JP2715412B2 - Music device with melody analysis function and automatic composer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention]   The present invention relates to a music device with a melody analysis function and an automatic music composition.
About the machine. [background]   One of the important factors to consider regarding the quality of an automatic composer
One is a song that humans have been familiar with, in other words,
The potential to create songs that are more musical than purely mechanical
Whether the composer has the ability
And   For example, in Japanese Patent Application No. 56-125603 (Japanese Patent Application No. 60-40027)
Is derived from a series of pitch data (for example, 12-scale data)
Individual pitch data is sampled at random and sampled.
If the satisfies the limited conditions,
The melody is adopted if the conditions are not met.
The sample is not sampled again and the condition is checked again.
An automatic composer of a system that repeats the above is disclosed. But
The melody generation process of this automatic composer is basically
Is a try-and-error method. Land pitch data
At the sample stage, the pitch sequence was completely disordered.
I'm up. If you keep this chaotic pitch line,
Cannot be realized as a di (by astronomical contingency
Good melody may be possible). So this
In order to bring some order into disorder, condition inspection and
This is a kind of filtering. this
In such cases, the degree of selection is an important factor. Sorting is too tight
Then, the generated melody will be one pattern
If it is too loose, the original disorder will dominate.
Would.   The above automatic composer is a melody
Rather than a melody that is elusive in style
Suitable for composing, mainly for listening training and performance training
It is effective as a music composition device for learning.
Music is generally difficult to transcribe and play). In this sense
I don't have the abilities I mentioned.   In view of these points, the applicant has received a request from the user.
The motif will be exhibited inspired by the empowered motif
Open, grow and control your song's consistency, variety and hierarchy
Has applied for an automatic composer that can control
"Automatic composer" filed on April 8, 1987 and 1987
Filed on May 20). As its basic configuration,
The means to evaluate and analyze the obtained motifs and the evaluation results
Melody to be generated based on a certain motif feature parameter
Melody system to plan outline, style and associate
Control parameter generation means and planned control parameters
A method for generating a melody specifically according to chord progression information
There is disclosed a lody generation execution unit.   The automatic composers according to these applications provide the overall characteristics of the song,
Controls the mood and style, planning to compose
It is possible to create songs that are rich in diversity and hierarchy
It is. This is one of the great advantages. Second
The feature is that the motif is entered by the user,
The point is that the composition is made based on the music. This feature is
Is effective in deepening the satisfaction of people and raising their awareness of composing
To work.   The third feature is that Japanese speech and non-Japanese speech are treated separately.
That is, the approach is adopted. That is, on
Motif evaluation means (motif feature parameter extraction method
Dan) contains non-harmonic and harmony sounds contained in the motif.
Function to extract harmony feature parameters
Melody generation means (melody control parameter generation means
Extracted harmony to the dan and melody generation execution means)
Characteristic parameters (for example, the number of harmony, the type of variance,
Melody in a format that reflects the number of non-harmonic tones
Features included. Such harmony / non-harmony
The distinction function makes the generated song rich in music rather than mechanical
Help to make things.   Further, although related to the third feature, the harmony feature pattern
The parameter extraction means is a melody-harmonic analyzer
Can be used. Learning through melody analyzer
The melody line shows the musical
Can easily learn the functions of working and harmony
You.   In the above application, it shall be related to melody analysis.
Thus, a non-harmonic classification extraction means is shown. This non-harmony
The logic of the sound classification and extraction means is the pitch of the motif (melody)
The sequence is converted to a pitch progression matrix, and this pitch progression
It is compared with the conditions (patterns) of vocal sounds, and if they match,
The corresponding sound in the chief (melody) is the non-harmonic
It is estimated that. Pitch progression from pitch train
Conversion to a row is performed by means of calculating the difference in pitch between the preceding and following notes
doing.   The present invention uses a new logic that is different from the logic described above.
It is characterized by extracting or classifying harmony sounds
You. [Object of the invention]   That is, the object of the present invention is a novel approach.
Melody that automatically performs harmony evaluation of the melody
An object of the present invention is to provide a music device with an analysis function. Another
Is an automatic composer of the type described above, i.e.
Such patent application, entitled "Automatic composer", April 8, 1987
Application (Japanese Patent Application No. 62-86571) and filed on May 20, 1987
The automatic composer described in (Japanese Patent Application No. 62-121037) was improved.
In particular, with respect to motif analysis.
To provide an automatic composer. [The gist of the invention]   To achieve the above object, according to the present invention, pitch and pitch
To input a string of sound data containing
Give the chief an entry and special features that characterize the input motif.
Motif feature parameter extraction method to extract signature parameters
A step and a code representing the musical background progression of the melody to be generated
Code progress storage means for storing at least code progress information
And the chord progression information and the extracted feature parameters
Melody generating means for generating a melody based on the
And the input motif in the automatic composer having
Code input means for inputting a code representing the music background
The motif feature parameter extracting means is provided
For each sound of the selected motif,
Chief tone name) and the input chord components
Compare the motif sound names in the input chords
If it contains a note with the same note name as
Is determined as a non-harmonic tone if not included
Harmony / non-harmonic sound determination means.
An automatic composer is provided.   In the case of this configuration, the following advantages are obtained. You
In other words, for users, especially those with a musical background
Is the music world you want to express, even with the same melody line
In relation to the world, some of the leading code candidates
From among these, you can select a favorite code. this
On the other hand, this automatic composer
The harmony / non-harmonic sound determination means included in the output means
To identify or extract non-harmonic sounds contained in the motif
You. Then, from this extraction result, the harmony feature parameters
(For example, the number of chords, the number of non-chords, the pitch pattern
Is obtained. This information is the melody of the automatic composer
Melody is generated according to this information.
You. Users don't like the melodies they've made
For example, the next most likely code is the motif code.
Can be entered. This time, the extracted harmony
Chief feature parameters will be different from before,
These parameters are reflected in the generated melody
So, the finished melody is naturally different from before
Become something. By following this process, you
It is hoped that the melody desired by the user will be obtained more effectively
I will be waiting.   In a preferred configuration example, the motif feature parameter extraction is performed.
The sounding means further includes, for each sound determined as a non-harmonic sound,
In accordance with the result of the above-mentioned harmony / non-harmony determination means.
Of the sound data representing the above motif
Type determining means for determining the type of non-harmonic sound of the sound
including.   In the embodiment described later, the melody generating unit includes
A means to add various non-harmonic sounds is included.
The motif is separated in relation to the categorized non-harmonic sound adding means.
On the side of analysis, the motif sound determined as a non-harmonic sound
Type determining means for determining the type of non-harmonic sound is provided.
You. Such non-harmonic type determination means and non-harmonic types
Combination with sound adding means makes the generated melody more
It is effective for fine control.   Further, according to the present invention, the pitch (MDi) and the pitch (MRi)
A sequence of melody sound data that contains
Melody input means (1) for inputting a melody,
Enter the chord information indicating the music background of the melody
The input code input means (1) and the melody information are displayed.
For each sound of the melody,
Check whether the sound is a chord sound.
If the sound is determined as a harmony, it is not a chord sound
If so, determine the sound as a non-harmonic sound
Sound determining means (9, Fig. 9) and pitch of the melody information
Removes data from the sequence of sounds determined as non-harmonic
And the pitch sequence of the harmony (LLi),
Dispersion chord forming means (9, FIG. 14) to be formed;
Of pitch movement in the pitch sequence of the distributed chords
Determine motion parameters to determine motion parameters
Means (9, FIG. 18).
A music device with a de-analysis function is provided.   According to this configuration, the dispersed chord as a melody skeleton
And that we can get that pitch train pattern
First, get information about the movement of the pitch
Can be   The above motion parameter determining means is a motion parameter
Neighboring sums in the pitch series of the above-mentioned dispersed chords as meters
Parameter representing the maximum motion formed between voices
And a maximum motion calculating means for calculating the data.   Further, the motion parameter determining means is a motion
Parameters in the pitch row of the dispersion chord
The pitch of each pair ((harmonic pair))
Homology progression parameter calculator for calculating the number of the same chord pairs
Steps may be included.   Further, according to the present invention, the pitch (MDi) and the pitch (MRi)
A sequence of melody sound data that contains
Melody input means (1) for inputting a melody,
Enter the chord information indicating the music background of the melody
The input code input means (1) and the melody information are displayed.
For each sound of the melody,
Checks whether the chord is represented by the chord represented by
If so, determine that sound as a harmony,
Harmonic sound / non-harmonic sound decision to determine the sound as non-harmonic sound
Setting means (FIG. 9 and FIG. 9) and the pitch length sequence of the melody information.
The note length of the sound determined as a non-harmonic tone
By absorbing the pitch determined as a harmony
And the length sequence of the harmony, that is, the rhythm pattern of the dispersed chord
Means for forming a distributed chord rhythm pattern (9, 15
Figure), with a melody analysis function
A music device is provided.   According to this configuration, in the inputted melody,
Removes the sound determined as a vocal sound and changes the melody to only a chord
The length sequence (the length sequence of the dispersed chords)
Obtainable. To form a sequence of dispersed chords,
The sound of each sound determined (presumed to be) as a non-harmonic sound
The long was determined as the next chord (presumed to be)
It is absorbed by the length of the sound. As a result, the resulting dispersed chord
Look at the length sequence representing the original length sequence of the melody
be able to. [Example]   Hereinafter, embodiments of the present invention will be described.   This embodiment is an example in which the principle of the present invention is applied to an automatic composer.
It is. The feature of the embodiment related to the present invention is that
Of the overall functions of the composer,
Function part for extracting the signature parameter (F10 in FIG. 2, ninth
11B, FIG. 14, FIG. 15, and FIG. 18). This departure
Audible / non-harmonic sound deciding means, non-harmonic sound type determiner
The best representation of these for columns is shown in FIG.
(A flowchart of one embodiment of the chord / non-harmonic sound determining means)
G), FIG. 10 (a flowchart of one embodiment of the type determining means)
FIG. 11A, a flow chart of another embodiment of the type determining means.
FIG. 11B (flow chart of still another embodiment of the type determining means).
Chart). Fig. 14 shows the input melody (mochi
Remove the data of sounds determined as non-harmonic tones
To form a pitch sequence of harmony, that is, a pitch sequence of dispersed chords.
5 is a flowchart illustrating one embodiment of a distributed chord forming means.
You. In this flowchart, the pitch sequence of the dispersed chords (chords
Each pitch LLk in the parameter of the type
It is expressed by a number, not a note name. This representation is
It does not depend on the contents of the notes (sound name information) of a specific chord
In a sense, it is an expression independent of the code. Fig. 18
Is the pitch movement in the pitch sequence of the dispersed chord
Motion parameters that determine the motion parameters
It is a flowchart of one aspect of parameter determination means. This
In the flowchart of
The largest motion between adjacent chords
(The maximum value PA of the difference between LLk + 1 and LLk_{1, 2}) And pitch
Are the same number of adjacent harmony pairs (PA_1,6).
This biggest motion PA_{1, 2}Is the “na”
It is one measure of "smoothness." Parameter PA_1,6Is the same
It is called the parameter of sound progression.   Fig. 15 shows the pitch sequence of the input melody (motif).
The length of the sound determined as a non-harmonic tone is next to it
By absorbing the pitch determined as a harmony,
The length sequence of the chords, ie, the rhythm pattern of the dispersed chords (harmony
Distributed chord rhythm pattern that forms the length pattern of the sound)
It is a flowchart showing one aspect of the forming means. this
In the flowchart, the pitch determined as a non-harmonic tone is
Determines which of the sounds before and after the harmony is determined
16 Determine according to the pulse scale weight as shown in Fig.
I have.   The basic part of this automatic composer is the above-mentioned patent application,
Patent application filed by the applicant, titled "Automatic composer", Showa
Filed on April 8, 1987 (Japanese Patent Application No. 62-86571) and 1987
The composer disclosed in the May 20 application (Japanese Patent Application No. 62-212037)
Referred to these patent applications because they belong to
No. In the attached drawings, details of the basic parts are shown in FIGS.
This is shown in Figure 49. <overall structure>   FIG. 1 shows the overall circuit configuration of the automatic composer. In the figure, 1
Is an input device, 2 is a chord structure sound memory, 3 is a chord progression
Memory, 4 is root data memory, 5 is scale weight data
Memory, 6 is motif memory, 7 is parameter B memo
, 8 is the phrase identification data memory, 9 is the CPU, 10 is the work
Memory, 11 is parameter C memory, 12 is learning data memo
, 13 is melody data memory, 14 is monitor, 15 is CR
T and 16 are stave printers, 17 is a tone generator, 18 is a
Command system 19 is an external storage device.   The motif memory 6 is input from the input device 1.
This is where motif (input melody) information is stored.
You. Motif information is a row of pitch and pitch (pitch) data
Be composed. When composing automatically, CPU9 uses this motif information
From the parameters that characterize the motif
Parameter).   The chord progression memory 3 is represented by a column of chord names.
The chord progress information to be displayed is stored. Chord progress information
Indicates that the user sequentially specifies a code from the input device 1.
May be entered, or a rough specification (for example,
CPU9 automatically responds to the chord progression
It may be generated dynamically. Automatic generation of chord progressions
Are, for example, basic code patterns
Code pattern) or allowed code
The logic of the connection is, for example,
A model of the Rukov chain can be used. However, the entire song
Mode is specified directly by the user or
Automatically generated by the application
Not that.   However, the motif code is specified by the user
The entered code is input from the input device 1, which is an important point.
Int. In other words, the chord progression memory 3 has few
At least the chord progression of the melody is stored. Mo
The chief code and the melody code following the motif
For the automatic connection with code or chord progression,
This can be realized by the Rukov chain model.   The chord composition sound memory 2 has various chord composition sounds (core
The pitch data of the current member is stored.
In this case, the contents of each address (code
Code), a specific code on the chord sound memory 2
The storage area of the card configuration sound data is now specified.
ing. The CPU 9 changes the chord at the
Chord progression memory 3 (for example, for each bar)
Forwards the address of the
Calculate the address on the code configuration sound memory 2 and configure the code.
Read each pitch data to be formed.   The root data of the chord is stored in the root data memory 4.
The scale weight data memory 5 stores the scale (note scale).
Shows the degree of effectiveness of each pitch that composes
Weight data, i.e. weighted note scale data
Data is stored (set of note scale data
Memory to store). When composing automatically, use an appropriate method
The scale is selected, and the weight data of this scale is read out.
You. The root data memory 4 stores the read scale weight data.
Used to shift root.   On the other hand, the parameter B memory 7 stores the melody flow.
To control consistency and diversity in
(Parameter B) is stored. Also, higher order
Music identification data memory 8
Is used. The memory 7 and the memory 8 have user
Area, that is, the parameters specified by the user
Read / write areas are prepared. Automatic composition
At this time, the CPU 9 sets the parameter B and the musical style identification data (after decoding).
Data), the above motif feature parameters, music progress section
Parameter C (meaning
Rodi planning information). Parameter C is generated
It has the property of controlling or characterizing the melody.
The generated parameter C is stored in the parameter C memory 11.
Is done.   The work memory 10 contains the process for the CPU 9
Intermediate data (for example, the melody being processed)
Data) is stored.   The melody data memory 13 constitutes a completed song
Melody data is stored.   Output the completed song to the monitor 14 as necessary
be able to. For example, the tone generator 17 and the sound system
You can audition through the system 18. Also, the staff notation
From the linter 16, a copy of the score can be obtained.   The user through the monitor 14 partially corrects the song
You may want to do that. In the present embodiment,
User corrects it via CRT15 and input device 1
Can be requested and modified in an interactive format.
Positive is executed. The corrected data is the learning data memory
Stored as knowledge in 12. For later automatic songs, CPU 9
Uses this knowledge to generate melodies.   The external storage device 19 is a backup copy of the completed song
Or learned knowledge, other copies, or alternatives
It is used as a resource for automatic music programs. <Automatic song function>   Next, refer to FIG. 2 for the overall functions of the automatic composer.
It will be described in the light of the above.   As shown in the figure, the main function is
Motif feature parameter extraction
Output means F10 and the information provided by this means F10 (comprehensive
To control the melody from the PA)
Melody control information that generates (inclusively shown by PC)
Information generation means F20 and the information PC provided by this means F20.
Therefore, a melody that specifically generates a melody (single melody)
There is a generation execution unit F30.   As will become more apparent from the following description, the features of the present invention
Is a motif in the information sent to the means F10
Chord information (chord data) assumed by the user
This chord information is utilized in the included points and in means F10.
Using the harmony characteristic parameters of the motif (number of harmony,
The number of chords, the type of variances, etc.)
You. Chord information is input through the input device 1 in FIG.
Information.   On the other hand, the melody control information generation means F20 is a motif feature
Ability to associate and plan melody based on parameter PA
The melody generation execution means F30 has this plan (PC
And express the melody according to the plan
Has rules to generate. In a broad sense, means F20 and F30
Constitutes a melody generating means. Meanwhile, the inference aspect
Motif feature parameter extraction means F10 and melody
It is basically in the reverse relationship with the generation execution means F30,
Motif feature parameter extraction means F10
(Input melody)
It has inference ability to derive meter PA, while Melody
The melody feature execution means F30 is a melody feature
Inference ability to derive a specific melody from parameter PC
Have. Planning space of melody control information generation means F20
Is vast, and the degree to which PA is reflected on PC is freely and
It can be controlled in various ways.   Parameter B shown in this F20 section
Another data is a data created by the user through the input device 1.
Data. Parameter B used and musical style identification data
If the data changes, the melody control information generation means F20 outputs
The control parameters (PC) also change.   Generally, a song changes as it progresses. But limited
Looking at the time spent, certain features are fixed. Own operation
The music machine takes this point into consideration, and considers it to be fixed.
It has the concept of a unit of "between." For example, motif features
Each function in the parameter extraction means F10 is based on a certain section as a unit.
To extract PA. Similarly, the melody control information generating means
F20 is the melody of the PC value assigned to each section
The melody generation and execution means F30 passes it to the generation and execution means F30.
The melody of the section is generated using the value of the PC of. Mochi
Of course, this means that all PCs
It does not mean that it will change (similarly for PA). PC
Depending on the type, the interval over which the value can be fixed is generally equal
No. Therefore, in principle, different PC types
And assign values to different sections by type.
It is also possible to make it occur. But this
This complicates the process. Therefore, in the embodiment, the type of PC
Far fewer types of segments, preferably all
Sections that can be common to PCs (sections with the greatest common divisor)
Between). Motif feature parameter extraction means F10
The same applies to.   In the configuration of FIGS. 9 to 49, the most simplified
As an example, each function F10, F20, F30 has a common section
Like that. "1 bar" is selected as this section
You. Of course, this automatic composer is limited to this
Instead, each function may have two or more different sections. An example
For example, the interval for the pitch sequence and the interval for the pitch sequence are always
It need not be the same. Means F20 is set to PC
While assigning values, chord progression is measured in measures.
It is also possible to proceed in a section that does not exist.   Please describe the details of main functions F10, F20, F30 in Fig. 2.
U.   F10 section, user input as input information
In addition to the motifs related to
The specified chord information (chord data) is sent. This
The motif code is the motif sound inside section F10
High pattern extraction means F11, especially non-harmonic sound extraction means (harmony
Sound / non-harmonic sound determining means) Used in F11-1A.
That is, the means F11-1A compares the received chord with the constituent tones.
Infer a motif note with the same note name as a harmony,
Motif notes that do not match constituent sounds are inferred to be non-harmonic
You. Therefore, the logic of the non-harmonic sound extraction means F11-1A
The structure is very simple.
If given, the harmony for each sound of the motif /
Non-harmonic sounds can be uniquely distinguished. Shy
However, if the user specifies different chord information,
Make another interpretation. Applicable to motif and melody lines
What chord is a composer, arranger or performer
Is related to the music world that he is trying to express
Cannot be determined. In other words, the present invention
A clue to specify the music you want to express
When the chord designation is changed, the
The melody is reflected in that
You.   In section F10, the non-harmonic classification
F11-1B is the harmony / non-
Analyzes the motif data sequence based on the results of the harmony identification
The type of non-harmonic to which each non-harmonic belongs
Is being determined. That is, harmony versus non-harmonic
Motif at a finer level than voice level
The harmony interpretation of is done. The meaning of this is that
It is also related to the melody generator (F20, F30) of the music machine.
The Rody generation unit includes various non-harmonic feature pattern parameters.
Part that generates data (included in the PC) (in F20
), And sums non-harmonic sounds according to these parameters.
A part to be added before or after a vocal sound or between harmony sounds (F31-2
A) There is a melody line
Control of Harmonic / Non-harmonic Character Patterns
doing.   Distributed chord pattern (pitch) extraction means (dispersed chord type formation
Means: F11-2 removes non-harmonic sound from motif
Pitch pattern, that is, the type of dispersed chord (LLi described later)
Is the part that extracts   On the other hand, motif duration (rhythm) pattern extraction means F12
Is a function that extracts features related to the length sequence of motifs.
You. Feature mini-pattern extraction means F12 that is a component
-1 extracts the characteristic small pitch sequence contained in the motif
Scattered chord (rhythm) pattern extraction means
Pattern formation means) F12-2 is included in the motif
The length of the non-harmonic tone is absorbed by the length of the
Only form a note length sequence.   Of the above factors, F11-1B, F11-2, and F12-2 are non-
It operates in response to the result of the harmony sound extracting means F11-1A. this
These means, in the configuration of FIG.
It is a function to extract data. More generally, harmony
Characteristic parameters in motifs and melody lines
Parameters that characterize the style and style of harmony and non-harmony
Data.   Section within the block of the melody control information generation means F20
Generates parameters depending on the section number indicated by counter F23.
The generated progress-dependent parameter generating means F22 is shown.
You. Some of these parameters are regulated by song progression.
This includes parameters that vary regularly,
The rule variation parameter generating means F22-1 shown by F22-1
Generated by Means for the parameter generated by means F22
Then, the random number generation means F24 operates, and
A controlled random number or variation to the same parameter
You can enter. The above F22 and F23 elements are inside the PC
Contains parameters with the above properties
For non-operational type
(For example, a database of parameters C)
C (melody control information) can be generated
It is shown for the purpose. Actually, in this embodiment,
Parameter C is generated in operation type, and this operation is executed.
The part to be executed is the calculation means indicated by F21. Calculation means F21
Is indicated by the motif feature parameters PA and I1 from the means F10.
Information of the parameter B, the small value indicated by the bar counter F23-1.
Section number, style identification data from style identification data generation means F25
Data as inputs, and these inputs as parameters
Calculate the meter C. The style identification data generation means F25
It has a clause counter F25-1 and uses it to
Information about a specific passage is selected from the separate data I2. easy
Some of the information about the clauses is the type of clause (repeated, expanded
Type) is included. The style identification data generation means F25 is small
Read the bar number indicated by the bar counter F23-1 and read the bar number
Inspecting the positional relationship between the number and the passage, and based on the inspection results
To decode (decode) related musical identification data. Solution
The read musical style identification data is passed to the calculating means F21. easy
The role of the formula identification data generating means F25 is
To give the following hierarchy.   As can be seen from the above description, the melody control information generation
The PC output from the means F20 varies, and some PCs
It is constant for a relatively long time, and one PC is a unit section
Fluctuates at a cycle equivalent to (one bar here)
Some PCs are affected by the music style,
To change to another value, and so on. Parameters
The value of B, the value of musical identification data, used by the calculating means F21
Depending on the type of function, even the same PA
Various PCs will occur.   In other words, the parameter PC is the local stylistic feature of the song.
Sign, a slightly wider range of stylistic features, appearing regularly
Stylistic features, hierarchy in the melody flow
Style, non-harmonic structural in melody flow
Characteristic style, regular or standard in the melody flow
Parameters such as the style of regular changes, the characteristics of the entire song, etc.
It is a data representation.   Melody generation execution means F30 has a pitch as its main element.
It has a sequence generation means F31 and a duration sequence generation means F32. sound
High row generation means F31 is in progress from code progress data I3
Code and melody control information generation means F20 from PC
Or the P whose variation was introduced by the random number generation means F31-4
Dispersion chord generation using C and generating a pitch sequence of dispersion chords
The raw means F31-1, before or after the generated dispersed chord or
In the meantime, according to the plan of the PC and according to the additional rails inside
Non-harmonic sound adding means F31-2 for adding a non-harmonic sound
ing. Each non-harmonic classification adding means F31-2A is a means F31-2.
1 is a configuration example. F31-3 is used pitch control means.
In the above means F31-1 or F31-2,
For the melody node candidates obtained in the process,
Has the ability to control the use of That is, use sound
High control means F31-3 is replaced with note scale generating means F31-3A
Generates weighted data for each pitch of the scale
The sound effect inspection means F31-3B gives the candidate sound based on the weight.
Have high effectiveness tested. Melodyno that passed the inspection
Is sent back to the means F31-1, F31-2, where the formal
Used as a melody note.   The length sequence generating means F32 is optimally combined with the optimal combining means F32-1.
Means F32-2 and feature pattern embedding means F32-3
Then, a note length sequence is generated according to the plan of the PC. Optimal coupling means
In the case of this embodiment, F32-1 and the optimal dividing means F32-2 are distributed.
Used to form chord durations. Optimal coupling means
F32-1 is the initial pitch sequence (for example, the dispersed chord of F10 (Liz
The variance sum extracted by the pattern extraction means F12-2
Sound pattern or its modified pattern)
To the number of notes of the distributed chord to be marked (given by the PC)
Combine note lengths with minimum number of joins. Similarly,
Appropriate division means F32-2 divides initial pitch sequence to distributed chord number PC
Divide the note length by the minimum number of divisions until In addition, both
Means F32-1 and F32-2 are separated by using a pulse scale.
Perform the split and join. Pulse scale is PC (melody meter
Image information) as a kind of image information.
However, in a specific example to be described later, the dividing and connecting
There is a pulse scale inside the rule. On the other hand,
Sign pattern embedding means F32-3 operates according to the related PC
And a characteristic mini-pattern
Put in. In this example, the feature
Two patterns are injected.   The control means F33 is provided for each element of the melody generation execution means F30.
It is used to control activation and data transfer between elements.
You. Melody which is the execution result of melody generation execution means F30
The melody data is stored in the melody data memory 13 (Fig. 1).
It is. <Details>   Figures 3 to 49 show various data and motif features.
Extraction of parameter A, generation of control parameter C, melody
The details of the specific occurrence of di are shown. Paper
For the sake of convenience, a detailed description of the present invention will be omitted except for the features of the present invention.
love. For those skilled in the art, clear descriptions such as flowcharts
Can easily understand the contents to the level that can be implemented.
U.   Here are a few things to help you understand. Fig. 3
FIG. 4 shows a list of main variables, and FIG.
The strike is shown. Each parameter shown in these lists
Symbols in the flow chart from FIG. 9 to FIG. 49
Match the symbols used. Smell these flows
Therefore, the ratio of the pitch data is as shown in FIG.
Continuous pitch on the chromatic scale
Integer value. For the duration, one bar
Is divided into 16 equal lengths. Therefore
The duration of the sixteenth note is “1” and the duration of the eighth note is “2”.
There is (the pitch data shown in the motif data column in FIG. 6)
See MDi and pitch data MRi).   The motif data, chord component sound data,
Data, musical identification data, scale weight data, and
And the data of parameter C in Fig. 7 as input conditions
The melody obtained in FIG. 8 is illustrated in FIG.   The overall process from creating a melody to a motif
First, the motif evaluation, that is, the motif feature parameter
This is the extraction of the meter (PA).
This is the calculation of the control parameter (PC) for the position section (measure).
After the PC is determined, the specific melody is generated.   Process for extracting motif feature parameter A from motif
The relationship is shown in FIGS. 9 to 20.   Here, FIG. 9, FIG. 10, FIG. 11A, and FIG.
Non-harmonic sound extraction and classification related to the features of the present invention
I will explain these in detail. Non-harmonic sound extraction   FIG. 9 illustrates a flowchart of non-harmonic sound extraction.
The basic logic is as follows. That is, Mochi
Even note names that do not match the chords
The tones are non-harmonic. Conversely, the sound of the chord is
Each of them is a logic for generating a harmony sound.   More specifically, in the sections from 9-1 to 9-4,
Pitch sequence of motif data in chief memory 6 {MDi}
To the area {HDi} prepared on the work memory 10
doing. In other words, all pitch data of the motif data
To HDi. Incidentally, in this example,
Each note of the chief is a pair of pitch data and pitch data
expressing. However, there is no concept of pitch in rests.
It can be expressed by putting a special value in the pitch data.
Is showing.   9-5 to 9-15 for the i-th note in array {HDi}
To determine if it is a harmony or non-harmonic
It is identified by collating with. The identification result is
In the case of "harmony", a value specific to HDi (minus 20 in the figure)
I remember by putting. j is the number of the chord component
Is shown. For convenience of explanation, in this example, the code is
Four independent voices or three independent voices, including two voices one octave apart
Only consider the case. Furthermore, adapt to this
Therefore, each chord constituting sound memory 3 (FIG. 1) has
There are four data as constituent sound data. This flow
In, the pitch data of the chord components is indicated by KDj
I have. The subscript j is the j-th chord component sound from the bottom
Is shown.   As shown in 9-7, the motif note of interest is
When resting, it is not a non-harmonic sound, so the next motif note
Move on to check. 9-8 and 9-9 are Mochi, respectively.
Where the note names of the notes
Yes (elimination of the concept of octave numbers). 9-10 is a note name
Checks for matches and mismatches.
Since the motif note is a "harmony", the next note
Move on to check. If there is a mismatch, the next chord sound
J is incremented in 9-12 to proceed with
And repeat steps 9-7 and below. J ≧ 4 is satisfied by 9-11
The motif notes of interest are all
Compared with the constituent sounds, but none of them match
That is, it is the case of "non-harmonic sound". Therefore,
Non-harmony in pitch data HDi of the motif note in 9-13
Substitute the sound identification value. 9-14 No is the motif note
It is the total number. Therefore, when i ≧ No holds,
Whether all notes in the music are non-harmonic
Has been identified. Therefore, the non-harmonic sound extraction process ends.
Complete. Classification of non-harmonic sounds   The classification of non-harmonic sounds is based on the extraction of non-harmonic sounds as described above.
By analyzing the pattern of the motif sound sequence based on the results
Is performed.   There are several possible classification logics. Figure 10 shows the
FIG. 11A is an example of a flow according to the classification logic of FIG.
FIG. 11B illustrates the second and third classification logics, respectively.
FIG.   Description will be made with reference to FIG. In 10-1, the motif note
Initialize the number i to “1”, and then, until HDi ≧ 0, i
Is incremented (10-1, 10-2). HDi ≧ 0
Indicates that the i-th note is a harmony. 10-4 and i
As inferred from incrementing,
Check from the beginning of the motif and go to the first harmony
Non-harmonic sounds are distinguished from others. In the illustrated flow
Non-harmonic HDi exits as minus 20 (Hon, HDi =
-20).   10-5 indicates that the note of interest is a harmony or rest
Check whether the HDi
Since there is no need to change, the processing proceeds to the next note processing. Unsuccessful
At this time, the second note is a non-harmonic sound. Focus on 10-6
The note before and the note after the same note
The height is HDi = -30
(Embroidery sound, 10-8). In other words, pay attention
We conclude that the note that is is an embroidery sound. When not established
Is HDi = -40 (elapsed sound, 10-7)
Run. That is, it is concluded that the sound is a passing sound.   Check if it is non-harmonic and the last note
At 10-9, i = No holds. This non-harmonic sound is an anomaly
And HDi = -60 is executed (10-10). 10 <11, i <N
When o holds, the disjunction for all notes
Since the voice classification has not been completed, i = i + 1 (10-4)
Return to the next note analysis.   The four types of HDi obtained as a result of the processing in FIG.
The name given to the chord identification value, that is,
"Elapsed sound", "embroidery sound" and "miss sound" are just explanations.
This is for your convenience. This is also true for FIGS. 11A and 11B.
The same is true.   The non-harmonic classification shown in Fig. 11A is almost the same as Fig. 10.
It is. The difference is that the processing shown in 11A-6 and 11A-7 is added.
That is the point. The calculation shown in 11A-6, that is,   a₁= (MD_i-1−MDi) × (MDi−MD_{i + 1}) Is the power of the pitch differences formed between three consecutive tones
It is a calculation.   Therefore, a₁When is positive, the three sounds have pitches
This means that it is monotonically changing (ascending or descending). This
11A-7 is used to determine whether or not HDI has been established.
I put Inus 40 and concluded it was a passing sound.   FIG. 11B shows the flow of the non-harmonic classification by the third logic.
It is a chart. Comparison with non-harmonic classification shown in Fig. 10
As can be understood from (b), the knowledge of
Items) have been increased, and the conditions for concluding that
has been edited.   In other words, three non-harmonic sounds (HDi) including before and after
Sound (MD_i-1, MDi, MD_{i + 1}) Is a monotonous pitch change (1
1B-6, 11B-8, 11B-9) and the previous sound
When the pitch difference is a semitone or a whole tone (when 11B-10 holds)
Concludes that non-harmonic sound (HDi) is a passing sound (HDi = -40)
Attach. When the condition of this elapsed sound is not satisfied,
It is concluded that the vocal sounds are impulsive. To elaborate a little more, 11
Embroidery sound conditions shown in B-6, 11B-8 to 11B-11
The condition of the sound that elapses, the condition of the missed sound shown in 11B-12,
Non-harmonic tones that do not correspond to any of
(HDi exits at minus 20).   The description will proceed to FIG. 12 and subsequent figures.   In the rhythm evaluation of FIG. 12, the pitch ratio is 3: 1, 1: 1,
One-to-one, two-to-one-to-one, one-to-two-to-one, one-to-one-to-two minis
How many patterns are included in the motif to be evaluated
Is checked, and the result is used as a counter for each pattern.
HR₁~ HR₆Is set to   The parameter extraction shown in FIG.
This processing is performed after non-harmonic sound extraction and classification, where
Various parameters A (PA) are extracted.   Details of the dispersed chord type forming means 13-1 are shown in FIG.
In this flow, the motif dispersion chord
I want a sound pattern. Sound type data is set in the array {LLi}
Is done. For example, {LLi} = {3, 2, 1, 0, 1}
Sometimes it means: The first “3” is at the beginning of the motif
The harmony that appears is all the harmony contained in the motif
Of which is the third height from the bottom, the next “2”
The second harmony that appears is the total harmony of the motif
And the second height from the bottom, and so on.
However, “0” indicates that the note is a rest
(Refer to the example in FIG. 14).   Details of 13-2 which is a dispersed chord rhythm pattern forming means
Is illustrated in FIG. In the flow of Fig. 15,
The pitch pattern of the harmony
Scattered chord pattern). In other words, mochi
The length of the non-harmonic tones contained in the
You. Fig. 16 shows the pulse scale inherent in the flow of Fig. 15.
And the motif dispersion chord using this pulse scale
I'm getting a pattern. Original pitch pattern of the motif
The basic principle of the conversion from
Follow. That is, the pulse of the current note (non-harmonic)
The weight on the scale and the weight on the pulse scale of the next note
If the next note start position SUM is heavier, the current note
The note length MRi is set to the harmony RH in front of it._NAbsorb the current sound
If the note start position SUMB is heavier, the note length MRi
Harmony RH_{N + 1}Absorb. Motif that is the execution result
The distributed chord (length) pattern is set in the array {RHi}
You.   FIG. 17 is an example of the detailed flow of 13-3 and 13-4.
You. Through the processing of extraction and classification of non-chords,
Is converted to {HDi}, and each HDi (i is the i-th
Indicates the pitch) if the sound is a harmony
If it is a non-harmonic sound, the type of non-harmonic sound
A unique value (identification value) that indicates Each non-sum in Fig. 17
In the extraction flow of the number of voices and harmony, the parameter PAij is initialized
After the conversion (first row), the value of HDi is checked and the condition is satisfied.
Each non-harmonic counter (PA_5,2, PA_{2, 2},
PA_{4, 4}, PA_3,3). Counter PA_{1, 3}Wa
Enter the number of voices.   Fig. 18 shows 13-5 (maximum motion calculation means) and 13-6
An example of the detailed flow of (same sound progress parameter calculating means)
Yes, here the variance chord sound type {LLi} (in the processing of FIG. 14)
By analyzing the obtained pattern)
Meter PA_{1, 2}(Between adjacent chords of a dispersed chord
LLk + 1
Given by the maximum value of the difference between
Meter PA_1,6(For each pair of adjacent chords
Represents the number of pairs having the same height).   FIG. 19 is an example of a detailed flow of 13-8. Liz ahead
The various motifs that can be included in the motif
The number of nirhythm patterns is secured in ｛HRi 、.
In the extraction of characteristic rhythm parameters, which mini pattern
Is the dominant pattern. For convenience,
In the flow of FIG. 19, the number HR of the three-to-one pitch pattern₁When
Number of one-to-one pitch patterns HR_ThreeOnly tested, 3 pairs
PA when there are many pitch patterns of 1_{6, 1}To “3”, so
PA when not available_{6, 1}Is set to “0”.   Fig. 20 shows an example of the detailed flow of 13-7,
Find the minimum pitch in the length sequence {HRi} of_3,3N
Is on.   As can be seen from the above specific description, motif features
The parameter extracting means F10 (FIG. 2) is provided outside the automatic composer.
Evaluate and analyze the motif data sent from the
In Fig. 2, parameters that are generally described by PA are extracted. Extraction
The fruits are used to generate PCs and melodies. An example
For example, the pitch type parameter ｛LLi｝
Is the initial parameter of the melody pitch type.
和 RHi ｛is the melody pitch sequence
Can be used as initial information. Other PAs are used for PC calculations
Here, for example, it is used as a DC component.   The processing executed by the parameter calculating means F21 in FIG.
Shown in Figure 1. Figure 25 generates melody continuously
Flow example (25-8, 25-27, 25-28 in the figure)
The indicated I is a bar number, and the bar of the bar number I is 25-10.
(The melody of (I-th bar) is created)
In 25-9, the processing of 21-4, 21-5, and 21-6 in FIG. 21 is executed.
Is performed. 21-1, 21-2 and 21-3 in FIG. 21 are used.
Input parameter PA, PB, SB
is there. Decoding of style identification data performed in 21-4 of FIG.
Please refer to FIG. 22, FIG. 23 and FIG.
In this example, just before the rust (deployment part),
(SBP₁~ SBP_Three) And hold
Data (SBH).   At 25-10 in Fig. 25, a melody for one measure is specifically generated.
Has formed. Other parts of Fig. 25 are motif data
And melody data transfer, ink of measure number to be generated
And so on.   The specific melody generation can be seen from FIG.
First, generation of a melody (dispersion chord) of only the harmony, then
, And the addition of each non-harmonic tone, and finally a rest and characteristic mini-pattern
Is a selective injection.   FIG. 26 shows the overall flow of generation of the dispersed chord.
Read chord constituent sounds in this flow 26-1, sound
Change floor weight 26-2, calculate optimal number of turns 26-4, code structure
Details of turn 26-5 of the sound generation are shown in Fig. 28;
Shown in FIGS., 30B and 31; FIGS. 32 and 33. Two
When 6-5 is completed, the scale weight data to be used
(Data with weights assigned to each note on the scale), and
Ongoing chords and their inversions, that is, the occurrence of dispersed chords
The pitch of each harmony used is fixed. Scale weight de
The melody notes of the melody
Used to check the validity of a candidate, if valid
Adopted as a melody note in the formula.   Fig. 35 shows the simplified flow from 26-6 to 26-42.
Where the pitch sequence of the distributed chord is determined
It is. Details of the decision from the sound before 26-15 shown in this
Is shown in FIG. Sum of pitches closest to last note of previous bar
Process the voice into the first melody note of this bar
is there.   When going from 26-11 through 26-12 to 26-29, use the sound pattern.
Is within the maintenance range, and during the maintenance range is the same as the previous measure
Alternatively, the opposite sound pattern (26-31) is created. For example, the eighth
The second measure shown in the (a) dispersion chord of the figure is the first measure
The sound type is ascending and (LL₁= 1, LL_Two= 2, LL_Three= 3, LL_Four
= 4), Sound type relation from the PC calculation unit for the second measure
Parameters are the notes that have a range from the first to the fourth
(PC_{1, 4}= 0, PC_{1, 3}= 4) and PC_{1, 13}By sound
Since the inversion of the type is specified, the line below (LL₁= 4, LL_Two=
3, LL_Three= 2, LL_Four= 1).   Measures that do not maintain the sound pattern, such as motif measures
In the case of, sound pattern formation by randomization starting from 26-16
Is executed. Parameter C in 26-16 to 26-28
And these parameters control the randomization.
Work to limit.   The effective pitch test is performed at 26-35 and 26-36.   26-44 is the determination of the duration pattern of the dispersed chord. Generate
The number of harmonies in the measure to be equal to the number of harmonies in the motif
Sometimes, the initial pattern (for example,
Turn) becomes the pitch pattern of the dispersed chord as it is.   The number of chords is greater than the number of chords included in the initial pattern
When the time is small, split or combine by pulse scale.
Is going. For details of 26-44, see FIGS.
See the figure. The basic logic of division is the minimum number of divisions
That there is a pulse scale
Or(Where T₀~ T_FifteenIs the pulse point and T₀Is small
The head of the knot, T₈Indicates the timing of the third beat and is shown below.
Number is the corresponding weight)
(Inherited in the rule) using the current pitch pattern
Cross the heaviest pulse point
When a note is split using its pulse point as the dividing line
It is said. The basic logic of the combination is the target number of notes
Combining with the minimum number of joins
Start with the lightest weight in the current duration sequence
This is a method in which a note being played is combined with a preceding note.   In addition of the bias (Fig. 42), the PC
Eccentric weight PC_{2, 2}, Additional position PC_2,3, Scale weight PC_{2, 1},
Ascending, descending PC_{2, 4}, PC with length limit_{2, 7}Etc., and
A bias is selectively added according to the internal bias logic.
You. The note length is calculated from the note length of the harmony that follows the pitch.
A part is taken to make the pitch length of the excitement.   Elapsed sound addition (Fig. 43) basically consists of harmony and harmony
Since the sound that connects between
To see the conditions related to
Decoding part of TA's plan (mainly the first column in Fig. 43)
Or if there is a sound that can be used as a transition between the harmonies
Search and inspection (mainly in the second column)
Pitch conversion (incorporation into pitch patterns) and pitch of evolved notes
Conversion (incorporation into pitch pattern)
3 rows).   The addition of embroidery sound (Fig. 44) has the same pitch for the front and rear chords
Only performed at that time (see first column). Other
There are various restrictions due to parameter C. For example, PC_{4, 4}
Means “0” means prohibit adding embroidery sound.
If it is "1", it can be added once, and if it is "2", it can be added continuously.
Can be added at any time, and can be added unlimitedly at "3"
Is intended. 44-
9 to 44-14 and 44-15.   In the third column, search for the effective embroidery pitch.
And incorporated into the pitch pattern {MED} (44-24 to 44
−30), receive a part of the pitch from the adjacent chord,
This is incorporated into the pitch pattern {MER}. The second
44-16 to 44-20 in the column indicate the pitch data,
This is an array shift in preparation for incorporating into a high pattern.   Similarly, the parameter C is also solved for the addition of a miss sound (Fig. 45).
Part to be read, part to exclude or inspect illegal addition of missed sounds
(E.g. 45-2, 45-3 and 45-4 to 45-16). 45-17
At the time of entry, it was decided that there was no problem with adding
You. 45-17 to 45-23 are distinctive processes shown at the bottom right of the flow
Is for In the rightmost column, between the previous sound and the next sound
In the pitch pattern
It is crowded. In addition, the number of notes increases in the case of a miss sound addition.
Absent. In other words, adding an anomaly changes the pitch of the last note of a bar.
It is.   Rest addition (Fig. 46) is added at the end of the last bar of the passage, etc.
It is often caught.   Characteristic rhythm generation (Fig. 47) uses characteristic mini-patterns.
According to the value of parameter C,
It is about to be incorporated. In this example, the feature rhythm generation is
It is the final process of generating the ody
Does not change the length pattern
Feature mini-panel within the range permitted by the PC.
Turn, here a pattern with a pitch ratio of 3 to 1
Built into the application. 47-9 to 47-12 are consecutive 2
Of three notes that can be converted to a 3: 1 pattern
We are checking to see if there are any. Inspection
In the case, the conversion is executed in the third column.   Parameter change by the correction learning in Fig. 48 and the learning in Fig. 49
The user can select the user's favorite parameter C
Related to the learning function used in individual measures
It is. In the correction learning (Fig. 48), after the song is completed,
Indirect measure parameters that the user does not like
Of the bar for which the learning function is specified
Enter the type and value of the specified parameter in the learning memory 12.
You. Parameter change by learning, as shown in Figure 13,
Performed in parameter calculation and generated internally by the composer
The parameter C selected by the user is
Used preferentially. <Summary and application to melody analyzer>   From what has been described in detail above,
The features of the moving composer are clear. In particular, motif rye
Model to extract its harmony feature parameters from the
Chord information intended by the user for the chief
The use of is notable. This is done
In the example, the function of non-chord extraction and classification of motifs is 2
Parts (harmonic / non-harmonic determining means and types of non-harmonics)
Decision means). That is,
In general, perform harmony analysis of motifs and melody lines
For analysts (typically those with music experience)
To distinguish between vocal sounds and harmony sounds, and what sounds are
Is not always distinct from non-harmonic tones
Recall a wide variety of melody lines
Rather, they can be said to be inseparable. here
In the present invention, the use of code information
More direct and rational, motif or melody
Of non-harmonic tones. Also another aspect
According to this invention, the present invention applies to a certain melody line.
Harmony or chords are related to the artist's musical space
In the idea that it is not generally determined in general
Rooted. Allows user to freely specify code
This means that in the area of application to automatic composers,
Allow the user to approach the musical expression (song) intended and expected
Make it easier. In other words, expanding the selection range for the user
It has greatly contributed.   The above embodiment applies the present invention to the field of automatic music composers.
Automatic harmony evaluation of the melody line
It is clear that the method can be applied to a melody analyzer
is there. Through the use of a melody analyzer, the learner
You will be able to effectively learn the typology of the line. Moreover,
The technology to code the melody also in a relatively short time
It is expected to learn. [The invention's effect]   As described in detail above, the present invention (especially claims)
The automatic composer according to the invention described in paragraphs 1 and 2) is
To separate each sound of the music into a harmonious sound and a non-harmonic sound.
Code information from the
The melody generation means of the composer determines the harmony / non-harmony
Based on the motif feature parameter that depends on the result of the means decision
Melody is generated. Therefore, the user
Melody depends on which code
The line changes. In other words, this automatic composer
Providing users with some good melody line candidates
can do. In particular, you can freely add chords
Gives the user an environment where they can create very effective songs
I can.   Claims (Claims 4 and 5 are dependent claims)
According to the music device with melody analysis function
From the pitch series of the melody, the dispersion sum, which can be called its prototype
It is possible to obtain the pattern of the pitch sequence of the
It is related to the pitch movement in the pitch sequence of the scattered chord.
Analysis information can be obtained. Claim 6
According to the music device with melody analysis function described in the section,
From the melody sequence, the variance
A chord rhythm pattern can be obtained.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is an overall configuration diagram of an automatic composer according to an embodiment of the present invention, FIG. 2 is a functional block diagram of the automatic composer, and FIG. FIG. 4 is a diagram showing a list of variables, FIG. 4 is a diagram showing a list of parameters C used in the automatic composer,
FIG. 5 is a diagram showing an example of pitch data, FIG. 6 is a diagram showing an example of input data for explaining the operation, FIG. 7 is a diagram showing an example of the value of the parameter C, and FIG. FIG. 9 is a flowchart showing a melody generation result for the data of FIG. 7 for each process, FIG. 9 is a flowchart of non-harmonic sound extraction, FIG. 10 is a flowchart of non-harmonic sound classification, and FIG. FIG. 11B is a flowchart of non-harmonic sound classification using still another logic, FIG. 12 is a flowchart of rhythm evaluation, FIG. 13 is a schematic flowchart of parameter extraction, and FIG. 14 is a flowchart of harmony type. A flowchart for extracting parameters, FIG. 15 is a flowchart for extracting a pitch pattern of a harmony, FIG. 16 is a diagram for explaining a pulse scale used in the flow of FIG. 15, and FIG. Figure Is a flowchart for extracting the number of each non-harmonic sound and the number of harmonic sounds, FIG. 18 shows smoothness,
Flowchart for extracting parameters of homophony progression, 19th
Fig. 20 is a flowchart for extracting characteristic rhythm parameters. Fig. 20 is a flowchart for extracting the minimum pitch.
FIG. 21 is a schematic flowchart of the parameter C calculation, FIG. 22 is a diagram showing the data format of the musical expression identification data memory, FIG. 23 is a flowchart of reading and decoding the musical expression data, FIG. FIG. 25 is a flow chart of melody generation, FIG. 26 is a flow chart of dispersion chord generation, FIG. 27 is a diagram showing an example of data of a chord constituent memory, a chord progression memory, a root note memory, and FIG. 28 is a chord constituent sound 29 is a diagram showing an example of scale weight data, FIG. 30A is a flowchart of scale weight change (1), FIG. 30B is a flowchart of scale weight data read, and FIG. 31 is a flowchart of scale weight data readout. flow chart of the weight change of the scale (2), the flow chart of FIG. 32 calculates the optimum turn number, the flowchart of turning in FIG. 33 chord member, FIG. 34 MED ₁ (measure A flowchart for determining the initial sound from the previous sound, FIG. 35 is a simplified flowchart of the main part of generating a pitch sequence of a dispersed chord, FIG. 36 is a flowchart for determining a pitch sequence of a distributed chord, and FIG. 37 is a sound. FIG. 38 is a flowchart of the optimal length dividing process, FIG. 39 is a detailed flowchart of the check in FIG. 38, FIG. 40 is a detailed flowchart of the shift in FIG. 38, FIG. 38 is a detailed flowchart of the execution in FIG. 38, FIG. 42 is a flowchart of adding a zigzag sound, FIG. 43 is a flowchart of adding an elapsed sound, FIG. 44 is a flowchart of adding an embroidery sound, FIG. FIG. 46 is a flowchart for adding a rest (breath), FIG. 47 is a flowchart for generating a characteristic rhythm, and FIG.
FIG. 48 is a flowchart of correction learning, and FIG. 49 is a flowchart of parameter change by learning. DESCRIPTION OF SYMBOLS 1 ... Input device, 2 ... Chord structure sound memory, 3 ... Chord progression memory, 9 ... CPU, 13 ... Melody data memory, F10 ... Motif feature parameter extraction means, F20 ...
... Melody control information generating means (control parameter generating means), F30... Melody generation executing means, F11-1A... Non-harmonic sound extracting means, F11-1B.

Claims (1)

(57) [Claims] A motif input means for inputting a sequence of sound data including pitch and duration information as a motif; a motif feature parameter extracting means for extracting a feature parameter characterizing the input motif; a music background of a melody to be generated A chord progression storage means for storing at least chord progression information indicating progress; and a melody generating means for generating a melody based on the chord progression information and the extracted characteristic parameters. Code input means for inputting a chord representing the musical background of the motif; and the motif characteristic parameter extracting means, for each sound of the input motif, a note name of the sound (hereinafter referred to as a motif sound name) and the chord input. Means of the chord input from the Includes a harmony / non-harmonic tone determining means for determining the tone as a harmony when the motif name is the same as the motif tone, and determining the tone as a non-harmonic tone when not included. An automatic composer characterized in that a feature parameter is extracted based on a decision result of a harmony / non-harmony deciding means. 2. In the automatic music composer according to claim 1,
The motif feature parameter extracting means further performs pattern analysis on a row of sound data representing the motif for each sound determined as a non-harmonic voice according to the determination result of the harmony / non-harmonic sound determining means. An automatic composer comprising type determining means for determining the type of non-harmonic sound. 3. Melody input means for inputting a sequence of melody sound data including pitch and duration information as melody information; code input means for inputting chord information representing a music background of the melody represented by the melody information; For each sound of the melody to represent, check whether the sound is a chord constituent sound of the above chord information,
A harmony / non-harmonic sound determining means for determining the sound as a chord sound if the sound is a chord sound, and a non-harmonic sound if the sound is not a chord sound; Dispersion chord forming means (9, 14th) for removing the data of the sound determined as a non-harmonic tone and forming a pitch sequence of a chord, that is, a pitch sequence of a dispersion chord.
FIG. 5) and a motion parameter determining means for determining a motion parameter related to a pitch movement in a pitch sequence of the formed dispersed chords. 4. 4. The automatic composer according to claim 3, wherein said motion parameter determining means includes, as said motion parameter, a parameter representing a maximum motion formed between adjacent chords in the pitch sequence of said dispersed chords. A music device with a melody analysis function, comprising: a maximum motion calculating means for calculating. 5. 4. The automatic composer according to claim 3, wherein said motion parameter determining means includes, as the motion parameter, a sum of the same pitch of each pair of adjacent harmony in the pitch sequence of the dispersion chord. A music device with a melody analysis function, comprising: a same-sound-progression-parameter calculating means for calculating the number of voice sound pairs. 6. Melody input means for inputting a sequence of melody sound data including pitch and duration information as melody information; code input means for inputting chord information representing a music background of the melody represented by the melody information; For each sound of the melody to be represented, it is checked whether or not the sound is a constituent sound of the chord represented by the chord information. If the constituent sound is a constituent sound, the sound is determined as a harmonic sound. Harmony / non-harmonic sound determination means, which determines the non-harmonic sound in the length sequence of the melody information, and absorbs the sound length of the sound determined as the non-harmonic sound to the adjacent harmony sound determined as the harmony sound. A distributed chord rhythm pattern forming means for forming a cadence length pattern of a harmony, that is, a distributed chord rhythm pattern. Music equipment.