining whether or not a predetermined pitch stable period can be ensured. Steps LB6 and LB9 are steps for determining a predetermined pitch stable section before and after the pitch transition. This is a processing block for shortening the number of transition clocks when it cannot be secured.

First, in step LB5, it is determined whether or not the number of transition clocks is ½ or more of the pronunciation length of the next note Ni + 1. If the next note sound length (ti + 1 to ti + 2) is ½ or more (YES), the process proceeds to step LB6, and in the second half of the next note Ni + 1, the pitch stable interval of the next note Ni + 1, that is, the next note pitch Pi + 1 changes. In order to ensure at least half of the next note length so that the slur may be continuous with short notes as a horizontal interval that does not,
Number of transition clocks = 1/2 of the note length of the next note (2)
As described above, after the number of transition clocks is shortened, the process proceeds to step LB7. On the other hand, if “NO” (NO) is determined in step LB5, it is determined that the pitch stable section of the next note Ni + 1 has already been secured, and the process immediately proceeds to step LB7.

In step LB7, the number of delayed clocks corresponding to the times ti to tid in FIG. 15 is calculated by the following equation (3) (note sound length is represented by the number of clocks, which is the same in the following). Proceed to LB8:

In step LB8, it is determined whether or not the number of delayed clocks is less than ½ of the note length of the note Ni. If the note sound length (ti to ti + 1) is less than ½ (YES), the process proceeds to step LB9 to shorten the number of transition clocks. In this shortening process, the note Ni is delayed so that the pitch of the first half does not change, and the second note Ni + 1 is kept horizontal without changing the pitch of the second half. The number of clocks is the number of delayed clocks = 1/2 of the note length of the note (4)
And the number of transition clocks is calculated by the following equation (5):

In step LB8, when the number of delayed clocks is ½ or more of the note sound length (NO), or after finishing the process of step LB9, the process proceeds to step LB10, and the transition pitch per clock is expressed by the following equation. Calculate by (6):

  Then, the process proceeds to the final step LB11. Based on the delay clock number, the transition clock number, and the transition pitch per clock generated as described above, the time point tid when the delay clock number has elapsed from the note-on timing ti of this note. Starting from, pitch bend event data PBLi that changes by a transition pitch per clock is generated for each clock, and this pitch bend event data is inserted by the number of transition clocks.

  According to the pitch bend event data PBLi created in this way, first, the note Ni until the time tid at which the number of delay clocks has elapsed from the note on timing ti given by the note on timing data (TM) of the note Ni. As a pitch stable section in which the note pitch pi given by the note-on event data NONi does not change, at least a half of the note length of the note Ni is secured.

  Next, during the transition period from the time point tid to the time point tis where the number of transition clocks elapses, only the transition pitch is generated for each clock of the next note Ni + 1 from the pitch pi of the note Ni toward the pitch pi + 1 of the next note Ni + 1. The pitch can be gradually changed. Note that the note-on timing ti + 1 given by the timing data (TM) of the note-on event data NONi + 1 of the next note Ni + 1 is almost the center timing of this transition section as shown in FIG.

  Thereafter, at the time tis delayed from the note-on timing ti + 1 of the next note Ni + 1, the next note pitch pi + 1 given by the note-on event data NONi + 1 of the next note Ni + 1 is reached. Thereby, a natural effect corresponding to the slur is obtained. From the time tis to the note-off timing (note-on timing of the next note) ti + 2, at least the next note Ni + 1 as a pitch stable section where the next note pitch pi + 1 does not change so that slurs may continue in short notes. The time of 1/2 of the sound length is secured. Therefore, when a slur is continued with short notes, the next note Ni + 1 represented by the basic note data NTi (note on timing ti + 1, pitch pi + 1) is sequentially subjected to the same processing as the previous note Ni to thereby obtain the next note. As soon as the pitch pi + 1 is reached, the next pitch transition can be started.

[Pitch bend event data creation process based on scouring]
FIG. 18 shows a flowchart of the pitch bend event data creation process based on the scream data in the reproduction melody data creation process in step M7 (FIG. 4) of the automatic melody generation process of FIGS. When a new hit point, i.e., note Ni is selected in step Y1, it is checked in step T2 whether or not the scoring data SYi of the hit point (note Ni) is "1". Then, the process proceeds to step Y4, and if not, the process proceeds directly to step Y4.

  In step Y3, the note starts from the vicinity of the timing of the note-on event data NONi corresponding to the hit point (note) Ni stored in the reproduction melody data memory in the RAM 3 and starts with the squeezing depth indicated by the screaming depth data Py. Pitch bend event data PBYi that gradually changes toward the pitch Pi + 1 of Ni + 1 is created and inserted.

  In step Y4, it is determined whether the hit point (note Ni) is the final hit point (final note). If the hit point is the final hit point, the pitch bend event data 1 creation process is terminated. If not, step Y1 is performed again. Returning to Steps Y1 to Y3, the process is repeated until it is determined in Step Y4 that the process has been completed up to the final hit point.

  Here, the procedure for creating the pitch bend event data PBYi that gradually changes according to the scooping data SYi in step Y3 will be described. The process in step Y3 is similar to the process of creating pitch bend event data PBLi using slur data SLi in step L3 of the slur-based data process in FIG. Although only specific enumerations are omitted and specific flowcharts such as FIGS. 16 and 17 are omitted, those skilled in the art will be able to realize them very easily.

[1] In scribing, as shown in FIG. 19, sound generation starts from a pitch Pi-Py that is slightly lower than the original pitch Pi of the note Ni, and the pitch is gradually increased to cause pitch transition. And the original pitch Pi is reached at the time point tis when the predetermined transition time has elapsed. Therefore, the scribing depth Py (cent) and the scribing transition time [unit: for example: millisecond (m.sec)] are set in advance and stored in the ROM 2 or the like, and the scribbling data SYi = "1" in step Y2. Is detected, these data are read out accordingly.

[2] Next, the read transition time is corrected with the tempo. As in the case of the slur, this correction is performed, for example, when the tempo value is “60” and the jerky transition time correction value (for example, 100 to 150%) and the tempo value is “120”. By presetting a scouring transition time correction value (for example, 66 to 100%) and storing it in the ROM 2 or the like, and linearly interpolating between the shacking transition time correction values with the shave transition time read in [1]. It is possible to employ a method of calculating a corrected squeezing transition time corresponding to the calculated sneezing transition time.

[3] Using the following equation (7), calculate the number of transition clocks corresponding to the crushed transition time corrected in [2]:

[4] If the calculated transition clock number is 1/2 or more of the note length of the note Ni, the transition clock number is shortened,
Number of transition clocks = 1/2 of the note length of the note (8)
And

[5] If the actual pronunciation length corresponding to the note Ni is short, such as the actual pronunciation length is shortened by the staccato data STi as shown in FIG. Or shorten the transition time. For example, if the actual sounding length clock number is less than “12”, squealing is stopped, and the actual sounding length clock number is “12” or more,
Number of transition clocks> Number of pronunciation length clocks-6 clocks (9)
In this case, the number of transition clocks is shortened by the following equation (10):
Number of transition clocks = Actual number of sound generation clocks-6 clocks (10)

[6] Further, based on the scooping depth Py read out in [1] and the number of transition clocks determined through [3] to [5], the following equation (11) Calculate the transition pitch:
Transition pitch = (shear depth) / (number of transition clocks) (11)

[7] Based on the number of transition clocks obtained as described above and the transition pitch per clock, it starts from the note-on timing ti of the note Ni and changes by the transition pitch per clock every clock. Pitch bend data PBYi is created and inserted by the number of transition clocks.

[Melody generation processing considering instrument conditions]
According to one embodiment of the present invention, a melody suitable for a musical instrument to be sounded (played) can be generated by setting a melody generation parameter in consideration of conditions under which the musical instrument can be played. FIG. 20 is a functional block diagram of a melody generation system that takes into account the musical instrument playable conditions according to one embodiment of the present invention. In this system, an external storage device 9 is used to provide a musical instrument independent melody generation database B1, a musical instrument performance condition database B2, and a performance condition change database B3. Then, in order to set the music conditions, the music style is designated in the music style designation block B4, and the instrument type is designated in the instrument type designation block B5. An appropriate desired melody is generated.

  Here, the instrument-independent melody generation database B1 stores various parameters such as key, oct / 16/16 triplet (note resolution parameter), time signature, etc. for each musical style. It has data for melody generation. On the other hand, in the musical instrument playable condition database B2, for each musical instrument type, a generated sound range, a jump pitch width (“jump” refers to performing at a short interval of 3 degrees or more), and a jump frequency. Various condition parameter information that can be played, expressed, or easily played by the musical instrument, such as expression performance information and key (wind instrument only), is prepared in advance as playable condition data. The “generated sound range” parameter is data dependent on the instrument, but the “jumping pitch width” parameter may be a combination of instrument independent data.

  Further, the playable condition changing database B3 has playable condition changing data for changing the musical instrument playable condition data for each musical style. The playable condition change data may be data for changing based on the original data, or may be given as absolute data in the case where it is assumed that the performance is performed with an electronic sound source with no restrictions on performance. . Further, it may be data that expands the original data.

  When the music style is designated by the music style designation block B4, the musical instrument type is designated by the musical instrument type designation block B5, and the music condition is set, the first extraction block B6 is independent of the musical instrument according to the designated musical style. Sex melody generation data is extracted, and in the second extraction block B7, instrument performance enabling condition data corresponding to the specified instrument is extracted. In accordance with the song style designation in the song style designation block B4, the performance condition change data corresponding to the designated song style is extracted from the performance condition change database B3, and in the data change block B8, The musical instrument performance condition data extracted in the second extraction block B7 is changed according to the performance condition change data. This data change may be performed when the original data from the second extraction block B7 is processed, and may be simply replaced when changed to absolute data.

  Then, in the melody generation block B9, based on the instrument-independent melody generation data extracted in the first extraction block B6 and the instrument performance condition data changed in the data change block B8, A desired melody composed of predetermined rhythm data, pitch data, facial expression data and the like satisfying the condition parameters suitable for the designated musical instrument with music conditions is generated. For the generation of a melody including facial expression data, the specific methods already described in detail with reference to FIGS.

  The melody generation data includes a parameter group that depends on the musical instrument to be played and a parameter group that does not depend on the musical instrument. As described above, the melody generation data that does not depend on the musical instrument includes a parameter indicating the resolution of the melody sound to be generated, such as “eighth system”, “16th system”, and “triplet system”, “3 There are parameters representing time signatures such as “time signature” and “four time signatures”, or parameters such as keys, which are prepared in advance in the instrument-independent melody generation database B1. Such melody generation data that does not depend on an instrument can be set by designating the song style in the song style designation block B4, or directly set in the block B4 regardless of the song style designation. Also good. However, since there are keys that are easy to play and keys that are difficult to play, the key can be handled as data depending on the instrument depending on the instrument type.

  FIG. 21 shows specific examples of various data for melody generation prepared in the melody generation system in consideration of the musical instrument playable condition according to one embodiment of the present invention. FIG. 21 (1) is a specific designation example of the instrument-independent data (B1). According to “Curve Style Designation”, as shown in the upper two rows of FIG. 21 (1), each condition parameter set in the columns (A) to (C) is designated. Then, the condition parameters in each column (A) to (C) are individually specified.

  FIG. 21 (2) is a specific example of instrument dependency data, that is, instrument performance condition data (B2). Here, “legato” is a smooth performance and refers to a performance in which sound is superimposed for a moment when moving from note to note, and “staccato” is a performance in which the sound is cut short. “Slur” is a performance in which, when moving from one sound to the next, the sound is continuously continued as before and the pitch is changed.

  FIG. 21 (3) is a specific example of the performance condition change data (B3). Although the performance condition change data corresponding to the song style designation such as “quiet” is instrument-independent, the “compression” rate may be dependent on the instrument. For example, in the case of a piano, since the sound range is wide, the compression ratio is set to ¼ in order to create a “quiet” atmosphere. That is, even in the same “quiet” style, the compression rate value may be varied depending on the musical instrument.

  Next, the processing of the melody generation data depending on the musical instrument to be played will be specifically described with reference to an example in which “human voice (soprano)” is designated in the musical instrument type designation block B5. For the generated sound range, the second extraction block B7 extracts, for example, a sound range condition parameter set in the range of “C4 to E5” from the musical instrument playable condition database B2. Here, when a song style is designated in the song style designation block B4 and, for example, a melody generation condition of “quiet” is given, this range is played in the data change block B8 via the performance condition change database B3. The parameter “C4 to E5” is changed, and “E4 to C5” (= the center of “C4 to E5” is maintained and the range is compressed to ½) is generated narrowly. Here, as a specific change processing method, there is a possibility of deviating from the original range when determined by the absolute range, so the process of “holding the center of the original range and compressing the range to ½” It is better to adopt a method. In this case, it may be about “1/2” instead of being completely set to “1/2”.

  For the jump distance (jump pitch width), for example, jump distance condition data set to “up to one octave” is extracted from the musical instrument playable condition database B2 by the second extraction block B7. Here, when a song style is designated in the song style designation block B4 and a melody generation condition of “quiet” is given, for example, this jump is made in the data change block B8 via the performance condition change database B3. Change the distance “up to 1 octave” and make it narrower, such as “up to 5 degrees”. Here, as a specific change processing method, a processing method of “compressing the original distance to ½” or the like is adopted. In this case, it may be about “1/2” instead of being completely set to “1/2”.

  Further, for the jump frequency, the second extraction block B7 extracts jump frequency condition data set to, for example, “30%” of the total number of notes from the musical instrument performance condition database B2. Here, when a song style is designated in the song style designation block B4 and a melody generation condition of “quiet” is given, for example, this jump is made in the data change block B8 via the performance condition change database B3. The frequency “30%” is changed, and the frequency is reduced to “15%” of the total number of notes. Here, as a specific change processing method, a processing method of “compressing the original numerical value to ½” or the like is adopted. In this case, it may be about “1/2” instead of being completely set to “1/2”.

  Assuming that the generated melody is sounded by an electronic sound source, the playable condition data set in the musical instrument playable condition database B2 is not able to be played by the corresponding natural musical instrument. The state may be set.

[Various Embodiments]
As described above, the performance data generation process of the present invention has been described according to the embodiment. For example, the present invention can be applied to the case where a facial expression is given to performance data having no facial expression recorded by the step recording method or performance data having no facial expression recognized by music score recognition or voice recognition.

  In the embodiment, as shown in FIG. 2, staccato data ST in which the presence / absence of staccato data in the original melody data is expressed by “1” / “0” is stored, and ST = “1” (with staccato) ), The staccato ratio (%) is determined with reference to the staccato ratio table, but the staccato ratio (“100%” when there is no staccato) is directly stored for all the note data. May be.

  The contents (arguments and numerical values) of the staccato frequency table and the staccato ratio table are not limited to those illustrated. The same applies to the slur frequency table.

  The facial expression accompanied by the pitch change is realized by inserting pitch bend event data at fixed timings, but is not limited thereto. For example, the pitch bend event data insertion timing may not be constant. Specifically, a template for a pitch change curve may be prepared, and data specifying this may be inserted. In this case, regarding correspondence to different transition times, a plurality of templates having different transition times may be prepared and a template may be selected according to the transition time, or the template may be deformed according to the transition time. It may be. Or you may make it insert the command which designates pitch changes, such as portamento control.

  The form of the system is not limited to the form of the electronic musical instrument, but may be the form of a personal computer + application software. When taking the form of an electronic musical instrument, the form is not limited to a keyboard musical instrument, and may be a stringed musical instrument type, a wind instrument type, a percussion instrument type, or the like. In addition, the sound source device, the automatic performance device, etc. are not limited to those built in one electronic musical instrument body, but each is a separate device, and each device is connected using communication means such as MIDI or various networks. There may be.

  Further, the processing program and various data used for the processing may be supplied to an electronic musical instrument or a personal computer from an external storage medium or from an external device via a communication interface.

  Next, in relation to automatic performance, the format of the created performance data includes “event + relative time” indicating the time of occurrence of the performance event as the time from the previous event, and the time of occurrence of the performance event. "Event + absolute time" expressed in absolute time within a song or measure, "pitch (rest) + note length" in which performance data is expressed in pitch and note length or rest and rest length, Any format may be used, such as a “solid method” in which a memory area is secured for each minimum performance resolution and the performance event is stored in the memory area corresponding to the time at which the performance event occurs.

  Also, on the memory, time-series performance data may be stored in a continuous area, or data stored in a scattered area may be separately managed as continuous data. (That is, it only needs to be managed as time-sequential data, and it does not matter whether the data is continuously stored in the memory).

  Finally, in relation to the MIDI interface, not only a dedicated MIDI interface but also a MIDI interface using a general-purpose interface such as RS-232C, USB (Universal Serial Bus), IEEE 1394 (I Triple 1394), etc. It may be configured. In this case, data other than MIDI messages may be transmitted and received simultaneously.

FIG. 1 is a block diagram showing a hardware configuration of a performance data creation apparatus according to an embodiment of the present invention. FIG. 2 is a diagram showing the data format of the original melody data used in one embodiment of the present invention. FIG. 3 is a diagram showing a data format of reproduction melody data used in one embodiment of the present invention. FIG. 4 is a first part (1/3) of the flowchart showing the automatic melody generation process according to one embodiment of the present invention. FIG. 5 is a second part (2/3) of the flowchart showing the automatic melody generation process according to the embodiment of the present invention. FIG. 6 is a third part (3/3) of the flowchart showing the automatic melody generation process according to the embodiment of the present invention. FIG. 7 is a diagram showing an example of a staccato frequency table used in one embodiment of the present invention. FIG. 8 is a diagram showing an example of a slur frequency table used in one embodiment of the present invention. FIG. 9 is a diagram showing an example of a scooping frequency table used in one embodiment of the present invention. FIG. 10 is a “pitch-time” characteristic diagram for explaining the effect of applying facial expression data according to an embodiment of the present invention. FIG. 11 is a part (1/2) of a flowchart showing the gate time creation process based on the staccato data in the reproduction melody data creation process according to the embodiment of the present invention. FIG. 12 is another part (2/2) of the flowchart showing the gate time creation processing based on the staccato data in the reproduction melody data creation processing according to the embodiment of the present invention. FIG. 13 is a diagram showing an example of a staccato ratio table used in one embodiment of the present invention. FIG. 14 is a flowchart showing pitch bend event data creation and note combination processing based on slur data in the reproduction melody data creation processing according to one embodiment of the present invention. FIG. 15 is a “pitch (cent) -time (number of clocks)” characteristic diagram for explaining a mode of creating pitch bend event data that gradually changes in slur data according to one embodiment of the present invention. FIG. 16 is a part (1/2) of a flowchart showing a pitch bend event data creation process in slur data during data processing based on slur data according to one embodiment of the present invention. FIG. 17 is another part (2/2) of the flowchart showing the pitch bend event data creation processing in the slur data during the data processing based on the slur data according to one embodiment of the present invention. FIG. 18 is a flowchart showing pitch bend event data creation processing based on squeaky data in the reproduction melody data creation processing according to one embodiment of the present invention. FIG. 19 is a “pitch (cent) -time (number of clocks)” characteristic diagram for describing a mode of creating pitch bend event data that gradually changes during scrambling according to one embodiment of the present invention. FIG. 20 is a functional block diagram of a melody generation system that takes into account the musical instrument playable conditions according to one embodiment of the present invention. FIG. 21 shows specific examples of various data for melody generation handled in the melody generation system in consideration of the musical instrument playable condition according to one embodiment of the present invention.

Explanation of symbols

N1, N2, ...; Ni, Ni + 1, Ni + 2, ... Basic notes (notes),
P1, P2, ...; Pi, Pi + 1 Basic note (note) pitch,
t1, t2, ...; ti, ti + 1, ti + 2 Note-on timing of basic notes (notes),
Py scooping depth,
Pis pitch difference between the basic notes Ni and Ni + 1,
tid When a horizontal stable period of a predetermined number of delay clocks has elapsed from the note on timing ti of the basic note Ni
In the tis slur, the note-on timing ti (pitch Pi) of the basic note Ni at the time of reaching the pitch Pi + 1 of the next note Ni + 1 through the pitch transition from the time point tid (pitch Pi) after the horizontal stabilization period of the note Ni. A point in time when the original pitch Pi is reached via a pitch transition from -Py).

Claims (2)

Music condition designating means for designating musical conditions that do not depend on an instrument, including at least key, note resolution, time signature, and song style for the performance data to be generated;
Instrument designating means for designating an instrument to be played with the performance data;
Parameter acquisition means for acquiring performance data generation parameters depending on the musical instrument designated by the musical instrument designation means, including at least a range, a jump distance, a jump frequency, and an expression method;
Parameter changing means for changing at least the jump distance, the jump frequency, and the expression method included in the performance data generation parameter acquired by the parameter acquisition means based on the musical composition included in the music condition specified by the music condition specifying means; ,
A performance data generating apparatus comprising: performance data generating means for generating performance data based on the music conditions specified by the music condition specifying means and the performance data generating parameters changed by the parameter changing means. A music condition specifying step for specifying music conditions that do not depend on an instrument, including at least key, note resolution, time signature, and song style for the performance data to be generated;
An instrument designating step for designating an instrument to be played with the performance data;
A parameter acquisition step for acquiring performance data generation parameters depending on the instrument specified in the instrument specifying step, including at least a range, a jump distance, a jump frequency, and an expression method;
A parameter changing step for changing at least the jumping distance, the jumping frequency, and the expression method included in the performance data generation parameters acquired in the parameter acquiring step based on the music style included in the music conditions specified in the music condition specifying step; ,
A program for causing a computer to execute a procedure comprising a performance data generation step for generating performance data based on the music conditions specified in the music condition specification step and the performance data generation parameters changed in the parameter change step is recorded. A computer-readable recording medium for generating performance data.

Images