JPH0511763A - Note property judging deivce and automatic accompaniment device - Google Patents

Abstract

PURPOSE:To judge a note property by a method suitable for the musical condition of every part in code progression in a device judging a note property from the code progression. CONSTITUTION:A note property judging device 10 has a first judging part 30, judging a key note in a new code section of code progression with a present key note as grounds, and a second judging part 40, judging a key note in a new code section without making a present key note grounds. The first judging part is effectively applied to a musical condition, in which the function of a code is specifiable from a key note before the code. The second judging part is effectively applied to a musical condition, in which the function of a code is difficult to be specified from a key note preceding the code, (including a condition having no preceding key note and an unknown condition). A note property judging device 10 can be applied to an automatic accompaniment device.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a music apparatus, and more particularly to a tonality determining apparatus for determining the tonality of chord progressions and an automatic accompaniment apparatus for performing automatic accompaniment.

[0002]

2. Description of the Related Art Electronic musical instruments having an automatic accompaniment function are already known. In this type of electronic musical instrument, chord progressions are input by sequentially specifying chords as a combination of key codes (note numbers) from a performance input device such as a keyboard. Inside the electronic musical instrument, there is provided a chord constituent note memory for storing the constituent notes of each chord of the chord set so as to indicate the pitch from the root (chord root note). By the chord root / type discrimination function, one of the input multiple key chords (note numbers) is assumed to be the chord root, and the pitch from the rest of the chord roots is obtained and can be compared with the chord constituent note memory. Generate the specified chord component sound data. The chord root / type discrimination function searches the chord constituent sound memory for the generated designated chord constituent sound data. The type and root of the chord are identified by finding a match between the designated chord constituent note data and the chord constituent note data for a given chord type in the chord constituent note memory. In this way, a chord progression in which each chord is represented by a root and a type is obtained. Further, an accompaniment pattern memory for storing accompaniment patterns is provided inside the electronic musical instrument. The accompaniment pattern is composed of horizontal (time) information and vertical (pitch) information of the accompaniment line. By the accompaniment decoding function, the stored vertical information of the accompaniment pattern (vertical data of each accompaniment sound) is converted into data indicating a specific pitch (pitch) according to the identified chord type and route.

This type of device lacks the ability to evaluate the function of each chord in chord progression. In general, it is not possible to specify the tonality, which is a set of pitch classes (pitch types) of sounds that can be used in the chord section, only from the chord type and root information in music. For example, chord C MAJOR (chord constituent notes C, E,
Speaking of G), if this chord has the function of I (tonic), as a set of suitable pitch classes, C, D, E, F, G, A, B (key C
Ionian scale) is desirable. This code C M
AJOR

[External character 1] If it has the function of (dominant), a suitable pitch class set is C, D, E, F, G, A, B ♭ (key F's mixolydian scale). Code C
MAJOR

[External character 2] If it has the function of (subdominant), C,
D, E, F #, G, A and B (key G's Lydian scale) are preferred as pitch class sets. That is, in terms of music, a preferable pitch class set (tonality) cannot be determined unless the type, route and function of the chord are specified. Nevertheless, the apparatus described above uses only the root and type of chord constituent note memory to decode the accompaniment pattern to determine the pitch of the accompaniment note. Therefore, there is a possibility that sounds with a pitch (pitch) unfavorable to automatic accompaniment may be mixed and become unnatural. In order to avoid this, the vertical information of the accompaniment pattern to be stored may be limited so as to include only the chord-constituting sounds, but this makes the accompaniment simple.

Japanese Patent Application No. Sho 62-33359 relating to the present applicant
No. 5 shows a tonality determination device that determines the tonality of chord progression and an automatic accompaniment device that uses the result of the tonality determination device to perform accompaniment. This tonality judging device judges the tonality of each chord by evaluating the tonality distance between chords in chord progression according to an evaluation algorithm. In principle, this tonality determination device can only make relatively low reliability tonality determination.

Another example of the tonality determining device and the automatic accompaniment device is shown in Japanese Patent Laid-Open No. 2-29787. This tonality judging device searches a given chord progression for a part having a particular chord type pattern and a particular root difference pattern to judge a tonality. However, in this tonality judging device, the key already judged for the preceding part of the chord progression is never used or referred to for the key judgment for the following chord. That is, this tonality determination device lacks the function of predicting the tonality of the following chord using the leading key as a clue. Therefore, the detection of modulation (key change) tends to be delayed. Due to such a delay, the key is erroneously determined in a part of the chord progression. An automatic accompaniment apparatus, which is an application of such a tonality determination apparatus, cannot perform a sufficiently satisfactory accompaniment for the reasons already described.

Recently, the applicant of the present application has proposed a tonality determination device for determining the key of the following code by using the key of the preceding key as a clue (Japanese Patent Application No. 3-68922, filed March 8, 1991). ). This tonality determination device has an identical key maintenance confirmation function that confirms whether the succeeding chord maintains the preceding key, and a transposition detection function that determines whether the succeeding chord is a key related to the preceding key. Have Therefore, this tonality determination device can fairly accurately determine the tonality of each chord in the chord progression, and in particular can detect the transposition to the intimate key without delay. However, it cannot handle all the various situations in real music. For example, in real music, a succeeding key may be a key unrelated to the preceding key. In some cases, there is no preceding key such as in the first part of the song, or the preceding key itself is not clear. In these areas of musical phenomenon, the tonality analysis ability of the tonality determination device is not sufficient, and there is room for improvement.

[0007]

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide various musical situations in chord progression (a situation in which a key in a preceding chord section is a clue of a key in a following chord section, a preceding section). Tonality judgment with improved tonality judgment ability of chord progression by dealing with situations in which the key does not become a clue in the section to follow, situations in which there is no preceding section, situations in which the key is unknown, etc.) It is to provide a device. A further object of the present invention is to provide an automatic accompaniment apparatus which utilizes such a tonality judgment apparatus to perform an accompaniment having a desired tonality.

[0008]

According to the present invention, chord progression assigning means for imparting chord progression in which each chord is represented by root and type, and data representing the current keynote relating to the current tonality are provided. Current keynote storage means for storing, first tonality determining means for determining the tonality of a new chord from the chord progression based on the current keynote, and tonality of a new chord from the chord progression And a second tonality determining means for determining by analyzing the chord pattern formed between the new chord and the chord preceding the new chord without being based on the current keynote. A characteristic tonality determination device is provided.

In the case of this configuration, the first tonality determining means is the current keynote (keynote preceded by a new code).
To determine the tonality of a new chord. on the other hand,
The second tonality determining means determines the tonality of the new chord by analyzing the chord pattern formed by the new chord and the chord preceding the new chord without using the current keynote as a clue. Therefore, this tonality determination device is not limited to a musical situation that can be a clue of the tonality in the preceding keynote or the chord section in which the key follows, but also in other musical situations such as a chord followed by the preceding keynote. It is possible to cope with a musical situation that cannot be a clue of tonality in a section, thereby improving tonality judgment ability.

In one aspect, the first tonality determining means is operable when the current keynote is specified, while the second tonality determining means is operable when the current keynote is not specified (current keynote). This may include the case where there is no note)
Alternatively, it can be operated when the first tonality judgment means cannot specify the tonality of the new chord (because of the musical situation in which the current keynote is not a clue).

In a preferred configuration example, the first tonality determining means stores the same keynote keeping code table storing means for storing a set of chords for keeping the keynote the same, and from one keynote to another keynote. Analyzing the chord progression using a transposition code sequence table storage means for storing a set of transposition code sequences indicating transposition, the current keynote, the same keynote maintenance code table storage means, and the transposition code sequence table storage means. Based on the first function name / keynote determination means for determining the function name and keynote of the new code, and the function name and keynote determined by the first function name / keynote determination means. The first key that generates tonality data that defines the pitch class set that can be used in the new chord section Data generating means, and the second tonality determining means includes key establishment code sequence table storage means for storing a set of key establishment code sequences for establishing a keynote, and the key establishment code sequence table storage means. A second function name / keynote determination means for determining the function name and keynote of the new chord by analyzing the chord progression, and a function name determined by the second function name / keynote determination means. Second tonality data generating means for generating tonality data defining a pitch class set usable in the new chord section based on the key note.

[0012] Preferably, the first function name / keynote determination means stores the same keynote maintenance code table storage means to determine whether the new code has a function of maintaining the current keynote. It is determined whether the key-maintaining search means for searching and the chord pattern formed by the new chord and the chord preceding it have the function of suggesting transposition from the current keynote to the related keynote related thereto. Therefore, the second function name / keynote determination means includes a modulation search means for searching the modulation code sequence table storage means, and the second function name / keynote determination means determines a code pattern formed by the new code and a code preceding it. In order to determine whether or not it has the function of establishing a particular keynote, the above key establishment code sequence A key establishment determination means for searching the table storage means, and the tonality determination device further includes the current key according to the determined relationship keynote when the relationship keynote is determined by the first tonality determination means. The first keynote updating means for updating the note storage means updates the current keynote storage means by the determined specific keynote when the specific keynote is determined by the second tonality determination means. Second key note updating means.

Further, according to the present invention, there is provided an automatic accompaniment apparatus which utilizes the tonality determining apparatus of the present invention to perform an accompaniment having a desired tonality. That is, the automatic accompaniment apparatus of the present invention has a chord progression giving means for giving a chord progression in which each chord is represented by a root and a type, and a current keynote storage means for storing data representing a current keynote relating to the current tonality. A first tonality determining means for determining the tonality of a new chord from the chord progression based on the current keynote, and a tonality of a new chord from the chord progression,
Second tonality determining means for determining by analyzing a chord pattern formed between a new chord and a chord preceding the new chord, not based on the current keynote, and the first tonality determination First accompaniment forming means, which operates when the tonality of the new chord is specified by the means, and forms an accompaniment according to the specified tonality in the section of the new chord, and the second tonality determination means And a second accompaniment forming means that operates when the tonality of the new chord is specified, and that forms an accompaniment according to the specified tonality in the section of the new chord.

In one configuration example, each of the first tonality determining means and the second tonality determining means uses the function name and keynote of the new chord as the tonality information which is the determination result, and the first tonality determining means. An accompaniment pattern generating means for generating an accompaniment pattern in accordance with the given function name, and means for providing the accompaniment forming means and the second accompaniment forming means respectively. And an accompaniment pitch generation means for transposing each pitch of the accompaniment pattern from the accompaniment pattern generation means in accordance with a given key note to generate an accompaniment pitch.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in more detail below with reference to the drawings. 1 (A) to 1 (E) show an automatic accompaniment apparatus incorporating the features of the present invention. The automatic accompaniment device is roughly divided into a tonality determination device and an accompaniment forming device. Figure 1
A tonality determination device is shown in FIG.

The purpose of the tonality determination device 10 is to set (set) a pitch class (pitch type) that can be used in each chord section (music time) based on the chord progression in which each chord is represented by a route and a type. Tonality data (TO
NALITY data). The tonality determination device 10 has a chord progression input device 2 as a main element.
0, a key / function progress extraction unit (consisting of 30, 40, 50 to 52), a chord progress knowledge database 60, and a tonality data generation unit 70. The chord progression input device 20
Each chord is given a chord progression defined by a root and a type. The key / function progress extraction unit is a chord progress input device 2
The chord progression knowledge database 60 shows keynotes and chord functions suggested by each chord in the chord progression from 0.
The keynote progression and the function progression that match the chord progression are extracted based on the music knowledge stored in. The tonality data generation unit 70 responds to the keynote progress and the function progress from the key / function progress extraction unit and generates tonality data representing a pitch class set usable in each chord section of the chord progression.

According to the features of this embodiment, the key / function progress extraction unit uses the keynote of the preceding section as a clue to determine the keynote and function of the following chord and the first judgment unit 30 and the preceding keynote. It includes a second determination unit 40 that analyzes a chord pattern formed by a succeeding chord and a preceding chord without using a clue to determine the function and keynote of the succeeding chord. In FIG. 1A, the preceding keynote is called a "current keynote" and the following chord is called a "new chord" from the chord progression. Hereinafter, description will be made using these terms.

Data representing the current keynote is stored in the current keynote memory 51. The contents of the current keynote memory 51 are stored in the keynote updating unit 5 every time a new keynote is detected by the first judging unit 30 or the third judging unit 41.
Updated via 2.

The first determination unit 30 operates when the current keynote is known (determined), and the current keynote memory 5
Keynote from 1 and chord progression knowledge database 60
The keynote and function of the new chord from the chord progression input device 20 are determined based on the music knowledge from.

On the other hand, the second judging unit 40 determines whether the current keynote is unknown (including the case where it does not exist) or the first judging unit 3
It operates when the judgment of 0 fails (when a particular keynote or function cannot be judged), and the chord progression input device 2
The chord pattern formed by the new chord from 0 and the chord preceding it is analyzed according to the musical knowledge of the chord progression knowledge database 60 to determine the key note and function of the new chord.

A key hold flag 50 is provided to store information as to whether the current keynote is fixed or unfixed (hold). The key hold flag 50 is initialized to a reset state (indicating that there is no current keynote) at the start of analysis of chord progression, and is set when the specific keynote is detected by the second determination unit 40, and the current keynote is set. Indicates that it has been confirmed. Even after the setting state (key confirmation state) is once entered, when neither the first determination unit 30 nor the second determination unit 40 can identify the keynote of the new chord that is the subsequent chord in the chord progression, the key is pressed again. It indicates that the current keynote is unknown because it is reset to the hold state.

The first judging unit 30 judges whether the new chord from the chord progression input device 20 maintains the current keynote according to the musical knowledge stored in the same keynote keeping chord table 61 of the chord progression knowledge database 60. It has the ability to judge the key maintenance. Here, the same keynote keeping code table 61 stores a set of chords for keeping the same keynote. In order to save the storage capacity, each code entry of the same keynote maintenance code table 61 is not the expression form by the combination of the keynote, the chord root and the chord type, but the form of the function name, that is, the chord function and chord type It is preferable to take the form of a combination of In this case, the key maintenance determination means in the first determination unit 30 can be configured by the function generation unit 31 and the key maintenance search unit 32 as illustrated. The function generator 31 evaluates the new chord from the chord progression input device using the current keynote in the current keynote memory 51 to generate the function of the new chord. The key keeping search unit 32 searches the same keynote keeping chord table 61 for a chord entry having a new chord function name consisting of the generated function and the new chord type (given from the chord progression input device 20). . If the search finds a code entry that matches the function name of the new chord (33), the assumption in the function generator 31, that is, the assumption that the current keynote is regarded as the keynote in the new chord section was correct. become. If this is the case, the function name (type and function) of the new chord and the keynote are passed to the tonality data generator 70, where tonality data representing the pitch class set that can be used in the new chord section is generated. To be done.

Further, when the same keynote keeping code table does not include the code corresponding to the new code having the function evaluated by the current keynote, the first judging section 30 determines that
A modulation determination means for checking the possibility of modulation is provided. The transposition determining means transposes the chord pattern formed by the new chord and the chord preceding it from the current keynote to the related keynote according to the musical knowledge stored in the transposition chord sequence table of the chord progression knowledge database 60. Determine whether to suggest. For this purpose, the transposed chord sequence table 62 stores a set of chord sequences that suggest transposition to the relevant keynote. Each transposed code sequence entry in table 62 can be represented in various formats. In one form, each transposed chord sequence entry includes a pivot chord and a post transposed chord that follows the pivot chord,
The pivot code is represented by the function name before modulation and the function name after modulation on the related keynote, and the code after modulation is represented by the function name after modulation. In this case, the modulation determination means is the related keynote generation unit 34 shown in FIG.
It can be configured by the function generation unit 35 and the modulation search unit 36.
The related keynote generation unit 34 generates a keynote (related keynote) related to the current keynote in the current keynote memory 51. Preferably, the related keynote generator 3
4 generates a plurality of related keynotes. Function generator 35
Receives the related keynote and the new code, and generates the function of the new code evaluated in each related keynote. Furthermore, the function generator 35 is a code preceding the new code (previous code).
Generates the function evaluated by the current keynote and the function evaluated by the previous code with each related keynote. Function generator 35
These functions generated in step (4) are input to the transposition search unit 36 together with the type of the new chord, the type of the immediately preceding chord, and the data of each related keynote. The transposition search unit 36 creates a transposition code sequence entry that corresponds to the input data for each related keynote (the function name of the immediately preceding code evaluated by the current keynote and the function name of the new code evaluated by the current keynote). The table 62 is searched.
If the transposing code sequence table 62 includes such a transposing code sequence for a certain related keynote, (37) the transposition from the current keynote to the related keynote is performed by the code pattern consisting of the immediately preceding code and the new code. It has been suggested that the function generator 3
The function viewed from the relational keynote of the new chord generated in step 5 represents the correct function of the new chord.
If this is the case, the related keynote, the type of the new chord, and the function of the new chord evaluated by the related keynote are passed to the tonality data generation unit 70, where the pitch class set that can be used in the new chord section is set. The tonality data that it represents is generated. The related keynote is also passed to the keynote updating unit 52, and the keynote updating unit 52 updates the current keynote memory 51 to the related keynote.

When the keynote and the function for the new chord have failed in the first judging section 30 (the key matching is the same keynote for the entries that match the respective input data as searched by the key keeping and searching section 32 and the transposing and searching section 37). When it is determined that the code is not in the maintenance code table 61 or the modulation code sequence table), the task of key / function determination for the new code is handed over to the second determination unit 40. Further, the second determination unit 40 takes charge of the task of key / function determination for the new code even when the current keynote is unknown.

The second judging section 40 forms the new chord and the preceding chord according to the musical knowledge stored in the key establishment chord sequence table 63 of the chord progression knowledge database 60 without depending on the information of the current keynote. Determine whether the chord pattern played suggests the establishment of a particular keynote. For this purpose, the key establishment code sequence table 63 stores a set of code sequences (key establishment code sequences) that establish keynotes or keys. Each key establishment code sequence entered in the table 63 has the following expression format, for example. The last code in the code sequence is expressed by the function name (function and type) evaluated in the keynote established by the code sequence, and the code preceding the last code is the type of the code and the code immediately after the root of the code. It is expressed by the pitch from the route (route difference). When the key establishment code sequence table 63 of this expression format is adopted, the second determination unit 40 can be configured by the route difference calculation unit 41 and the key establishment search unit 42. The root difference calculation unit 41 calculates a root difference between adjacent chords in a chord pattern to be analyzed from the chord progression input device 20, that is, a chord pattern consisting of a new chord and one or more chords preceding it. As a result, the analysis code pattern in which the new code is the last code of the pattern is re-expressed by the type pattern and the route difference pattern. The key establishment search unit 42 searches the key establishment code sequence table 63 for a key establishment code sequence having such an analysis code pattern. An entry for such a key establishment code sequence is shown in Table 6.
3 (43), the analysis code pattern ending with the new chord establishes the keynote, and the type of keynote established is the function of the last chord given from the key establishment chord sequence found. And the route of the new chord given from the chord progression input device 20. That is, the keynote whose function value is the pitch of the route from the keynote is the keynote established by the analysis code pattern.

When the function of the new chord and the keynote can be specified by the successful search of the key establishment search section 42, the second determination section 40 sets the key hold flag 50 to indicate that the key has been established. Further, the second determination unit 40 updates the string keynote memory 51 to the newly confirmed keynote via the keynote updating unit 52. Furthermore, the second determination unit 40
The determined new chord function and keynote are passed to the tonality data generation unit 70 together with the new chord type information.
Here, the pitch class set that can be used in the new chord section is determined.

On the other hand, when the function of the new chord and the key note cannot be specified due to the failure of the key establishment search section 42, the second determination section 40 indicates that the key has become unknown, and therefore the key hold flag 50 is set. To reset. In this case, the tonality data generation unit 70 generates tonality data that defines the pitch class set that can be used in the new chord section by the method described below.

In accordance with the features of this embodiment, the tonality data generating section 70 (including the first determining section 30 and the second determining section 40).
When the function / keynote of the new chord can be identified by the key / function determination means, the first method (by narrow tonality interpretation) generates tonality data that defines the pitch class set that can be used in the new chord section. However, if the function and keynote of the new chord cannot be specified, the pitch class set that can be used in the section of the new chord is defined by the second method different from the first method (by wide tonality interpretation). Generate data.

The tonality data generator 70 reads the state of the key hold flag 50 in order to know whether or not the function of the new code and the keynote can be specified by the judgment of the key / function judging means. In the set state (key hold release state), the function and the key note are specified, and in the reset state (key hold state), they are not specified. In the “key hold release state”, the tonality data generation unit 70 causes the first determination unit 30 to
Alternatively, the pitch class set that can be used in the new chord section is determined based on the new chord function specified by the second determination unit 40, the keynote, and the type of the new chord given from the chord progression input device 20. Instead of the combination of the function of the new chord, the keynote and the type, the usable pitch class set may be determined based on the equivalent combination. As a generalization, the tonality data generator 70 uses a pitch class set that can be used by combining at least two of the new chord function, the root and the keynote (for example, the new chord function and the root) and the new chord type. Can be determined. Such a pitch class set is suitable for the above combinations,
The melody using it works to clarify the tonality. In order to obtain such a pitch class set, it is preferable to provide a function name / scale conversion table 71 in the tonality data generation unit 70 and convert the function name (combination of function and type) of the new chord into a scale. The usable pitch class set is defined by the combination of the scale converted from the function name and the keynote (when the fundamental note which is the starting note of the scale in the table 71 is defined as having a predetermined pitch from the keynote). . Alternatively, the usable pitch class set is defined by the combination of the converted scale and the root of the new chord (when the fundamental of the scale = the chord root).

On the other hand, when the key hold flag 50 is in the "key hold state", that is, when neither of the judgment units 30 and 40 can specify the function of the new chord and the key note, the tonality data generation unit 70 A wide tonality interpretation based on the type and route of the new chord (with the possibility of unspecified new chord functions or possible multiple functions of the new chord) defines the pitch class set that can be used in the new chord section.
This pitch class set serves to allow multiple tonality possibilities when used in a melody. Usually, this pitch class set is a part of the pitch class set that can be used in the section of the new chord when the function is specified. Preferably, the pitch class set that can be used when the function cannot be specified is defined by the pitch class set that can be commonly used for all of the plurality of functions that the new chord may take.

A method of determining a pitch class set that can be used by this method is illustrated in FIG. 1 (F). According to the method of FIG. 1 (F) (common PC method), a plurality of possible functions are assumed from the chord type, and the usable scale (pitch class set) is set when the chord type has each function. The pitch class common to these scales is determined, and this is used as the pitch class set that can be used when the function cannot be specified. For example, if the root of the new chord is C and the type is Major, the functions that the new chord can take from the type "Major"

[External character 3] Or

[External character 2]

[External character 1]. The function of the code C Major

Pitch class set (PC
S) is C, D, E, F, G, A, B, and functions

PCS for [gaiji 2] C, D, E, F #, G, A, B And features

PCS is C, D, E, F, G, A, B ♭ for [External character 1]. Therefore, the pitch classes common to these PCSs are C, D, E, G, and A, which are the pitch class sets that can be used in the section of the new chord C Major when the function and the keynote cannot be specified. Stipulate. This pitch class set (C, D, E, G, A) has three functions

[External character 4] Pitch class set for each possibility (each of which has a character that defines a particular tonality) is composed of a common pitch class, so a wide tonality (possibility of three individual keys C, G, F) ) Is shown. In other words, if the melody based on this pitch class set (C, D, E, G, A) is used in the section of the chord C Major whose function and keynote could not be specified, the key before and after this section is C, G, Any of F (Pitch class set (C, D,
E, F, G, A, B) melody, pitch class set to clarify key G (C, D, E, F #, G, A, B) melody, pitch class set to clarify key F (C,
(D, E, F, G, A, B ♭), whichever is used), acts so as not to impair the character of the key before and after this. For example, the pitch class set (C,
If a melody of D, E, F, G, A, B) is played, then the listener clearly knows that the key is C. Also, the chord C Maj played prior to that
Since the transposing feeling is not experienced between the melody in the or section and the subsequent melody in which the key C is clarified, it is noticed that the key of the preceding melody is C.

In this manner, the tonality determining apparatus determines the tonality of the tonality by determining the pitch class set for the chord section whose function and keynote cannot be specified according to a wide tonality interpretation according to the chord type and route. It is possible to obtain a sequence of tonality data (pitch class set) that does not introduce flow defects (wrong transpositions). On the other hand, even if it is difficult to specify the function of the chord with the conventional tonality determination device, the function of the chord is specified without force (without the musical basis) (the possibility of a plurality of functions that the chord can have is Among them, one function is arbitrarily selected), and the pitch class set corresponding to this unfounded chord function is determined as the transposition. As a result, the progression of tonality obtained from chord progression may include defects and erroneous transpositions.

Returning to FIG. 1A, the tonality data generator 70
Includes a type / scale conversion table 72 for converting a chord type into a scale in order to generate a pitch class set usable in a chord section whose function cannot be specified. The combination of the converted scale and the root of the chord defines the usable pitch class set. If desired, a route type / pitch class set conversion table memory may be used to obtain a usable pitch class set directly from the combination of route and type, but this is disadvantageous in terms of storage capacity. For example, the route type / pitch class set conversion table can be implemented by a lookup table memory that returns a common PC as illustrated in FIG. 1F for each route type combination input. Further, the type / scale conversion table 72 can be realized by a look-up table that returns the scale of the pitch structure expression with the scale start sound as the chord root for each chord type input. For example, the pitch structure expression data of the scale for the type “Major” can be expressed by (0, 2, 4, 7, 9) (here,
"0" represents the fundamental note of the scale having the same pitch class as the chord root, and "2" represents the second note of the scale having a pitch of twice the semitone (numerical value "1") from the fundamental note of the scale. The same). Pitch classes CB will be expressed as "0" to "11", respectively. If the pitch class of the chord root is "C", that is, data "0", the pitch that can be used in the section of the chord C Major whose function cannot be specified by adding the chord root data "0" to the data of each note of the scale. Class set data (0, 2, 4, 7, 9), that is, pitch class set (C, D, E, G, A) is obtained.

In summary, the tonality determination device 1 shown in FIG.
The main features and advantages of 0 are as follows. (A) Unlike the conventional tonality determination device, the first tonality determination function (30, 61, 62, 70) that determines the tonality in the following chord section by using the leading key as a clue and the leading key It has both of the second tonality judgment functions (40, 63, 70) for judging the tonality of the chord section without using as a clue. (B) The tonality determination function complements each other's weak points. That is, the first tonality determining function has a high tonality determining ability with respect to a musical situation in which a key in the preceding section can be easily used to determine the key in the subsequent section. In such a musical situation, the second tonality determination function tends to delay the determination of the correct key and the detection of the transposition in the subsequent section. On the other hand, the problem solving ability (tonality determination ability) for the musical situation in which it is difficult to use the key of the preceding section as a clue is excellent in the second tonality determination function. Therefore, by combining the first tonality determination function and the second tonality determination function, it is possible to provide a tonality determination device having a tonality determination capability that cannot be achieved by the conventional tonality determination device. (C) Consideration is given not only to a musical situation in which a key can be specified (a situation presumed by a conventional tonality determination device) but also in a musical situation in which a key cannot be specified (50). That is,
We also take into account situations where chord progressions include chords whose function cannot be specified. (D) For a chord whose function can be specified (by the key / function judging means 30 and 40), the tonality in the chord section is narrowly interpreted (71), while for a code whose function cannot be specified. Widely interprets the tonality in that chord section (72). This gives a tonality progression with a natural flow. (F) That is, for the code whose function can be specified,
The pitch class set that can be used in the chord section is determined according to the combination of the chord, the type, the root and the specified function (or a combination equivalent thereto), while for the chord whose function cannot be specified, The pitch class set that can be used in the chord section is determined from the chord type and route in consideration of the possibility of multiple functions that the chord can take.

By combining the tonality determining device as described above with the accompaniment forming device, it is possible to provide an automatic accompaniment device which plays an accompaniment having a rich tonality and a natural tonality flow in accordance with the input of chord progression. .

The accompaniment forming apparatus for forming an accompaniment by using the tonality determining apparatus as shown in FIG. 1A can be realized in various modes. The basic configuration is shown in a functional block diagram in FIG.

In FIG. 1B, the accompaniment forming device 90 forms a first accompaniment forming device 190 for forming an accompaniment in a chord section in which a function can be specified during chord progression, and an accompaniment in a chord section for which a function cannot be specified. And a second accompaniment forming device 290 for forming the. For this purpose, the first accompaniment forming device 190 has a key hold flag 50 (which is included in the tonality determination device 10 of FIG. 1A but also shown in FIG. 1B for clarity). It operates in the "release key hold" state, and forms an accompaniment based on the combination of chord function and type and key note. Here, the function of the chord and the keynote are the first determination unit 30 of FIG.
Alternatively, it is given from the second judging unit 40 (when the function and keynote of the chord can be specified in each judging unit 30, 40), and the chord type is given from the chord progression input device 20. First accompaniment forming device 190
Instead of the combination of the chord's function, type, and keynote, it may be configured to receive a combination having information equivalent to the combination (for example, the chord's function, type, and root). Alternatively, the tonality data generator 70
(In the section of the chord, that is, the chord in which the function is specified), the information of the available pitch class sets may be received. In any case, the first accompaniment forming device results in the formation of an accompaniment having a pitch of a pitch class included in the pitch class set usable in the musical time of the chord having the specified function. Therefore, the first accompaniment forming apparatus 1 configured to receive a combination of chord functions, types, and keynotes, or a combination equivalent thereto.
Numeral 90, whether explicit or implicit, can be used in some way inside, corresponding to a part of the tonality data generator 70, that is, in the section of the code whose function is specified by the above combination. Tonal pitch class setting means for determining or defining a specific pitch class set. For example, if there is a memory that stores accompaniment data that composes accompaniment sounds in a pitch class selected from a set of pitch classes that can be used for a certain combination of chord function, type, and keynote, information about that combination is given externally. It is assumed that the first accompaniment forming apparatus is configured to select the memory and output the accompaniment data in such a case.
In the case of this configuration, the tonality pitch class setting means is realized by the memory and the means for selecting the memory according to the combination information. Further, when the above memory is configured for each music style, means for defining the tonal pitch class set is realized by considering the music style in addition to the above combinations.

In FIG. 1B, the first accompaniment forming device 190.
A first accompaniment pattern generator 192 for generating an accompaniment pattern suitable for the combination of the chord functions and types.
And an adder 196 (pitch generating means) that adds the pitch data of the accompaniment pattern output from the generation unit 192 to the data of the keynote or chord root to generate the accompaniment tone pitch to be actually played. is doing. When the pitch data of the first accompaniment pattern generation unit 192 is expressed by the pitch from the key note, the actual accompaniment pitch can be obtained by adding the key note to it. Alternatively, when the pitch data of the first accompaniment pattern generator 192 is represented by the pitch from the chord root, the chord root is added to the pitch data to obtain the actual accompaniment pitch.

On the other hand, the second accompaniment forming apparatus 290 operates when the key hold flag 50 is in the "key hold" state (indicating that the function of the chord cannot be specified), and the type of chord whose function cannot be specified. And form an accompaniment based on the route. The second accompaniment forming apparatus 290 may be configured to receive the information of the pitch class set that can be used in the chord section, which is generated by the tonality data generation unit 70, instead of the combination of types of routes. In any case,
The second accompaniment forming device forms an accompaniment having a pitch of a pitch class included in such a pitch class set. Therefore, the second accompaniment forming device configured to receive the combination of the type and the route corresponds to a part of the tonality data generating unit 70 in some way inside, that is, the pitch usable in the chord section whose function cannot be specified. It will include means for determining and defining a class set from its chord type and root according to a broad tonality interpretation.

In FIG. 1B, the second accompaniment forming device 290 generates a second accompaniment pattern generating section 292 for generating an accompaniment pattern according to the chord type, and pitch data of the generated accompaniment pattern (representing the pitch from the root). And the data of the chord root are added to generate an actual accompaniment note pitch.

FIG. 1C shows the first accompaniment pattern generator 19.
2 is a first accompaniment forming apparatus 190M in which the configuration example of FIG. 2 is clarified, and FIG.
The first accompaniment forming apparatus 190M shown in FIG. 1C includes a plurality of accompaniment pattern memories 193-1 to 193-n (9 in total) prepared for each chord function / type (or scale).
1). Each accompaniment pattern memory stores an accompaniment pattern having a pitch content (pitch line) that matches a specific function / type combination (or a specific scale). For example, the first accompaniment pattern memory 193-1 has a function

[External character 3] (tonic) is used to store an accompaniment pattern having a pitch content that conforms to a major chord or an Ionian scale. Further, the accompaniment forming device 190M has a selection unit 194. The selection unit 194 determines a plurality of accompaniment pattern memories 193-1 to 193 according to the function / type (or scale) of the chord from the tonality determination device 10.
Select an accompaniment pattern for the same chord function / type (or the same scale) from -n. The accompaniment sound pitch data selected by the selection unit 194 among the accompaniment sound pitch data P1 to Pn of the plurality of accompaniment pattern memories 193-1 to 193-n is indicated by Ps. The selected pitch data Ps is added by the adder 196 to the data representing the keynote or chord root from the tonality determination device 10. The output data of the adder 196 represents the final pitch of the accompaniment sound. It can be seen that the plurality of accompaniment pattern memories 193 and the selection unit 194 are an example of the configuration of the first accompaniment pattern generation unit 192 of FIG.

In the first accompaniment forming apparatus 190M of FIG. 1C, a plurality of accompaniment pattern memories 193-1 to 193-n are provided.
The rhythm components (sequences of accompaniment sound lengths) included in the plurality of accompaniment patterns in can be mutually independent. In other words, the first accompaniment forming device 190M in FIG. 1C can form an accompaniment line having a rhythm depending on the scale (or the combination of the function and type of chord) from the tonality determination device 10. Therefore, it is possible to perform an accompaniment in which the rhythm changes as the chord changes during chord progression.

In the second accompaniment forming device 290M shown in FIG. 1D, a plurality of accompaniment pattern memories 293 are formed by the same method.
-1 to 293-m (the whole is shown by 293) and the selection unit 29
4 and the second accompaniment pattern generator 29 of FIG.
2 has been achieved. However, each accompaniment pattern memory 29
The accompaniment patterns 3-1 to 293-m have pitch contents adapted to individual chord types (or scales). The selection unit 294 selects the chord type (or the type / scale conversion table 72) given from the chord progression input device 20.
The accompaniment pattern is selected according to the scale given by. FIG. 1E shows another configuration example of the accompaniment forming apparatus as 90M.

In FIG. 1E, the accompaniment forming apparatus 90M includes an accompaniment pattern memory 91 for storing a standard accompaniment pattern for each accompaniment style. The accompaniment pattern stored is repeatedly read by a reading unit (not shown).
The reference pitch of the accompaniment sound read from the accompaniment pattern memory 91 is indicated by P. The accompaniment forming apparatus 90 further includes a pitch change data table memory 94. The difference pitch data ΔP for changing the pitch data P from the accompaniment pattern memory 91 is stored in each table element in the pitch change data table memory 94. The table element of the pitch change data table memory 94 is specified (addressed) by the combination of the scale and the pitch of the accompaniment note to be changed. The data representing the scale from the tonality data generation unit 70 and the accompaniment tone reference pitch data from the accompaniment pattern memory 91 are input to the address unit 95, where an address indicating a table element of the pitch change data table memory 94 is generated. . In the case where the scale input to the address section 93 is a code that can specify the function, the scale that corresponds to the function name is given from the function name / scale conversion table 71, whereas the code that cannot specify the function In the case of, it is the type corresponding scale given from the type / scale conversion table 72. So instead of undergoing such a scale,
The address section 93 may be modified so that it receives the function and type when the key is released (when the code function is fixed) and receives the type when the key is held (when the code function is undefined). Pitch change data in the addressed table element △
P is added to the reference pitch data P from the accompaniment pattern memory 91 by the adder 95. The pitch data output from the adder 95 has a pitch content (for example, represented by a pitch from a keynote) that conforms to the function name corresponding scale when the key hold is released, and a pitch that conforms to the type corresponding scale when the key is held. Has content (expressed by the pitch from the chord root). This pitch data is further added by the second adder 96 to the data from the selection unit 97 to become the pitch of the actual accompaniment sound. The selection unit 97 displays the current keynote memory 52 when releasing the key hold (when confirming the key).
From the chord progression input device 20 is selected when the key is held. In the case of releasing the key hold, as in the case of the key hold, if the contents of the pitch change table memory 94 are set so that the data output from the adder 95 represents the pitch from the chord root, the selector 97 can be used. Is unnecessary, and the actual accompaniment tone pitch can be obtained by adding the chord root to the output of the adder 95 regardless of whether the key is held or released.

As can be seen from the above description, the accompaniment forming device 90M of FIG. 1E also has a function corresponding to that of the first accompaniment forming device 190 and the second accompaniment forming device 290 shown in FIG. 1B. is there. The configuration of FIG. 1E has an advantage that the storage capacity required for accompaniment formation can be minimized.

Specific examples will be described in detail below with reference to FIGS. 2 to 33. FIG. 2 is a block diagram of hardware of an automatic accompaniment apparatus according to a specific embodiment. C
The PU 100 controls the entire automatic accompaniment apparatus. ROM1
In 02, a program executed by the CPU 100 and fixed data (including a chord progression knowledge database) are stored. The RAM 104 is used as a work memory under the control of the CPU 100. The input device 106 includes a keyboard for inputting a melody and a chord. Musical sound generator 108
Generates a tone signal under the control of the CPU 100. The sound system 110 receives a musical tone signal and outputs it to the outside. The timer 112 measures the elapse of a predetermined time, and the CP
A timer interrupt signal for activating the timer interrupt routine (FIG. 4) is given to U100. The display device 114 can display the chord progression, the function progression corresponding to the chord progression, the keynote progression, and the tonality (pitch class set) progression.

FIG. 3 shows the main flow of the operation of the CPU 100 shown in FIG. At power-on, the CPU 100 initializes the tonality determination system (3-1). This initialization process 3-1 includes temporary variables (FIG. 6 (A),
A process of setting (B) and (C) to a predetermined value is included. In 3-2, the CPU 100 key-scans the input device 106 in a normal manner. In 3-3, the CPU 100 controls the musical sound generating device 108 based on the melody key information input from the melody input area of the keyboard to generate a melody sound. In 3-4, the CPU 100 controls the display device 114 to display the input chord progression evaluation result (for example, the progress of the pitch class set). For example, the available pitch class set is displayed in the navigator corresponding to the keys of the keyboard, which facilitates the ad lib performance by the user.

FIG. 4 shows a timer interrupt routine which is periodically activated by the time-out of the timer 112 and executed by the CPU 100. 4-1 CP
The U100 examines the accompaniment key information (acquired by the key scan 4-2) and determines the type and route of the designated chord in a known manner. When a new code is detected (4-2), the CPU 100 determines the function and keynote of the new code (4-3). Next, the CPU 100 generates tonality data representing a pitch class set usable in the new chord section (4-4). Finally CPU100
Forms accompaniment data based on the tonality data, and controls the musical sound generator 108 to generate an accompaniment sound signal.

FIG. 5 shows a chord component sound table CKT stored in the ROM 102. Chord composition sound table CKT
Stores a set of chord constituent tones for each chord type. The constituent sound data of "0" represents the root sound. The other constituent tones are represented by the pitch between them. "1"
Represents the pitch of a semitone (2nd minor), and "2" is the whole tone (2nd major).
Degree), and in the same way, "11" is a long 7
Indicates the pitch of a degree. The data of "15" is a dummy,
Used for triad chords consisting of three constituent notes to unify the number of storage locations for each chord in the chord constituent note table CKT to four. The chord component sound table CKT can also be used when the chord type is specified in the chord discrimination 4-1.

FIGS. 6A, 6B, and 6C are temporary variables (RAM 10) used and referred to in the flowcharts described later.
4)). Variable (register) CD
N represents the new code obtained in code discrimination 4-1. C
The DN is composed of a part CDNr representing the root of the new chord and a part CDNt representing the type of the new chord. For example, the new code of C MAJOR is expressed by CDNr = 0 and CDNt = 0. The variable (register) CDB represents the code immediately before the new code (previous code). CDB is a portion CDBr representing the root of the immediately preceding code and a portion CDBt representing the type of the immediately preceding code.
It consists of and. The variable (register) FDN represents the function name of the new code. The FDN is composed of a part FDNd representing the scale frequency (function) in which the new chord is evaluated with the current keynote, and a part FDNt (the same data as CDNt) representing the type of the new chord. For example,

[External character 5] The function name of is indicated by FDNd = 2 and FDNt = 0. The variable (register) FDB represents the function name of the immediately preceding chord, and a part FDBd indicating the scale frequency (function) obtained by evaluating the immediately preceding chord with the current keynote and a part FDBt (the same data as CDBt) indicating the type of the immediately preceding chord. Consists of. The variable (register) TDN represents the current tonality under the situation where the keynote is specified, and consists of a part TDNk representing the current keynote and a part TDNs representing the current scale. TDNk and TDNs define the pitch class set that can be used in the code section in which the function can be specified. For example, C, D, E, F,
The tonality of C Ionian consisting of G, A, and B pitch classes is expressed by TDNk = 0 and TDNs = 0. On the other hand, the pitch class set that can be used in the chord section whose function cannot be specified is defined by CDNr and TDNs. The variable (table) TDK stores a plurality (here, four types) of relational tonality. Each tonality data TDK [i] of the table TDK is a part TDK representing a keynote.
It consists of k [i] and a portion TDKs [i] representing the scale. The keynote part of the first tonality data TDK [0] at the first address of the table TDK represents the related keynote which is the dominant (general key) of the current keynote. Similarly, in each keynote portion of TDK [1], TDK [2], and TDK [3], keynote data representing the subdominant (lower key) of the current keynote, the dominant of the dominant, and the subdominant of the subdominant are stored. Remembered.
Variable (table) FDK is related keynote TDKk
The function names of the new code CDN and the immediately preceding code CDB evaluated in [i] are stored. Each function name FDK [i] is composed of a part FDKd [i] representing the scale frequency of the code and a part FDKt [i] representing the code type. Table FDK
In the even addresses, the function name of the new code evaluated by each related keynote is stored, and in the odd address, the function name of the immediately preceding code evaluated by each related keynote is stored. Specifically, the data FDK at address 0 of the table FDK
[0] consists of the scale frequency of the new chord evaluated by the dominant of the current keynote and the new chord type. The data FDK [1] at address 1 represents the function name in the dominant key of the immediately preceding code. Similarly, FDK
[2] and FDK [3] represent the function name of the new code in the subdominant key and the function name of the immediately preceding code, respectively, and FDK [4] and FDK [5] represent the new code and the immediately preceding code evaluated by the dominant key of the dominant. FDK [6] and FDK [7] represent the function names of the new code and the immediately preceding code evaluated by the subdominant subdominant key. Regarding the related keynote, the address i of the relationship tonality table TDK is the address 2i (for the new code) and the address (2i + 1) of the table FDK.
It corresponds to (just before the code). The variable i is used as a pointer to elements on various tables.
The variable (register) CDF represents the code two before the CDN, and is composed of a root CDFr and a type CDFt. A variable (register) CDG represents a code three codes before the CDN, and is composed of a root CDGr and a type CDGt. The tonic hold flag CAOS indicates whether the key is fixed (whether the function of the analyzed chord is specified). When inputting the chord progression, the tonic hold flag CAOS is initialized to the "key hold state" (CAOS = 1).
While the chord progression is being input, the tonic hold flag CAOS is in the "key fixed state" when the function of the new chord can be specified in the key note / function determination routine 4-3, and the routine 4-3 is executed.
When the function of the code cannot be specified by, the status is "key hold status". The variable RDTb represents the root difference between the new code and the code immediately before. Similarly, the variable RDTf represents the root difference between the code one before and the code two before, and the variable RDTf
DTg represents the root difference between the code two before and the code three before. The variable ANT represents the pitch of the accompaniment sound actually played.

FIG. 7 shows details of the keynote function judgment 4-3 in the interrupt routine (FIG. 4). First 7
At -1, the CPU 100 checks whether the key is on hold, that is, whether the current keynote is fixed. In detail,
As shown in FIG. 8, when the tonic hold flag CAOS is "1", the current keynote is indefinite, and when CAOS = 0, the current keynote is indefinite. When the current keynote is confirmed, the CPU 100 changes the new code to the function name evaluated by the current keynote (7-2). Specifically, as shown in FIG. 9, the type CDNt of the new chord is set to FDNt (9-1), and the root of the new chord (pitch class)
CDNr is the scale frequency F from the current keynote TDNk
Convert to DNd (9-2).

In the next routine 7-3, the CPU 100 searches the same keynote keeping code table for the function name (function expression) FDN of the new chord. The same keynote keeping code table is placed in the ROM 102, an example of which is shown in FIG. 10 as OFT. The same keynote keeping code table OFT stores a set of chords for keeping keynotes. Each of the stored code data is in the form of a function name, and is composed of a part indicating scale frequency (function) and a part indicating code type. For example, the data (7, 9) at the address 14 of the same keynote keeping code table OFT is the chord function name.

[External character 6] Represents Data "15" indicating the end of the table is stored at the final address (here, 28) of the table OFT.

Same keynote maintenance code table OFT
If contains a code that matches the function name FDN of the new chord, the assumption in 7-2, that is, the assumption that the keynote for the new chord is the current keynote, was correct,
It can be said that the new code is a code (having a function) that maintains the current keynote. Details of the search routine 7-3 are shown in FIG. First, (11-1) the pointer i to the same keynote keeping code table OFT is initialized to the head position "0". Element OFT of the same keynote keeping code table OFT indicated by pointer i
Function name FD of the new code that evaluated [i] with the current keynote
Compare with N (11-3). If they match, the search routine 7-3 ends with success, and after confirmation (7-4), the keynote function determination flow (FIG. 7) also ends. At this time, the keynote in the new chord section is the same as the current keynote, and is indicated by TDNk. The correct function of the new code is also indicated by the FDN. If they do not match, the pointer i is incremented to compare the function name with the next table element OFT [i]. If the same keynote maintenance code table OFT [i] does not include an element that matches the function name FDN of the new code, 11-2
Then, the table end condition OFT [i] = 15 is satisfied, and the process proceeds to 7-5 through 7-4 in FIG.

In 7-5, the CPU 100 generates the function name FDB of the immediately preceding code according to the current keynote TDNk. Specifically, the CPU 100 sets the type variable FDNt to the type CDBt of the immediately preceding code and sets the scale frequency variable FDNd of the immediately preceding code to (CDBr + 12-TDNk) m.
Obtain according to od12.

Next, the CPU 100 searches the parallel code sequence table for a code sequence (function code pattern) consisting of the function name FDB of the immediately preceding code and the function name of the new code (7-6). The parallel code sequence table is stored in the ROM 102, and one example thereof is shown as MCST in FIG. The parallel key code sequence table MCST stores a set of function code patterns that suggests a change from a major key to a minor key (for example, C to Am) within the same key signature. In the memory map of FIG. 12 (see the format), one function code pair supporting the change to parallel minor is stored in each two consecutive addresses of the parallel code sequence table, and the even number address is the first of the function code pair. The function name (scale frequency and code type) of the code is stored, and the function name of the code after the code pair is stored at the odd address. Further, data "15" indicating the end of the table is stored at the final address (here 84) of the table MCST.

Parallel tone code sequence table MCST
If a function code pattern (chord pair) that matches the function code pattern consisting of the function name of the immediately preceding code and the function name of the new code is included in the same key signature, It can be concluded that a change from major to minor is suggested.

Parallel tone code sequence table MCST
The details of the search routine 7-6 for the are shown in FIG. First, in 13-1, the pointer i is initialized to "0". The function name FDB of the immediately preceding code is compared with the table element MCST [i] indicated by the pointer i, and the function name FDN of the new code is compared.
Is compared with the next table element MCST [i + 1] to test the match / mismatch of the function code pattern (14-
3). If they match, the search routine 7-7 ends with success, and after confirmation (7-7), the keynote and function determination flow (FIG. 13) also ends. As a result, the keynote (key signature here) for the new chord section is the current keynote TDNk.
, And the correct function of the new code is indicated by FDN. When the matching test 13-3 does not match, the table address pointer i is incremented by 2 to perform the matching test of the next function code pattern.
When the parallel code sequence table MCST does not include the same function code pattern composed of the function name FDB of the immediately preceding code and the function name FDN of the new code,
In 13-2, the end of the table (MCST [i] = 15) is detected, and the process branches 7-7 in FIG. 7 to proceed to the pivot modulation inspection routine shown in 7-8.

In the pivot modulation inspection, the possibility of modulation from the current keynote to another keynote is checked by using the immediately preceding code and the new code. For this purpose, the ROM 102 is provided with a modulation code sequence table. The modulation code sequence table is composed of the pivot code table PDB shown in FIG. 14 and the post-modulation code table MDB shown in FIG. The pivot chord table PDB stores a set of chord function names (frequency and type) that can be used in the keynote before modulation. Post-modulation chord table MD
In B, a set of function names (frequency and type) of chords that can be used in the keynote after modulation is stored.

The details of the pivot modulation inspection routine 7-8 are shown in FIG. First, in 16-1, the CPU 100 generates four types of relational keynotes (dominant, subdominant, dominant dominant, subdominant subdominant) related to the current keynote TDNk.
As a result, TDKk [0] is the dominant keynote, TD
Kk [1] is the subdominant keynote, TDKk
[2] TDKk, the dominant keynote of the dominant
[3] represents each pitch class of the subdominant subdominant keynote. Next, the CPU 100
In -2-16-5, the function name of the new code and the function name of the immediately preceding code evaluated by each of the four related keynotes are generated. As a result, FDK [0] is the first relational keynote, that is, the function name of the new chord evaluated by the dominant keynote (scale frequency FDKk [0] and type FD
Kd [0]), FDK [1] represents the function name of the immediately preceding code evaluated by the dominant keynote, and FDK [0]
[2] and FDK [3] represent the function names of the new chord and the immediately preceding chord evaluated with the second relation keynote (subdominant keynote), and FDK [4] and FDK [5] have the third relation. The key names (dominant dominant keynotes) of the new chords and the preceding chords are evaluated, and FDK [6] and FDK [7] are the fourth related keynotes (subdominant subdominant keynotes). Each function name of the evaluated new code and the immediately preceding code is shown. The scale frequency FDKd [i × 2] of the new chord evaluated by the (i + 1) th relational keynote is the root CDNr of the new chord and the (i + 1) th relational keynote TDKk.
Using [i] and (CDNr + 12-TDKk [i])
The function of the immediately preceding chord given by mod12 and similarly evaluated at the (i + 1) th related keynote, that is, the scale frequency of the root CDBr of the immediately preceding chord from the (i + 1) th related keynote TDKk [i] FDKd [i
X2 + 1] is (CDBr + 12-TDKk [i]) m
given in od12.

In 16-6 to 16-9, the CPU 10
0 examines the possibility of transposing from the current keynote to each of the first to fourth (i = 0 to 3) related keynotes. The code sequence consisting of the immediately preceding code CDB and the new code CDN is the (i +) th key from the current keynote TDNk.
The conditions for suggesting the modulation to the 1) th relational keynote TDKk [i] are as follows. First, the function name FDB of the immediately preceding code evaluated with the current keynote TDKk is included in the pivot code table PDB [],
The function name FDK [i × 2 + 1] of the immediately preceding chord evaluated by the related keynote TDKk [i] is included in the post-modulation chord table MDB [], and thirdly, the related keynote T
Function name FDK [i × of new code evaluated by DKk [i]
2] is included in the post-modulation code table MDB [] (16-7). If all of these conditions are met, the pivot modulation inspection routine (FIG. 16) ends with "fit". At this point, the keynote for the new chord section (the keynote related to the condition) is TDK
It is stored in k [i]. The correct function of the new code is stored in FDK [i × 2]. So CPU
100 proceeds from 7-9 in FIG. 7 to 7-10, and as shown in FIG. 17 for details, sets TDKk [i] in TDNk to update the current keynote (17-1), and FDN in FDN.
The contents of DK [i × 2] are set (17-2).

If the modulation condition is not satisfied for any of the related keynotes, the pivot modulation inspection routine (see FIG. 1).
6) ends with "not compatible", and the processing branches from 7-9 to the key establishment code sequence table search routine 7
Proceed to -11. If the current keynote is indeterminate (7-1)
This routine 7-11 is also executed.

The key establishment code sequence table referenced by the search routine 7-11 stores a set of code sequences that establishes or suggests a key. Key establishment chord sequences are generally characterized by the fact that the constituent notes of the chords that make up the chord sequence provide a set of pitch classes that suggest a particular key. For example, the code pattern Dminor → G
7th-> CMAJOR first code Dmino
The component sound of r is D, F, A, the second chord G7th is the G, B, D, F, the third chord C MAJOR
The constituent tones are C, E and G, and the pitch class set (C, D, E, F, G, A and B) is obtained by collecting these constituent tones.
Is obtained. This pitch class set defines C major in C major. Therefore, this code pattern is a code sequence that establishes a particular key.
By the way, if you convert this chord pattern into a function name pattern with Keynote C,

[External character 7] Becomes

On the contrary, when the above chord pattern (in the root type expression) is given, in order to judge the keynote suggested by the chord pattern, the following music rule (key establishment chord sequence knowledge) is used. ) Is prepared. Music rule: {condition part} In a chord pattern consisting of two chords before → chord before one → new chord,
(A) The type of new code is "MAJOR",
(B) The previous code type is "7th",
(C) The root of the previous chord is “completely 5 degrees above” the root of the new chord, (d) the type of the chord two before is “minor”, and (e) the chord two before If the root of is "completely 4 degrees below" the root of the previous code, the function of the new code is

[External character 8] Is. The code pattern Dminor → G7th →
C MAJOR satisfies the conditional part of this rule,
The function of the new code C MAJOR is

It is concluded that [gaiji 8]. In addition, this feature

It can be concluded that the keynote suggested by the chord pattern is "C" from [External character 8] and the root "C" of the new chord. (The above description is an outline of the keynote / function determination by the key establishment code sequence table search in this embodiment.)

The code sequences for dominant progression and subdominant progression are typical key establishment code sequences. Tables 1 to 3 show key establishment code sequences belonging to the dominant progress, and Tables 4 to 6 show key establishment code sequences belonging to the sub-dominant progress.

[Table 1]

[Table 2]

[Table 3]

[Table 4]

[Table 5]

[Table 6]

FIG. 18 shows the memory format of the key establishment code sequence table CPD referred to in the search routine 7-11. The key establishment code sequence table CPD stores a set of key establishment code sequences including key establishment code sequences including the code sequences as shown in Tables 1-6. According to the illustrated memory format, the first address CPD [i] of each key establishment code sequence entry is its type part CPDt [i].
Store the type of the last code of the key establishment code sequence (the code to be compared with the new code from the chord progression input device), the next address CPD [i + 1] is one in its type part CPDt [i + 1]. The code type of the previous route is stored, and the route part CPPr [i +
1] stores the difference between the route of the last code and the route of the last code. Similarly, the subsequent addresses store up to the condition of the first code of the key establishment code sequence (the type and the difference between the route and the next route), and the next address stores the key in the type part (upper digit). The data end mark "100" of the established code sequence is stored, and the function of the last code of the code sequence is stored as a conclusion in the root part (lower digit). The key establishment code sequence table CPD includes a code sequence consisting of four codes as the longest key establishment code sequence (of course, it is not limited to this).

FIG. 19 shows the details of the search routine 7-11 for searching the key establishment code sequence table CPD. First, the head of the key establishment code sequence table CPD is located by i = 0 (19-1), and RDTb
= CDNr-CDBr calculates the difference between the root of the new chord and the root of the immediately preceding chord (19-2). In the illustrated flow, another route difference RDTf (the root CDBr of the immediately preceding code and the root C of the immediately preceding code C
Difference from DFr), RDTg (root C of the code two before)
Although the difference between DFr and the root CDGr of the code three codes before is calculated in a loop (19-6, 19-1).
0), in order to increase the search speed, in the actual program, the route differences RDTb, RDTf, RD
Tg is calculated. In the following 19-3 to 19-16, the key establishment code sequence that matches the input code pattern ending with the new code is searched.

If the input code pattern (of length 2) consisting of the new code and the preceding code satisfies the condition of the key establishment code sequence of length 2 in the key establishment code sequence table CPD, it is 19-3. 19-4 is established. That is, CDNt = CPDt [i], CDBt = CP
Dt [i + 1], RDTb = CPDr [i + 1], CP
Dt [i + 2] = 100 is established. In this case, the search routine proceeds to 19-5, locates the conclusion of the key establishment code sequence (the function of the specified new code) by i = i + 2, and "fits" to the keynote / function determination routine. return.

An input code pattern (of length 3) consisting of a new code, a code one code before and a code two codes before is of a certain key establishment code sequence (of length 3) in the key establishment code sequence table CPD. If the condition is satisfied, 1
9-3 established, 19-4 not established, 19-7 (CDFt = C
PDt [i + 2], RDTf = CPDr [i + 2]) and 19-8 (CPDt [i + 3] = 100).
In this case, the search routine proceeds to 19-9 and i = i + 3.
Locates the conclusion of that key establishment code sequence and ends with "conforming".

Similarly, when the input code pattern of length 4 from the new code to the code immediately before the third code satisfies the condition of a certain key establishment code sequence (of length 4) in the table CPD, 19-3 Established, 19-4 Not established, 19-
7 established, 19-8 not established, 19-11 (CDGt = CP
Dt [i + 3], RDTg = CPDr [i + 3]) holds. Therefore, the search routine is 19-12 and i = i + 4.
Locates the conclusion of that key establishment code sequence and ends with "conforming".

If the input code pattern does not satisfy the condition of the key establishment code sequence of interest, 19-3
Either 19-7 or 19-11 is not established. In this case, the search routine locates the next key establishment code sequence according to 19-13 to 19-16 and returns 1 for the matching test of this sequence and the input code pattern.
Return to 9-3 to continue the search. 1 if the input code pattern does not match any key establishment code sequence in the key establishment code sequence table CPD, 1
In 9-16, the end of the search of the table CPD is detected, and the search routine returns "not compatible" to the keynote / function determination routine.

The keynote / function determination routine (FIG. 7) checks at 7-12 whether or not the key establishment code sequence table search routine ends with "match".
If “matching”, key hold release 7-1 3 and key setting 7-14 are executed. That is, as shown in FIG.
The hold state of the key is released by AOS = 0, and as shown in FIG. 22, the root of the new code and the search routine 7-11.
From the conclusion of the key establishment code sequence found in (the function of the last code), TDNk = CDNr + 12-CPD
The key note TDNk is obtained by r [i] mod12 (22-1), and the conclusion of the key establishment code sequence is determined by FDNd = CPDr [i].
(22-2). On the other hand, when the key establishment code sequence table search routine ends with "not compatible", the keynote / function determination routine sets the key hold state by CAOS = 1 (FIG. 20) (7-1).
5).

FIG. 23 shows a flow of the scale judgment routine 4-4. The scale determination routine 4-4 is executed after the keynote / function determination routine 4-3 (FIG. 7).
As can be seen from the flow of FIG. 7, when the function of the new code can be specified by the keynote / function determination routine 4-3, the key hold flag CAOS indicates the key hold release state “0”.
If the function of the new code cannot be identified,
The key hold flag CAOS becomes the key hold state "1".
In the former case (23-1), the scale determination routine determines the scale in the section of the new code by the function name / scale conversion table in order to interpret the tonality narrowly (2
3-2). In the latter case, the scale determination routine determines the scale in the section of the new chord by the chord type / scale conversion table in order to widely interpret the tonality (23-3).

FIG. 24 shows an example of the structure of the function name / scale conversion table as SCT. The function name / scale conversion table SCT stores the correspondence between the function name (function and type) of the code and the scale. In the case of the memory map of FIG. 24 (see format), each of the table SCT is 2
One code function name (scale frequency and type) is stored in an even-numbered address of one continuous address, and a scale name that matches the function name is stored in an odd-numbered address that is the next address. For example, address 6 and 7 are function names

[External character 9] Mixolydian MIXOLYDIAN (data 2) is shown as a scale suitable for (data 7, 0). However, as shown in the addresses 60, 62, 64, special code types (code types Aug, dim,
as well as

[External character 6] For a code of SUS4) other than SUS4, there is a property that a suitable scale is determined irrespective of the scale frequency from the keynote of the root of the code. In order to show this property (function), data "14" is stored in the frequency area (function part) of these special codes. In other words, the scales for these codes are obtained by type / scale conversion. Further, the code "15" indicating the end of the table is stored at the final address (here 66) of the table SCT.

A detailed flow of the function name / scale conversion routine 23-2 is shown in FIG. At the time of entering this flow, the correct function name of the new code is stored in the FDN as a result of the keynote / function determination process (FIG. 7).
Nd represents the scale frequency (function) of the new code, and FD
Nt represents the type of new code. First 25-1
CPU100 is function name / scale conversion table SCT
The pointer i for is initialized to "0". Pointer i
In the loop of 25-2 to 25-5, from the table top (i = 0) to the table end (SCTd [i] = 15)
Is incremented by 2. The function name FDN of the new code is compared with the table element SCT [i] pointed to by the pointer i (25-4). If they match, the next table element S
The scale identification data stored in CT [i + 1] is set in the scale section TDNs of the tonality data memory TDN (25-7). If the special code that identifies the scale only by type is a new code, type F of the new code
A table element SCTd [i] having data SCTt [i] matching DNt and having data SCTd [i] = 14 indicating a special code is found (25-
3). In that case, the scale identification data stored in the next table element SCT [i + 1] is stored in the tonality data memory TD.
It is set in the scale part TDNs of N (25-6).

The table SCT can be modified to perform the function name / scale conversion for the special code. Further, in the illustrated table SCT, the function name and the scale are not in one-to-one correspondence, but when used for accompaniment (not in scale display), there is a one-to-one correspondence, that is, a unique scale name for each function name. Table SCT to return
Different for each function name (with pitch line)
It is desirable to have accompaniment.

FIG. 26 shows an example of the structure of the code type / scale conversion table as CFD. The code type / scale conversion table CFD stores the scale for each code type. Such a scale can be defined according to the common PC scheme as described with respect to FIG. Therefore, in the code type / scale conversion routine 23-2, as shown in FIG. 27, TDNs = CFD
With [CDNt], the scale data for the new code type CDNt is fetched from the conversion table CFD and set in TDNs.

FIG. 28 shows the details of the accompaniment process 4-5 (FIG. 4). FIG. 29 shows an example of the standard accompaniment pattern used for each music style in this flow. This accompaniment pattern is C as a keynote and chord function name

[External character 10] Is assumed. Accompaniment pattern memory AM shown in FIG.
The data representing the accompaniment pattern of FIG. 29 is stored in (ROM 102). Specifically, accompaniment pattern memory AM
The data at the address is the pitch part (first part) AM _P
And comprising a pitch change table PCT portion representing a table column number (second part) AM _T shown in FIG. 31.
According to the flow of FIG. 28, the CPU 100 repeatedly reads the accompaniment pattern memory AM. When the first portion of the read data represents a pitch, the CPU 100 modifies the pitch via the pitch change table PCT in order to adapt the pitch to the tonality, and sends the modified pitch data to the musical sound generator 108. , Generate a tone signal.

Pitch change table PC shown in FIG.
T stores difference pitch data for adapting the pitch from the accompaniment pattern memory AM to the tonality. Each column of this table PCT represents each pitch of the reference accompaniment pattern,
Particular column by column pointers AM _T related to the pitch read out from the accompaniment pattern memory AM in operation is designated. Each row of the table PCT represents each scale, and a specific row is designated by the scale data from the scale determination routine 4-4 during operation. At the intersection of a column and a row (at a specified address for the row and column), the difference pitch data used to change the pitch represented by that column according to the scale represented by that row is stored. Pitch change table P
CT is roughly divided into a function-compatible scale group and a type-compatible scale group. The function-corresponding scale group corresponds to the scale group obtained from the function name / scale conversion table SCT, and the type-corresponding scale group corresponds to the scale group obtained from the type / scale conversion table CFD.
However, since both tables SCT and CFD include non-judgment codes, the areas of the pitch change table PCT corresponding to these codes are made common.

In the flow of FIG. 28, the CPU 100 first advances the accompaniment pattern pointer j by 1 (28-
1), when j reaches the size of the accompaniment pattern memory AM (16 in the case of FIG. 30) (28-2), j is returned to "0" indicating the beginning of the accompaniment pattern memory AM (28-3). Next, the CPU 100 checks whether or not the first data part AM _P [j] of the accompaniment pattern memory AM pointed to by the pointer j is data indicating a pitch (28-4). If this is not the case, the accompaniment processing routine is ended without doing anything. AM
_{When P} [j] represents a pitch, 28-5
At 28-8, pitch data of accompaniment sound is generated.

Specifically, when the current key state is not the key hold state (28-5) and the new code (current code) is not the non-judgment code (28-6), as shown in 28-7, Data ANT representing the pitch of the accompaniment tone actually played is generated by ANT = AM _P [j] + PCT (TDN _S ) (AM _N [j]) + TDN _K. That is, the reference pitch data AM _P [j] from the accompaniment pattern memory AM is added with the difference pitch data in the row TDN _S , AM _N [j] column of the pitch change table PCT, and the result is added to the current keynote TD.
N _K is added to obtain real accompaniment pitch data ANT. Here, the difference pitch data PCT (TDN _S ) (AM _N [j]) is the keynote C and the chord function name.

The pitch AM _P [j] of the standard accompaniment pattern written so as to conform to [External character 10] is the function name of the current chord (or the function name corresponding to this function name) specified by the keynote / function determination routine (FIG. 7). It is for changing so as to fit the scale). Therefore, AM _P [j] + PCT
The term (TDN _S ) (AM _N [j]) represents the accompaniment pitch when the keynote is C and the function name is the function name of the current chord. Therefore, the current keynote data TDN _S
By adding, the accompaniment pitch that matches the current music situation, that is, the accompaniment pitch that matches the current keynote and the current chord function name can be obtained.

When the current key state is the key hold state, as shown in 28-8, the actual accompaniment pitch data ANT is obtained by ANT = AM _P [j] + PCT (TDN _S ) (AM _N [j]) + CDNr. To generate. That is,
The TDN _S row of pitch change table PCT based pitch data from the accompaniment pattern memory AM, AM _N [j]
The difference pitch data in the row is added, and the current chord root CDNr is further added to the result to obtain the real accompaniment pitch data ANT. The difference pitch data in this case is a scale "IONIA" starting with the fundamental note C of the same pitch class as the chord root C.
N ", that is, pitch class set (C, D, E, F,
G, A, B) to change the pitch AM _P [j] of the reference accompaniment pattern written so as to conform to a scale in which the type of the current chord is interpreted with wide tonality. . Therefore, AM _P [j] + PC of this equation
The term T (TDN _S ) (AM _N [j]) represents the pitch when the fundamental is C (the chord root is C) and the scale corresponds to the type of the current chord. Therefore, if the data CDNr of the current chord root is added to this term, the accompaniment pitch suitable for the current music situation, that is, the usable pitch class set determined according to a wide tonality interpretation from the root and type of the current chord, is applied. Accompaniment tone pitch is obtained.

Even if the current key state is the key hold release state, if the current chord is the key non-judgment code, the real accompaniment tone pitch data ANT is obtained according to the key hold state (2
8-8). Finally, the CPU 100 sends a note-on command including the pitch data of the accompaniment sound thus generated to the musical sound generation device 108 to generate a musical sound of the corresponding pitch (28-7).

FIG. 32 shows an accompaniment example by this apparatus. As a chord progression, C Major → F Ma
jor → G Major → C Major is used. From this chord progression, the keynote is "C" and the function name progression is based on the keynote / function judgment (Fig. 7).

[External character 11] Is obtained (when C is given as a keynote of the section preceding this chord progression). Therefore, the accompaniment of the section of C Major is formed without changing the pitch content of the accompaniment pattern memory AM. However, the pitch line C5 → E5 → B4 → E5 → A of the reference accompaniment pattern stored in the section of the chord F Major.
Among 4 → E5 → B4 → E5, E5 is raised by a semitone according to the pitch change data “1” in the pitch change table PCT (chord function name

[External character 12] So C5 → F5 → B4 → F5 → A4 → F5 → B4 →
The pitch line of F5 is formed. Also, the code G M
In the ajor section, the code function name is

[External character 9] Therefore, according to the pitch change table PCT, C5 in the pitch line of the reference accompaniment pattern is semitone lowered to B4 by the pitch change data "1" E5, B4,
A4 is lowered by the pitch change data "2" for all notes D
5, A4, G4, resulting in B4 → D5 → A
A pitch line of 4 → D5 → G4 → D5 → A4 → D5 is formed.

FIG. 33 shows an accompaniment example when the key is held by this apparatus. In detail, FIG. 33 shows that when the key is on hold (the current keynote does not exist or the music situation is unknown),
When a chord C Major is given, the accompaniment played in this chord section is shown. The code C Major given under such a music situation
On the other hand, the keynote / function determination routine (FIG. 7) cannot specify the function of the code C Major. Therefore, the pitch class sets that can be used in the section of the chord C Major are C, D, E, G, and A (see FIG. 1).
(See (F)). In this case, as a result of executing the type / scale conversion routine, the pitch change table PCT S
The row of CALE1 (the row corresponding to the type Major) is selected to change the pitch from the reference accompaniment pattern memory AM. As a result, the pitch line of the reference accompaniment pattern, C5 → E5 → B4 → E5 → A4 → E5 → B4 → E
5 is a pitch line (accompaniment pitch line of FIG. 33) C5 which conforms to the pitch class set C, D, E, G, A.
It is changed to E5 → G4 → E5 → A4 → E5 → G4 → E5. Comparing both pitch lines, it can be seen that the standard accompaniment pitch B4 has changed to G4.
4 is the result of adding to the pitch data). The reason for the change is that the pitch class B of the pitch B4 is not included in the usable pitch class sets C, D, E, G and A. Since the accompaniment of FIG. 33 has the possibility of three keys C, F, and G, even if the subsequent accompaniment is to clarify any of key C, key F, and key G, the tonality is the same. Acts so as not to damage.

As described above, in the present automatic accompaniment apparatus, when the chord function cannot be specified, the accompaniment that allows a wide tonality from the chord type and the root is performed in the chord section, so that the musical piece The tonality flow of the entire accompaniment can be made natural.

[0086]

[Modification] The description of the embodiment has been completed, but various modifications are possible within the scope of the present invention. For example, a plurality of tonality determination modules may all be operated to determine the tonality of the same chord in the chord progression, and the final tonality determination may be performed based on the resulting tonality determination results. it can. For example, in a situation in which the key of the preceding section is known, the function of the new chord is judged not only by the first judging unit 30 which uses the preceding key information for judging the function of the new chord, but also by using the preceding key information as a clue. The second determination unit 40 also operates. Then, the result of the first determination unit 30 and the result of the second determination unit 40 are combined according to a predetermined logic to determine the final tonality. The following can be used as the predetermined combination logic. (A) In the case where only one of the plurality of determining units (for example, the first determining unit) can specify the function of the new code (when the determination is successful) and the other determining units (the second determining units) fail in the determination. Uses the result of the judgment unit of the judgment success in the final tonality determination. (B) When a plurality of judgment units gives the same judgment result as the function of the new chord in the judgment success, the judgment result is finally judged. (C) When multiple judgment sections succeed in judgment but different judgment results are obtained as a function of the new chord, the final adjustment is performed by the method of (c1) or (c2) below. Determine sex. (C1) Among the determination results of the plurality of determination modules that have succeeded in the determination, the determination result in which the length of the code sequence used for the determination is the longest is used for the final tonality determination. For example, the determination result of the first determination unit is functional.

It is [external character 3], and it is assumed that the code sequence knowledge of length 2 (code number 2) is used for the determination, and the determination result of the second determination unit functions.

[External character 1] and if the code sequence knowledge of length 3 is used for the determination, the determination result of the second determination unit is “new code function”.

[External character 1] "is used to determine the final tonality (c2).
The tonality that allows a plurality of different functions, which are the determination results of the plurality of determination modules that have been successfully determined, is determined as a possibility. Can also use the logic of (c2)). For example, for the code C Major, the first determination unit determines the function of this code.

It is judged as [External character 3], and the second judgment unit confirms the function of this code.

Function when it is judged as [External character 1]

Pitch class set C that can be used for [gaiji 3],
Function with D, E, F, G, A, B

Pitch class set C that can be used for [gaiji]
The pitch classes C, D, E, F, G, and A common to both D, E, F, G, A, and B ♭ are determined as the pitch class set usable in the section of the chord C Major.

The data structure or program structure of the chord progression knowledge base 60 can take various forms. For example,
As the same keynote maintaining code table 61, in each bit memory designated by a combination of type and function, it indicates whether the combination holds the same keynote (bit "1") or not (bit "0"). Store bit information. For example, if the total number of types E16 functions is 12, and if one word is a 16-bit memory (a memory that stores 16 bits at each address), the same keynote maintenance table 61 is used for consecutive 12-address memories. Can be realized. In this case, the same keynote search routine determines from the type and function of the input code, the corresponding address and the bit position in the word at that address (information whether the input code type and function maintain the same keynote. Since it is possible to directly calculate (the place where is stored), high-speed search and high-speed same keynote determination can be achieved.

Alternatively, if the data in each table memory is ordered and a high-speed search such as a binary search (binary tree search) can be performed, the processing speed of tonality determination can be further increased, and real-time processing can be performed. Desirable for applications. For example, in order to adapt the key establishment code sequence table 63 to a binary search, key entry code sequences to be entered into the table are arranged in ascending order of data indicating the type of code at the end of the sequence, and those having the same type of end code are The data representing one root difference (difference between the root of the code immediately before and the root of the termination code) are arranged in ascending order, and thereafter, the key establishment code sequence is similarly arranged. Each key establishment code sequence memory is, for example, the end code type in the second part of the first address, the first route difference in the first part of the next address, the previous code type in the second part, and so on. The format is such that the sequence end mark is stored in the first part of the address and the function of the termination code is stored in the second part.
A sequence start index table is provided, and for each address of the key establishment code sequence table, the sequence start index indicating the address on the key establishment code sequence table at which the code sequence stored as part of that address starts is stored. Let In operation, the key establishment search routine (A1) first retrieves the key establishment code sequence in the middle of the key establishment code sequence table (converting table size x 1/2 with the index table). (A2) The type data x1 of the code at the end of the sequence is compared with the type data y1 of the new code of the input code pattern. (A3)
If x1> y1, the key establishment code sequence in the center of the first half of the key establishment code sequence table is extracted (current address × 1/2 is converted by the index table), and if (A4) x1 <y1, the key establishment code sequence is obtained. The key establishment code sequence in the center of the second half of the table is taken out (current address + (table size-current address) x 1/2 is converted by the index table). x1
= Y1, the first root difference data x2 of the sequence and the first root difference data y2 of the input code pattern are compared,
If x2> y2, execute (A3). If x2 <y2, execute (A3).
4) is executed, and if x1 = y1 and x2 = y2, the type data x3 of the code immediately preceding the sequence is compared with the type data y3 of the code immediately preceding the input code pattern to obtain x3.
> Y3 (expressed as x1, x2, x3> y1, y2, y3), execute (A3), x3 <y3 (x1, x2,
If x3 <y1, y2, y3), execute (A4). Similarly, x1, x2, x3 ...> y1, y2, y3 ...
If ... (A3) is x1, x2, x3 ... <y1, y2,
If y3 ... (A4) is executed. (A5) Then (A
3) or (A4), the key establishment code sequence extracted again is compared with the input code pattern, and x1 ...> y1
...... Then, the key establishment code sequence in the center of the 1/4 area of the table in front is taken out, and x1 …… <y1 ……
Then, the key establishment code sequence in the center of the 1/4 area of the rear table is taken out. By continuing the above processing, the search areas are 1, 1/2, 1/4, 1/8 ...
... will converge. The search end condition is (r
1) When the end mark of the sequence is seen while the comparison between the input code pattern and the condition part of the key establishment sequence remains the same, (r2) the address of this key establishment code sequence is It is “not compatible” when it is the same as the address of the key establishment code sequence in the search path. If such a binary tree search is performed, the search for the table having 127 key establishment code sequences can be completed by comparison at most about 7 times (comparison between the input code pattern and the key establishment sequence).

In addition, AND. If the chord sequence table is composed of OR-type music rules, the storage capacity can be saved. For example, (a) (the termination code type is t
1) and (the first route difference is r1 and the preceding code type is t2) or (the first route difference is r2 and the preceding code type is t3), the function of the termination code is f1. AND. The OR-type music rule can be stored with data t1, r1, t2, f1, a sequence end mark (or sequence length), and an AND / OR identification mark bit. Therefore, if the two music rules of AND format are (b) (the end chord type is t1) and (the first root difference is r1) and (the preceding chord type is t2), the end chord function is f1. Yes, (c)
If (termination code type is t1), (first route difference is r2), and (previous code type is t3), the function of the termination code is f1, compared with the termination code type t1 and function f1. , It is possible to avoid repetition of data for the end mark.

[0090]

As described in detail above, the tonality determination device of the present invention uses the current keynote as a clue to analyze the chord progression to determine the tonality in each chord section. Means and second tonality determination means for analyzing the chord progression and determining the tonality in each chord section without using the current keynote as a clue. The bitonality determining means act so as to cover each other's defects. Therefore, it is possible to provide a tonality determination device in which the tonality determination capability is improved as a whole, the correct key detection delay is small, the erroneous key determination occurrence rate is low, and the transposition detection error is small.

[Brief description of drawings]

FIG. 1A is a functional block diagram showing an aspect of a tonality determination device according to the present invention.

FIG. 1B is a functional block diagram of an accompaniment forming apparatus that forms an accompaniment by using the determination result of the tonality determination apparatus as shown in FIG.

FIG. 1 (C) is a functional block diagram showing a configuration example of a first accompaniment forming device shown in (B).

FIG. 1D is a functional block diagram of the second accompaniment forming device shown in FIG.

FIG. 1 (E) is a functional block diagram of another configuration example of the accompaniment forming apparatus.

FIG. 1 (F) is a diagram showing a table to which a common PC method is applied in order to obtain a pitch class set that can be used in a chord section whose function cannot be specified.

FIG. 2 is a hardware block diagram of an automatic accompaniment apparatus according to a specific embodiment of the present invention.

3 is a flowchart of a main program executed by the CPU of FIG.

FIG. 4 is a flowchart of an interrupt routine executed by the CPU of FIG.

FIG. 5 is a diagram showing a chord component sound table.

FIG. 6A is a diagram showing a part of a list of temporary variables.

FIG. 6B is a diagram showing a part of a list of temporary variables.

FIG. 6C is a diagram showing a part of a list of temporary variables.

FIG. 7 is a flowchart of keynote / function determination.

FIG. 8 is a flowchart of a current key hold inspection.

FIG. 9 is a flowchart of function generation of a new code.

FIG. 10 is a diagram illustrating an example of the same keynote keeping code table.

FIG. 11 is a flowchart for searching the same keynote maintenance code table for a new code.

FIG. 12 is a diagram illustrating a parallel code sequence table.

FIG. 13 is a flowchart for searching a parallel tone code sequence table for a code pair consisting of an immediately preceding code and a new code.

FIG. 14 is a diagram illustrating a pivot code table.

FIG. 15 is a diagram illustrating a post-modulation code table.

FIG. 16 is a flowchart of a pivot modulation inspection including a search of a pivot code table and a post-modulation code table for a code pair consisting of an immediately preceding code and a new code.

FIG. 17 is a flowchart of keynote updating.

FIG. 18 is a diagram illustrating a key establishment code sequence table.

FIG. 19 is a flowchart for searching the key establishment code sequence table for an input code pattern including a new code.

FIG. 20 is a flowchart of key hold setting.

FIG. 21 is a flowchart of key hold release.

FIG. 22 is a flowchart of key setting.

FIG. 23 is a flowchart of scale determination.

FIG. 24 is a diagram illustrating a function name / scale conversion table.

FIG. 25 is a flowchart of a function name / scale conversion routine.

FIG. 26 is a diagram illustrating a code type / scale conversion table.

FIG. 27 is a flowchart of a chord type / scale conversion routine.

FIG. 28 is a flowchart of an accompaniment process.

FIG. 29 is a diagram showing a score of an example of an accompaniment pattern.

FIG. 30 is a diagram showing an accompaniment pattern memory for storing the accompaniment pattern of FIG. 29.

FIG. 31 is a diagram showing a pitch change table.

FIG. 32 is a diagram showing a score of an accompaniment example.

FIG. 33 is a diagram showing a score of an accompaniment example when keys are held.

[Explanation of symbols]

10 Tonality determination device 90, 90M Accompaniment forming device 20 Chord progression input device 30 First determination unit (determines the function and keynote of a new chord using the current keynote as a clue) 40 Second determination unit (current keynote The function of the new chord and the keynote are judged without using a clue. 60 Chord progress knowledge database 50 Key hold flag (indicating whether the current keynote is unknown) 51 Current keynote memory 52 Keynote updating section 70 Tonality data Generator (determines the pitch class set that can be used in each chord section) 71 Function name / scale conversion table (returns the scale that can be used in the chord section where the function can be specified) 73 Type / scale conversion table (function Returns the scale that can be used in the chord section that could not be specified.) 190 1st accompaniment forming device (specified function 290 Second accompaniment forming device (forming accompaniment for chord segment whose function could not be specified) 100 CPU 102 ROM 104 RAM 106 Input device

Claims (7)

[Claims] 1. A chord progression giving means for giving a chord progression in which each chord is represented by a root and a type, a current keynote storage means for storing data representing a current keynote relating to a current tonality, and the chord progression. A new tonality determining means for determining the tonality of a new chord based on the current keynote, and a tonality of a new chord from the chord progression, based on the current keynote. Tonality determining device that determines by analyzing a code pattern formed between a new code and a code preceding the new code. 2. The tonality determining apparatus according to claim 1, wherein the first tonality determining unit is operable when the current keynote is specified, and the second tonality determining unit is the current tonality determining unit. A tonality determination device which is operable when a key note is not specified. 3. The tonality determining apparatus according to claim 1, wherein the second tonality determining unit is operable when the tonality of the new chord is not specified by the first tonality determining unit. A tonality determination device characterized by being present. 4. The tonality determining apparatus according to claim 1, wherein the first tonality determining means stores the same keynote keeping code table storing means for storing a set of codes for keeping the keynotes the same, and a certain key. Modulation code sequence table storage means for storing a set of modulation code sequences indicating a modulation from a note to another keynote, the current keynote, the same keynote maintenance code table storage means, and the modulation code sequence table storage means A first function name / keynote determination means for determining the function name and keynote of the new chord by analyzing the chord progression using, and the function name determined by the first function name / keynote determination means. Tonality data that defines the pitch class set that can be used in the new chord section based on First tonality data generating means for generating, and the second tonality determining means stores key establishment code sequence table storage means for storing a set of key establishment code sequences that establish a keynote, and the key establishment The second function name / keynote determination means for analyzing the chord progression using the chord sequence table storage means to determine the function name and keynote of the new chord, and the second function name / keynote determination means. Second tonality data generating means for generating tonality data defining a pitch class set usable in the new chord section based on the determined function name and keynote. Tonality determination device. 5. The tonality determining apparatus according to claim 4, wherein the first function name / keynote determining means determines whether or not the new chord has a function of maintaining the current keynote. A key pattern maintained by the key keeping and searching means for searching the same key note keeping code table storing means, and the chord pattern formed by the new chord and the chord preceding it changes the transposition from the current key note to the related key note. A modulation search means for searching the modulation code sequence table storage means to determine whether or not it has a suggested function; and the second function name / keynote determination means includes the new code and the preceding code. To determine whether the chord pattern formed by the chord with which it has the function of establishing a particular keynote And a key establishment determination means for searching the key establishment code sequence table storage means, wherein the tonality determination device is determined when the related keynote is determined by the first tonality determination means. The first keynote updating means for updating the current keynote storage means by the related keynote, if the specific keynote is determined by the second tonality determination means, the first keynote updating means is determined by the determined specific keynote. And a second keynote updating means for updating the current keynote storage means, and a tonality determination device comprising: 6. A chord progression giving means for giving a chord progression in which each chord is represented by a root and a type, a current keynote storage means for storing data representing a current keynote relating to the current tonality, and the chord progression. A new tonality determining means for determining the tonality of a new chord based on the current keynote, and a tonality of a new chord from the chord progression, based on the current keynote. Tonality determining means for determining by analyzing a code pattern formed between a new code and a code preceding the new code, and the first tonality determining means determines the tonality of the new code. A first accompaniment forming means that operates when specified, and forms an accompaniment according to the specified tonality in the section of the new chord, and the new chord determination means by the second tonality determination means. Automatic accompaniment apparatus tonality operates when specified, and having a second accompaniment forming means for forming an accompaniment in accordance this specified tonality in the new code section of the. 7. The automatic accompaniment apparatus according to claim 6, wherein each of the first tonality determination means and the second tonality determination means includes a function name of the new chord as tonality information which is a determination result. A keynote is provided to each of the first accompaniment forming means and the second accompaniment forming means, and each of the first accompaniment forming means and the second accompaniment forming means generates an accompaniment pattern according to a given function name. Automatic accompaniment pitch generating means for transposing each pitch of the accompaniment pattern from the accompaniment pattern generating means in accordance with a given key note to generate an accompaniment pitch Accompaniment device.