JP2995237B2 - Tonality determination device - Google Patents

Description

Description TECHNICAL FIELD [0001] The present invention relates to a tonality discriminating apparatus for automatically discriminating the tonality (tonality) of a given music piece.

<Background Art of the Invention> Conventionally, various tonality determination devices for determining the tonality (tonality) of input music have been studied.

When these tonality discriminating devices are of high quality, the knowledge and know-how of a specialist are required to create an algorithm for tonality judgment. Alternatively, if the accuracy of the determination is to be improved, the types of tonality that can be determined must be limited as a result. For example, even if it is possible to distinguish between minor and major, it is not possible to distinguish other tones such as Okinawan tone and blues.

<Object of the Invention> The present invention has been made in view of the above points,
It is an object of the present invention to provide a tonality discriminating device which can discriminate various tones (tonality) while having a relatively simple configuration.

<Structure and operation of the invention> That is, according to the present invention, in order to achieve the above object,
Music information supply means for supplying music information composed of a plurality of notes; appearance frequency storage means for storing the appearance frequency of each scale sound for each of a plurality of keys; Accumulating note lengths for each note name and performing a predetermined calculation for each predetermined number of notes based on the accumulated value and the appearance frequency of each of a plurality of keys stored in the appearance frequency storage means. A similarity calculating means for calculating the similarity between the music information and each of the plurality of keys up to that point in real time, and a display means for sequentially updating and displaying the similarity for each key calculated by the similarity calculating means. Are provided.

FIG. 1 shows an example of this configuration. The music information supply means GS includes, for example, a keyboard or is connected to a keyboard so that music information can be input from this keyboard. This music information is represented by a pitch and a pitch.

Appearance frequency storage means (Membership function supply means) MS
Includes, for example, storage means for storing in advance a membership function representing the frequency of appearance of each scale for a plurality of tones (tonality). The storage contents of this storage means are based on the result of analyzing a plurality of music pieces.

The similarity calculation means CA calculates the similarity between the music information from the music information supply means GS and the membership functions for a plurality of keys from the appearance frequency storage means (membership function supply means) MS. . Specifically, the musical note length of the musical composition information supplied from the musical composition information supply means GS is accumulated for each note name, and the accumulated value and the appearance frequency storage means (membership function supply Means) By performing a predetermined calculation based on the appearance frequency of each of the plurality of keys stored in the MS, the similarity between the music information and each of the plurality of keys is calculated in real time up to that point. By calculating the similarity, it becomes clear what tone the music has. Then, the similarity for each key, which is the calculation result of the similarity calculating means CA, is displayed on the display means (output means) CU. This display means (output means) OU informs the user of the tonality determination result in various display forms. Therefore, if the membership function of each key is prepared in advance, the tonality determination of the music can be performed only by executing a relatively simple calculation.

<Example> Hereinafter, the present invention will be described in detail with reference to an example shown in the drawings.

FIG. 2 shows the overall configuration, and the CPU 1 controls the overall operation of the tonality determination device. This CPU1 has ROM
2, a RAM 3, a MIDI interface 4, and an LCD driver 5 via a common bus 6. In addition, a switch section for inputting control information is provided as necessary (not shown).

The ROM 2 stores a program for the operation of the CPU 1 and membership functions for a plurality of tones (tonality). This membership function will be described later in detail.

The RAM 3 is used as a work memory or the like when the CPU 1 executes various calculations. Also, the MIDI interface 4 includes
An interface to which performance information is given from an external electronic musical instrument EM. MIDI means MUSICAL INSTRUMENT DIGIT
Abbreviation for AL INTERFACE. This MIDI interface 4
The performance information given via the
It is memorized in M3.

The CPU 1 executes a similarity calculation described later based on the music information input to the RAM 3 and the membership function stored in the ROM 2. Then, in accordance with the result of the similarity calculation, a signal indicating the tonality of the input music information is given to the LCD driver 5, and the tonality of the music is displayed according to a predetermined display pattern on the LCD 7.

Here, the CPU 1 constitutes the similarity calculating means CA described with reference to FIG.
The MS, the MIDI interface 4 and the RAM 3 constitute the music information providing means GS, and the LCD driver 5 and the LCD 7 constitute the output means OU.

Next, a membership function stored in the ROM 2 in advance will be described.

First, the membership function m of the C major (C _Maj )
_CMaj will be described. This membership function m _CMaj
Is obtained by preparing a plurality of music scores, preferably many, of a music piece that is recognized as a C major by the manufacturer side, and extracting the features.

Specifically, for example, a score as shown in FIG. 3 is prepared. This score is a C major song. And 1
A unit time obtained by dividing a bar into 96, for example, is 1t, and a total sum of the number of unit times appearing in the scale C appearing in the music is obtained. Here, this value is set as T (C).
Now, for example, T (C) = 480t.

Similarly, the total appearance time of the scale C ^# appearing in the music is obtained. This value is set as T (C ^# ). Now, for example, T (C ^# ) = 0. Similarly, the sum of the appearance times is calculated for the 12 scales (C and B). For example, this song is as follows. Note that t is omitted.

T (C) = 480 T (C ^# ) = 0 T (D) = 192 T (D ^# ) = 0 T (E) = 336 T (F) = 32 T (F ^# ) = 16 T (G) = 608 T (G ^# ) = 48 T (A) = 848 T (A ^# ) = 48 T (B) = 432 Here, the scale vector is defined as follows.

T _h = ｛T (C), T (C ^# ), T (D), T (D ^# ), T
(E), T (E ^# ), T (F), T (F ^# ), T (G), T
^{(G #), T (A} ), T (A #), T (B)} Thus in this case, T _h = {480,0,192,0,336,32,16,608,48,84
8, 48, 432｝.

Then, normalized so the sum of the squares of the T _h is 1, this is a standard scale vector T _h '.

T _h ′ = ｛T (C) / D, T (C ^# ) / D, T (D) / D, T
(D ^# ) / D, T (E) / D, T (E ^# ) / D, T (F) / D, T
(F ^# ) / D, T (G) / D, T (G ^# ) / D, T (A) / D, T
(A ^# ) / D, T (B) / D｝ where It is. Σ is the sum of the scale α, that is, C and B.

 In the above example,

T _h ′ = ｛0.372,0,0.148,0,0.261,0.025,0.012,0.472,
0.037, 0.659, 0.037, 0.335｝ For other songs, obtain T _h ′ by the procedure described above. For example, the above processing is executed for 100 pieces of music, and T _h '
₀ ~T _{h '99} is obtained. Then, a total sum vector ST of the standard scale vectors for 100 pieces of music is obtained.

_{ST = T h '0 + T} h' 1 + ...... + T h '98 + T h' 99 = {ΣT h 'i (C), ΣT h' i (C #), ......, ΣT h 'i (A # ), {T _h ′ _i (B)}} is the sum for song i and thus for 100 songs.

 As an example, ST is as follows.

ST = ｛52.8,1.1,15.5,0,22.2,2.4,1.2,51.9,3.1,54.
5,3.9,35.2｝ Here, this vector is standardized.

_{ST '= T h' 0 /} D '+ T h' 1 / D '+ ...... + T h' 98 / D '+ T h' 99 / D '= {ΣT h' i (C) / D ', ΣT n' _i (C ^# ) /
_{D ', ......, ΣT h'} i (A #) / D ', ΣT h' i (B) / D '} where And Σ is the sum of the scale α, that is, C and B.

 In the above example, the following values are taken.

ST '= ｛0.516,0.011,0.152,0,0.217,0.023,0.012,0.
508,0.030,0.533,0.038,0.344｝ Each component of this vector ST ′ is defined as membership function m
_CMaj (TH). That is, the membership function m _CMaj
(TH) is a function having the scale TH as a variable, and the above example is as shown in FIG.

In this way, when the membership function m _CMaj (TH) is obtained for the _tonality of C, that is, _CMaj , C ^# _Maj
The membership function of m _C ^# _Maj (TH) = m _CMaj (TH-
1 / 2Z). Here, 1 / 2Z represents a semitone. As will be understood with reference to FIG. 5, for example, C-1 / 2Z = C + 11 / 2Z = B. Therefore, the following relationship is obtained.

m _C ^# _Maj (C) = m _CMaj (B) m _C ^# _Maj (C ^# ) = m _CMaj (C) ... m _C ^# _Maj (B) = m _CMaj (A ^# ) Resulting membership function m _C ^# M _Maj (TH) is the sixth
It looks like the figure.

Similarly, membership functions m _DMaj (TH) to m _BMaj (TH) for D _{Maj to} B _Maj are obtained. This means that the major membership function has been determined based on the 12 root sounds.

Next, the membership function of the minor system, m _Cm (TH) to m _Bm
(TH) is obtained in the same procedure as above, for example, by preparing 100 music scores. _{M Cm} (TH) of this membership function
And shows an example of ^m _C ^# m _(TH) in FIG. 7 and FIG. 8.

For blues songs, that is, for major blues and minor blues songs, for example, select 100 songs each,
_Applying the same processing as above, the respective membership functions m _CMajBS (TH) to m _BMajBS (TH) and m _CmBS (TH) to m _BmBS
(TH). 9 and 10 _show m _CMajBS (TH) and m _CmBS (TH) of the above membership functions, respectively.
And an example.

Thus, the membership functions of major, minor, major blues, and minor blues have been determined, but it goes without saying that the membership functions of keys that require judgment can also be determined in the same manner as above. No.

Then, the membership function obtained in advance is stored in the ROM 2 of FIG. As will be understood from the above description, the membership function for the tonic other than C can be obtained by shifting the designation of the TH of the membership function with the tonic as C. The following four membership functions are stored.

m _CMaj (TH) See FIG. 4, m _Cm (TH) See FIG. 7, m _CMajBS (TH) See FIG. 9, m _CmBS (TH) See FIG.

 Next, the operation of this embodiment will be described.

In this embodiment, when a predetermined number of note inputs are made, key discrimination is started in real time,
The resulting information about the tonality of the music is graphically displayed.

FIG. 11 shows an operation flow of the CPU 1. First, FIG.
At 0, initialization processing is performed. Thereafter, input / output processing 200 for performance information input via the MIDI interface 4 is performed. Next 300, note ON / O
If the presence or absence of FF is detected and new information indicating note ON / OFF is input via the MIDI interface 4, a key determination process of 400 is executed.
Return from 00 to 200.

The key determination process 400 is, specifically, first, at 401, a note
Whether the input is ON or OFF is detected, and when a new input is ON, the process proceeds to 402 to detect the note name (scale) n of the note ON. Then, in 403, the flag register O (n) is set to 1. This flag register O (n) is provided on the RAM3. Subsequently, at 404, the ON time measurement counter T
After clearing (n), the process starts and the measurement of the ON time for the new input scale is started. The ON time measurement counter T (n) is constituted by the CPU 1 and the RAM 3, and automatically performs time counting when a count start is set. Then, the process returns from 404 to 200.

If OFF is detected in 401, the process proceeds to 405, and the note name (scale) n of the note OFF is detected. Follow 4
At 06, the flag register O (n) is set to 0, and 40
At 7 the counting operation of the ON time measurement counter T (n) is stopped, and at 408 this ON time measurement counter T (n)
The content of (n) is added to the total ON time register STI (n). This total ON time register STI (n) has elements for each musical scale and can be considered as a vector.

At 409, it is determined whether or not the number of inputted notes exceeds a predetermined number, that is, it is determined that the register P (OFF) = 20. If P (OFF) <20, the register P is determined at 409.
After incrementing (OFF) by 1, the process returns to 200.

If YES is determined in 409, the process proceeds to 411 and the register P
After decreasing the value of (OFF) by 10, proceed to 412. That is, 40
As can be understood from the operations of 9, 410, and 411, in the initial state, after a note input exceeding 20 is made, a key judgment process is performed, and thereafter, when a new 10 note input is made, the tone is adjusted each time. The determination process is performed.

At 412, the square root F of the sum of squares of each component of the total time register STI (n) in the following equation is calculated.

In this equation, Σ means that a sum is taken for each scale n, that is, for C and B.

In the following 413, the total ON time register STI (n)
Are standardized, that is, each component of the total ON time register STI (n) is calculated as FTI in order to obtain a standardized total ON time register STI '(n) (which can also be considered as a vector). Divide by.

After performing this processing (pre-processing), each membership function stored in ROM2 and the standardized total
The similarity calculation with the ON time register STI '(n) is executed. 414C to 414B is the similarity calculation with the major membership function, 415C to 415B is the similarity calculation with the minor membership function, and 416C to 416B
Are the similarity calculation with the membership function of the major blues system, and 417C to 417B are the calculation of the similarity with the membership function of the minor blues system.

To explain one example, the membership function of C _Maj , M _CMaj
(N) (where n is the same as the above-mentioned TH and indicates each scale C-B) and the standardized total ON time register ST
Calculation of the similarity with each component of I '(n) is as follows.

S _CMaj = {STI '(n) * m _CMaj (n) In this equation, Σ is the sum for each scale n (C-B).

When each operation as shown in FIG. 11 is performed between each membership function and the standardized total ON time register STI '(n), the operation results are S _{CMaj to} S _BMaj and S _Cm to
12 * 4 = 48 values of S _Bm , S _{CMajBS to} S _BMajBS , S _{CmBS to} S _BmBS are obtained.

Then, after this series of processing, the process returns to 200, and the CPU 1
The similarity calculation results S _{CMaj to} S _BMaj , S _{Cm to} S
_Bm, S _CMajBS ~S _BMajBS, sends the S _CmBS ~S _BmBS, drives the LCD 7. FIG. 12 shows one display mode of the LCD 7. In other words, the LCD7 displays four pie charts,
For each main tone of minor, major blues, and minables, 48 calculation results can be shown.
Therefore, the user can visually confirm the key of the currently input musical piece from the display of the four pie charts.

In the example of FIG. 12, F _Maj showing the largest value
Are the most promising candidates, and the next candidates are A ^* _m , C
_Maj, and the like and G _m.

As described above, one embodiment of the present invention has been described. According to this embodiment, it is possible to detect and display the key (tonality) of a music input from a keyboard or the like in real time,
In addition, even if there is a modulation or the like in the middle of the music, the display form is sequentially changed in accordance with the modulation, which is preferable. In this case, with respect to the modulation, the moving-accumulation is performed such that the accumulated value of the ON-time data of each scale stored in the total ON-time register STI '(n) relates only to the latest predetermined number of notes. Then, a better result can be obtained.

Further, in the above embodiment, the probabilities of the respective keys are graphically displayed, so that the relationship between the respective keys can be understood.
It is convenient for learning music.

According to the above-described embodiment, the membership function expressed in semitone units is prepared. However, the membership function may be expressed with a smaller pitch.
For example, FIG. 13 shows a membership function m _CMajBS (TH) of C _MajBS taken in fine pitch units so that similarity can be calculated for bend information. By performing such a process, the accuracy of the determination result can be improved for a specific key such as blues.

Further, in the above embodiment, as described with reference to FIG. 12, for each key, the likelihood that the song is applicable is graphically displayed, but it is not always necessary to adopt such a display form, for example, Alternatively, only the key name determined to have the highest similarity as a result of the similarity calculation may be displayed in characters. Alternatively, a plurality of key candidates and their probabilities may be numerically displayed.

Further, according to the above embodiment, in addition to actually playing the music on the keyboard and inputting the music information, the performance information can be supplied in various forms. For example, voice input from a microphone or automatic performance information coded in advance can be given to the tonality determination device.

Also, by supplying the output of the discrimination result of the tonality discriminating device to the automatic accompaniment device and controlling the key of the automatic accompaniment, it is also possible to execute the optimum automatic accompaniment. In this way, the output of the key discrimination result can be used for various purposes (singer identification, karaoke imitation identification, etc.) as well as simply displaying.

<Effects of the Invention> As described above, according to the present invention, if the appearance frequency (membership function) of each scale note in each key is prepared in advance, the tonality determination of the music can be performed by performing a relatively simple calculation. The tonality to be determined is not limited,
Also, there is no need for complicated algorithms for key determination,
It is valid.

[Brief description of the drawings]

BRIEF DESCRIPTION OF THE DRAWINGS The drawings show the present invention, and FIG.
FIG. 3 is a block diagram of one embodiment, FIG. 3 is a diagram showing a musical score for explaining a membership function, FIG. 4 is a diagram showing a membership function of a C major, and FIG. FIG. 6 shows a membership function of C ^# major, FIG. 7 shows a membership function of C minor, FIG. 8 shows a membership function of C ^# minor FIG. 9, FIG. 9 is a diagram showing a membership function of C major blues, FIG. 10 is a diagram showing a membership function of C minor blues, FIG.
FIG. 12 is a diagram showing a processing flow of U, FIG. 12 is a diagram showing a display form, and FIG. 13 is a diagram showing a state in which a membership function is expressed in intervals smaller than a semitone. GS: music information supply means, MS: membership function supply means, CA: similarity calculation means, OU: output means, 1 ...
... CPU, 2 ... ROM, 3 ... RAM, 4 ... MIDI interface, 5 ... LCD driver, 7 ... LCD.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G10H 1/00-1/00 102 G10H 1/36-1/42 G09B 15/00-15/08 G10G 1 / 00-3/04

Claims (1)

(57) [Claims] 1. A music information supply means for supplying music information comprising a plurality of musical notes; an appearance frequency storage means for storing an appearance frequency of each scale sound for each of a plurality of keys; The note length of the notes in the music information is accumulated for each note name, and for each predetermined number of notes, a predetermined value is determined based on the accumulated value and the appearance frequency of each of the plurality of keys stored in the appearance frequency storage means. Calculating the similarity between the music information up to that point and each of the plurality of keys in real time, and sequentially updating the similarity for each key calculated by the similarity calculating means. A tonality discriminating device comprising: a display means for displaying.