Music notation creating device, training device, music notation creating method and training method

Abstract

A musical score writing device according to the present invention comprises a reception unit and an estimation unit, wherein the reception unit receives a musical note sequence that is composed of a plurality of musical notes. The estimation unit uses a trained model to estimate attribute information and each musical note for writing a musical score. The trained model is a machine learning model that has learned an input-output relation between a reference musical note sequence made of a plurality of reference notes, and reference attribute information and each reference note for writing a musical score. A training device according to the present invention comprises a first acquisition unit, a second acquisition unit, and a construction unit. The first acquisition unit acquires a reference musical note sequence. The second acquisition unit obtains each reference musical note and reference attribute information. The construction unit constructs a trained model.

Description

[Technical field]

[0001] The present invention relates to a musical notation creating device, a training device, a musical notation creating method, and a training method for creating musical scores. [Background technology]

[0002] As a technique for creating musical scores, for example, Patent Document 1 and Patent Document 2 are known. Patent Document 1 describes the generation of musical score display data by analyzing automatic performance data in MIDI (Musical Instrument Digital Interface) format. Patent Document 2 describes the extraction of note characteristics from music data objects such as standard MIDI files, the determination of related note syllables based on the note characteristics, and the generation of visual musical scores according to the note characteristics.

[0003] [Patent Document 1] JP 2005-195827 A [Patent Document 2] Special Publication No. 2018-533076 Summary of the Invention [Problem to be solved by the invention]

[0004] In practical musical scores, not only the notes but also various attribute information of the notes are written. However, the technology of Patent Document 1 or Patent Document 2 cannot estimate the attribute information from the MIDI data. Therefore, it is difficult to create practical musical scores.

[0005] An object of the present invention is to provide a musical score creation device, a training device, a musical score creation method, and a training method that are capable of creating practical musical scores. [Means for solving the problem]

[0006] A musical score creation device according to one aspect of the present invention includes a receiving unit that receives a sequence of musical notes consisting of a plurality of notes, and an estimation unit that uses a trained model to estimate each note and attribute information for creating a musical score, where the trained model is a machine learning model that has mastered the input / output relationship between a reference note sequence consisting of a plurality of reference notes, and each reference note and reference attribute information for creating a musical score.

[0007] A musical score creation device according to another aspect of the present invention includes a receiving unit that accepts an input note token string, which is performance data including information on notes, parts and beats; an estimation unit that converts the image score into a score element token string including note drawing, attribute and bar information, creates a learning note token string from the score element token string, and estimates the score token string from the input note token string using a trained model that has been trained to use the learning note token string as input and score tokens as output; and a creation unit that creates an image score from the score token string.

[0008] A training device according to yet another aspect of the present invention includes a first acquisition unit that acquires a reference note sequence consisting of a plurality of reference notes, a second acquisition unit that acquires each reference note and reference attribute information for creating a musical score, and a construction unit that constructs a trained model that has acquired the input / output relationship between the reference note sequence and each reference note and the reference attribute information.

[0009] A method for creating musical scores according to yet another aspect of the present invention receives a sequence of musical notes consisting of a plurality of notes, and uses a trained model to estimate each note and attribute information for creating the musical score, the trained model being a machine learning model that has mastered the input / output relationship between a reference sequence of musical notes consisting of a plurality of reference notes, and each reference note and reference attribute information for creating the musical score, and is executed by a computer.

[0010] A training method according to yet another aspect of the present invention includes obtaining a reference note sequence consisting of a plurality of reference notes, obtaining each of the reference notes and reference attribute information for creating a musical score, constructing a trained model that has acquired the input / output relationship between the reference note sequence and each of the reference notes and the reference attribute information, and executing the method by a computer. A program according to yet another aspect of the present invention is a program for causing a computer to execute a method for creating musical scores, the program causing the computer to execute a process of accepting a sequence of musical notes consisting of a plurality of notes, and a process of estimating each note and attribute information for creating the musical score using a trained model, the trained model being a machine learning model that has mastered the input / output relationship between a reference sequence of musical notes consisting of a plurality of reference notes, and each reference note and reference attribute information for creating the musical score. A program according to yet another aspect of the present invention is a program for causing a computer to execute a training method, which causes the computer to execute a process of acquiring a reference note sequence consisting of a plurality of reference notes, a process of acquiring each reference note and reference attribute information for creating musical score, and a process of constructing a trained model that has acquired the input / output relationship between the reference note sequence and each reference note and the reference attribute information. Effect of the Invention

[0011] According to the present invention, practical musical scores can be created. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 is a block diagram showing the configuration of a processing system including a musical score creating device and a training device according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram showing an example of a learning note token sequence in each training data. [Diagram 3] FIG. 3 is a piano roll shown by the training note token sequence of FIG. [Figure 4] FIG. 4 is a diagram showing an example of a musical score element token sequence in each training data. [Diagram 5] FIG. 5 shows the musical score represented by the musical score element token string in FIG. [Figure 6] FIG. 6 is a diagram showing other examples of musical score element token strings in each training data. [Figure 7] FIG. 7 is a diagram showing another example of a musical score element token string describing a clef. [Figure 8] FIG. 8 is a diagram showing another example of a musical score element token string describing a clef. [Figure 9] FIG. 9 is a diagram showing an example of a musical score element token string describing a voice part. [Figure 10] FIG. 10 is a block diagram showing the configuration of the training device and the musical score creating device. [Figure 11] FIG. 11 is a diagram showing an example of a visual musical score. [Figure 12] FIG. 12 is a flowchart showing an example of training processing by the training device of FIG. [Figure 13] FIG. 13 is a flow chart showing an example of a musical score creation process performed by the musical score creation device of FIG. [Figure 14] FIG. 14 is a diagram for explaining the operation of the reception unit in another embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] (1) Processing system configuration Hereinafter, a musical score creation device, a training device, a musical score creation method, and a training method according to an embodiment of the present invention will be described in detail with reference to the drawings. Fig. 1 is a block diagram showing the configuration of a processing system including a musical score creation device and a training device according to an embodiment of the present invention. As shown in Fig. 1, the processing system 100 includes a RAM (random access memory) 110, a ROM (read only memory) 120, a CPU (central processing unit) 130, a storage unit 140, an operation unit 150, and a display unit 160.

[0014] The processing system 100 is realized by a computer such as a personal computer, a tablet terminal, or a smartphone. Alternatively, the processing system 100 may be realized by the cooperative operation of multiple computers connected by a communication path such as Ethernet, or may be realized by an electronic musical instrument with a performance function such as an electronic piano.

[0015] The RAM 110, the ROM 120, the CPU 130, the storage unit 140, the operation unit 150 and the display unit 160 are connected to a bus 170. The RAM 110, the ROM 120 and the CPU 130 constitute the training device 10 and the musical score creation device 20. In this embodiment, the training device 10 and the musical score creation device 20 are constituted by a common processing system 100, but they may also be constituted by separate processing systems.

[0016] The RAM 110 is, for example, a volatile memory, and is used as a working area for the CPU 130. The ROM 120 is, for example, a non-volatile memory, and stores a training program and a music score creation program. The CPU 130 performs training processing by executing the training program stored in the ROM 120 on the RAM 110. The CPU 130 also performs music score creation processing by executing the music score creation program stored in the ROM 120 on the RAM 110. The training processing and the music score creation processing will be described in detail later.

[0017] The training program or the music notation creation program may be stored in the storage unit 140 instead of the ROM 120. Alternatively, the training program or the music notation creation program may be provided in a form stored in a computer-readable storage medium and installed in the ROM 120 or the storage unit 140. Alternatively, when the processing system 100 is connected to a network such as the Internet, the training program or the music notation creation program distributed from a server (including a cloud server) on the network may be installed in the ROM 120 or the storage unit 140.

[0018] The storage unit 140 includes a storage medium such as a hard disk, an optical disk, a magnetic disk, or a memory card, and stores the trained model M and a plurality of training data D. The trained model M or each of the training data D may be stored in a computer-readable storage medium instead of being stored in the storage unit 140. Alternatively, when the processing system 100 is connected to a network, the trained model M or each of the training data D may be stored in a server on the network.

[0019] The trained model M is a machine learning model trained to estimate each note and attribute information for creating a musical score, and is constructed using multiple training data D. The training data D indicates a set of a reference note sequence, each reference note, and each reference attribute information. The reference note sequence is shown as a learning note token sequence consisting of multiple reference notes that can be generated from MIDI, for example. Each reference note and reference attribute information is shown as a musical score element token sequence.

[0020] The training data D may be image data showing an image score shown in FIG. 5, which will be described later. In this case, a training note token string and a score element token string are created from the image score shown by the training data D. The trained model M is constructed by learning the input / output relationship between the training note token string and the score element token string. The training note token string and the score element token string are described in detail below.

[0021] (2) Learning note token sequence In this embodiment, the training note token sequence includes a part and a metrical structure in addition to a reference note sequence. Fig. 2 is a diagram showing an example of a training note token sequence in each training data D. Fig. 3 is a piano roll shown by the training note token sequence A in Fig. 2.

[0022] As shown in FIG. 2, the learning note token string A is described by a number of tokens including tokens A0 to A24, which are basically arranged in chronological order. Each token is a symbolized musical element, and some tokens have attributes. The attributes of a token are described in the latter half of the token (after the underscore). The learning note token string A in FIG. 2 is data excerpted from the first two bars of a song.

[0023] Token A0 indicates a part. As token A0, "R" and "L" indicate the right hand and left hand parts, respectively. In this example, the right hand token string is placed after "R". After that, "L" is placed, and after "L", the left hand token string is placed. "R" and the right hand token string may be placed after the left hand token string. Also, token A0 is placed at the beginning of learning note token string A, that is, before the reference note string (tokens A1 to A24), but may be placed at any position in learning note token string A. If there is no distinction between parts, learning note token string A does not include token A0.

[0024] Tokens A1 to A24 correspond to the reference note sequence. The reference notes in the reference note sequence are indicated by a pair of pitch and value. The pitch is described by the "note" attribute of tokens A1, A3, etc. The value is described by the "len" attribute of tokens A2, A4, etc. In the example of FIG. 2, a reference note with a pitch of "73" and a duration of 36 units is indicated by a pair of tokens A1 and A2, and a reference note with a pitch of "69" and a duration of 36 units is indicated by a pair of tokens A3 and A4. In the piano roll of FIG. 3, the key "C5" corresponds to the pitch "72".

[0025] "bar", "beat", and "pos" are tokens that indicate the metrical structure. In the learning note token sequence A, "bar" separates bars, and "beat" separates beats. The position of the reference note within a beat is described by the "pos" attribute. In the example in Figure 2, one bar is 4 beats. The length of one beat is 12 units.

[0026] Tokens A1 through A12 (6 units long of token A12) indicate the reference note string of the first measure. Therefore, tokens A1 through A12 are divided into measures by the "bar" before token A1 and the "bar" after token A12. Furthermore, the first measure is divided into beats by the three "beats" after token A4. Similarly, the remaining part of token A12 through A24 (6 units long of token A24) indicate the reference note string of the second measure.

[0027] (3) Musical score element token sequence In this embodiment, the musical score element token string includes information on note drawings, attributes, and bars for creating an image musical score. Fig. 4 is a diagram showing an example of a musical score element token string in each training data D. Fig. 5 is a musical score represented by the musical score element token string B in Fig. 4.

[0028] As shown in Fig. 4, the musical score element token string B is described by a plurality of tokens including tokens B1 to B38 basically arranged in chronological order. Like the tokens in the training note token string A, some tokens have attributes. The attributes of a token are described in the latter half of the token. Also, like the training note token string A, the musical score element token string B may include tokens of "R" and "L" that indicate parts.

[0029] In the musical score element token string B, bars are also separated by "bar." In the example of FIG. 4, the range separated by "bar" before token B1 and "bar" after token B15 corresponds to the first bar. Therefore, tokens B1 to B15 correspond to the first bar of the training note token string A in FIG. 2. Similarly, the range separated by "bar" before token B16 and "bar" after token B38 corresponds to the second bar. Therefore, tokens B16 to B38 correspond to the second bar of the training note token string A.

[0030] In musical score element token string B, the reference note in the reference note string is also indicated by a pair of pitch and note value. The pitch is described by the "note" attribute, and the note value is described by the "len" attribute. In learning note token string A, "len_12" corresponds to one beat, while in musical score element token string B, "len_1" corresponds to one beat. The direction of the stem of the reference note is described by the "stem" attribute. If the "stem" attribute is "down", the stem is drawn extending downward from the note head. On the other hand, if the "stem" attribute is "up", the stem is drawn extending upward from the note head.

[0031] In the example of Fig. 4, tokens B3-B6 indicate reference note N1 in Fig. 5, tokens B7-B10 indicate reference note N2, tokens B11-B14 indicate reference note N3, and tokens B16-B19 indicate reference note N4. Tokens B21-B24 indicate reference note N5, tokens B26-B29 indicate reference note N6, tokens B30-B33 indicate reference note N7, and tokens B34-B37 indicate reference note N8. In tokens B9, B13, etc., the attribute "len" is written as a fraction such as 1 / 2, but may also be written as a decimal such as 0.5.

[0032] A reference rest in a reference note sequence is described by the "rest" token. The duration of the reference rest is described by the "len" attribute, just like the reference note. Using the "beam" token, multiple reference notes, such as eighth notes or sixteenth notes, can be beamed together. The start and end positions of the beam are described by the "start" and "stop" attributes of "beam", respectively.

[0033] 6 is a diagram showing another example of the score element token string B in each training data D. The upper part of FIG. 6 shows a part of the score element token string B, and the lower part shows an image score corresponding to the score element token string B in the upper part. The same is true in FIG. 7 to FIG. 9 described later. Tokens B7 to B14 in the score element token string B in FIG. 6 are the same as tokens B7 to B14 in the score element token string B in FIG. 4.

[0034] As shown in Fig. 6, "beam_start" is placed before the token B7, and "beam_stop" is placed after the token B14. That is, the tokens B7 to B10 corresponding to the reference note N2 and the tokens B11 to B14 corresponding to the reference note N3 are sandwiched between "beam_start" and "beam_stop". As a result, as shown by the dashed line in Fig. 6, the reference note N2 and the reference note N3 are connected by a beam in the image score.

[0035] (4) Reference attribute information In addition to the tokens for drawing notes and rests, the musical score element token string B includes tokens describing key signatures, division and joining of note values, and clefs or voices as reference attribute information. Specific examples of reference attribute information in the musical score element token string B are explained below. For an explanation of the musical score element token string B describing key signatures, division and joining of note values, and clefs, please refer to Figures 4 and 5.

[0036] As shown by token B2 in Fig. 4, the key signature is described by the token "key". The type of key signature is described by an attribute of "key". For example, sharps and naturals are described by the attributes "sharp" and "natural" of "key", respectively. Furthermore, the number of key signatures is described by further attributes of "key". Thus, the three sharps enclosed by the dashed line in Fig. 5 are described by token B2. Tokens describing key signatures appear at the beginning of a line and at key signature change positions in the image score.

[0037] Splitting and joining of note values ​​is indicated by a performance symbol tie surrounded by a two-dot chain line in Figure 5. A performance symbol tie is described by a "tie" token, as shown by tokens B15, B20, B25, and B38 in Figure 4. The start and end positions of a performance symbol tie are described by the attributes "start" and "stop" of "tie", respectively.

[0038] As shown by token B1 in Figure 4, a clef is described by the token "clef". The type of clef is described by the attributes of "clef". For example, the treble clef and bass clef are described by the attributes "treble" and "bass" of "clef", respectively. Therefore, the treble clef is described by token B1 as clef C in Figure 5. Tokens describing clefs appear at the beginning of a line and at the position where the clef changes in the image score.

[0039] 7 and 8 are diagrams showing other examples of musical score element token strings B describing cleves. The octave line one octave above, surrounded by a dashed line in FIG. 7, is described by the "8va" token. The octave line one octave below, surrounded by a dashed line in FIG. 8, is described by the "8vb" token. The start and end positions of the octave line are described by the attributes "start" and "stop" of "8va" or "8vb", respectively.

[0040] Fig. 9 is a diagram showing an example of a musical score element token string B describing a voice part. The start position and end position of one voice part enclosed by a dashed line in Fig. 9 are described by a pair of "voice" and " / voice" tokens. The start position and end position of the other voice part enclosed by a two-dot dashed line in Fig. 9 are described by another pair of "voice" and " / voice" tokens placed after the above pair of "voice" and " / voice".

[0041] (5) Training equipment Fig. 10 is a block diagram showing the configuration of the training device 10 and the musical score creation device 20. As shown in Fig. 10, the training device 10 includes, as functional units, a first acquisition unit 11, a second acquisition unit 12, and a construction unit 13. The functional units of the training device 10 are realized by the CPU 130 in Fig. 1 executing a training program. At least a part of the functional units of the training device 10 may be realized by hardware such as electronic circuits.

[0042] The first acquisition unit 11 acquires a learning note token string A including a reference note string, a part, and a metrical structure based on each training data D stored in the memory unit 140 etc. In this example, the learning note token string A is acquired by extracting a part of a token string from a musical score element token string B acquired by the second acquisition unit 12 described later.

[0043] The second acquisition unit 12 acquires a musical score element token string B including information on note drawings, attributes, and bars based on each training data D stored in the storage unit 140 or the like. In this example, the image score is analyzed to extract the note drawings, attributes, and bars included in the image score in chronological order. In addition, each of the note drawings, attributes, and bars extracted in chronological order is converted into a token according to a predetermined conversion table. In this way, the musical score element token string B is acquired.

[0044] The construction unit 13 causes the machine learning model to perform learning for each training data D, using the learning note token string A acquired by the first acquisition unit 11 as input and the musical score element token string B acquired by the second acquisition unit 12 as output. By repeating machine learning for multiple training data D, the construction unit 13 constructs a trained model M that shows the input / output relationship between the learning note token string A and the musical score element token string B.

[0045] In this example, the construction unit 13 constructs the trained model M by training a Transformer, but the embodiment is not limited to this. The construction unit 13 may construct the trained model M by training another type of machine learning model that handles time series. The trained model M constructed by the construction unit 13 is stored in, for example, the storage unit 140. The trained model M constructed by the construction unit 13 may be stored in a server on a network, etc.

[0046] The musical score creation device 20 includes, as functional units, a receiving unit 21, an estimating unit 22, a first determining unit 23, a second determining unit 24, and a generating unit 25. The functional units of the musical score creation device 20 are realized by the CPU 130 in Fig. 1 executing a musical score creation program. At least a part of the functional units of the musical score creation device 20 may be realized by hardware such as electronic circuits. Furthermore, the musical score creation device 20 may be incorporated into musical score engraving software or a digital audio workstation (DAW).

[0047] The receiving unit 21 receives an input note token string including a note string consisting of a plurality of notes. A user can generate an input note token string by operating the operation unit 150 and provide the input note token string to the receiving unit 21. The input note token string has the same structure as the learning note token string A in Fig. 2. In other words, the input note token string has a part and a metrical structure in addition to a note string.

[0048] The estimation unit 22 estimates a score token string including notes and attribute information for creating a musical score from the input note token string, using a trained model M stored in the storage unit 140 or the like. The score token string indicates a token string corresponding to the input note token string accepted by the acceptance unit 21, and is estimated based on the note string, part, and metrical structure. Since the input note token string has a similar structure to the learning note token string A, the score token string has a similar structure to the score element token string B.

[0049] The first determination unit 23 determines an accidental based on the score token sequence estimated by the estimation unit 22. The accidental is determined, for example, from the key signature and pitch in the score token sequence. The accidental of the preceding note may be further used to determine the subsequent accidental. The second determination unit 24 determines a time signature based on the score token sequence estimated by the estimation unit 22. The time signature is determined, for example, from the number of beats in each bar in the score token sequence.

[0050] The generating unit 25 generates score information indicating a score in which each note and attribute information is written from the score token sequence estimated by the estimating unit 22. That is, the generating unit 25 functions as a creating unit and generates score information in a score format from the score token sequence. The score information may be text data in MusicXML format or the like. The display unit 160 displays an image score indicated by the score information generated by the generating unit 25.

[0051] Fig. 11 is a diagram showing an example of a visual score. As shown in Fig. 11, the visual score may further include an accidental X determined by the first determination unit 23. The visual score may further include a time signature Y determined by the second determination unit 24. Here, as long as there is no change in the time signature, the time signature Y may be written only at the beginning of the score.

[0052] (6) Training and music notation processing Fig. 12 is a flow chart showing an example of training processing by the training device 10 of Fig. 10. The training processing of Fig. 12 is performed by the CPU 130 of Fig. 1 executing a training program. First, the second acquisition unit 12 acquires a musical score element token string B from each training data D (step S1). The first acquisition unit 11 acquires a learning note token string A corresponding to the musical score element token string B from the musical score element token string B acquired in step S1 (step S2).

[0053] Next, the construction unit 13 performs machine learning for each training data D, using the musical score element token string B acquired in step S1 as an output token and the learning note token string A acquired in step S2 as an input token (step S3). Next, the construction unit 13 determines whether or not sufficient machine learning has been performed (step S4). If the machine learning is insufficient, the construction unit 13 returns to step S3. Steps S3 and S4 are repeated while changing the parameters until sufficient machine learning has been performed. The number of times the machine learning is repeated varies depending on the quality conditions that the trained model M to be constructed should satisfy.

[0054] When sufficient machine learning has been performed, the construction unit 13 saves the input / output relationship between the learning note token string A and the musical score element token string B acquired by the machine learning in step S3 as a trained model M (step S5). This ends the training process.

[0055] Fig. 13 is a flow chart showing an example of a score creation process by the musical score creation device 20 of Fig. 10. The musical score creation process of Fig. 13 is performed by the CPU 130 of Fig. 1 executing a musical score creation program. First, the reception unit 21 receives an input note token string (step S11). Next, the estimation unit 22 estimates a musical score token string from the input note token string received in step S11, using the trained model M stored in step S5 of the training process (step S12).

[0056] Next, the first determination unit 23 determines an accidental based on the score token sequence estimated in step S12 (step S13). The second determination unit 24 determines a time signature based on the score token sequence estimated in step S12 (step S14). Steps S13 and S14 may be executed either first or simultaneously.

[0057] Then, the generating unit 25 generates score information based on the score token sequence estimated in step S12, the accidentals determined in step S13, and the time signature determined in step S14 (step S15). Based on the generated score information, a visual score may be displayed on the display unit 160. This completes the score creation process.

[0058] (7) Effects of the embodiment As described above, the musical score creation device 20 according to this embodiment includes a receiving unit 21 that receives a sequence of musical notes consisting of a plurality of musical notes, and an estimating unit 22 that estimates each note and attribute information for creating a musical score by using a trained model M. The trained model M is a machine learning model that has learned the input / output relationship between a reference sequence of musical notes consisting of a plurality of reference notes, and each reference note and reference attribute information for creating a musical score.

[0059] According to this configuration, since each note and attribute information corresponding to a sequence of notes is estimated using the trained model M, it becomes possible to write not only the notes but also the attribute information on the musical score, thereby making it possible to create a practical musical score.

[0060] The musical score creation device 20 may further include a generating unit 25 that generates musical score information indicating a musical score in which the estimated notes and attribute information are described. In this case, the user does not need to generate musical score information from the notes and attribute information, thereby improving usability.

[0061] In other words, the musical score creation device 20 in this embodiment may include a receiving unit 21 that accepts an input note token string, which is performance data including information on notes, parts, and beats, an estimation unit 22 that converts the image score into a score element token string including note drawing, attribute, and bar information, creates a learning note token string from the score element token string, and estimates the score token string from the input note token string using a trained model M that has been trained to input the learning note token string and output score tokens, and a creation unit that creates an image score from the score token string.

[0062] The estimation unit 22 may estimate a key signature as the attribute information. The estimation unit 22 may estimate division and joining of note values ​​as the attribute information. The estimation unit 22 may estimate a clef as the attribute information. The estimation unit 22 may estimate a voice part as the attribute information. The musical score creation device 20 may further include a first determination unit 23 that determines an accidental based on the estimated notes and attribute information. The musical score creation device 20 may further include a second determination unit 24 that determines a time signature based on the estimated notes and attribute information. In these cases, a more practical musical score can be created.

[0063] The training device 10 according to this embodiment includes a first acquisition unit 11 that acquires a reference note sequence consisting of a plurality of reference notes, a second acquisition unit 12 that acquires each reference note and reference attribute information for creating a musical score, and a construction unit 13 that constructs a trained model M that has learned the input / output relationships between the reference note sequence, each reference note, and the reference attribute information. With this configuration, it is possible to easily construct a trained model M that has learned the input / output relationships between the reference note sequence, each reference note, and the reference attribute information.

[0064] (8) Other embodiments In the above embodiment, the learning note token string A includes a part and a metrical structure, but the embodiment is not limited to this. The learning note token string A only needs to include a reference note string, and does not need to include a part and a metrical structure. The same applies to the input note token string. Also, the score element token string B includes bar information, but the embodiment is not limited to this. The score element token string B only needs to include reference notes and reference attribute information, and does not need to include bar information. The same applies to the score token string.

[0065] In the above embodiment, the musical score creation device 20 includes the generating unit 25, but the embodiment is not limited to this. A user can create a musical score based on the musical score token sequence estimated by the estimating unit 22. Therefore, the musical score creation device 20 does not need to include the generating unit 25.

[0066] In the above embodiment, the musical score creation device 20 includes the first judgment unit 23 and the second judgment unit 24, but the embodiment is not limited to this. If accidentals do not need to be written in the musical score, the musical score creation device 20 does not need to include the first judgment unit 23. If time signatures do not need to be written in the musical score, the musical score creation device 20 does not need to include the second judgment unit 24.

[0067] In the above embodiment, the user generates an input note token string by operating the operation unit 150 and provides it to the receiving unit 21, but the embodiment is not limited to this. Fig. 14 is a diagram for explaining the operation of the receiving unit 21 in another embodiment. As shown in the upper part of Fig. 14, the user may provide waveform data generated by playing the piano or the like to the receiving unit 21.

[0068] In this case, as shown in the lower part of Fig. 14, the receiving unit 21 converts the given waveform data into MIDI data and obtains the input note token string from the converted MIDI data. Therefore, the receiving unit 21 receives the input note token string in the form of waveform data. With this configuration, it is possible to generate a musical score that describes a performance from the waveform data of the performance.

[0069] In the above embodiment, the receiving unit 21 may receive an input note token string in which right-hand part tokens and left-hand part tokens are mixed. Even in this case, by using an appropriately trained trained model M, it is possible to estimate a musical score token string in which right-hand part tokens and left-hand part tokens are separated.

Claims

1. A reception unit that receives a musical note string consisting of a plurality of musical notes; An estimation unit that estimates each note and attribute information for creating a musical score using the trained model, The trained model is a machine learning model that has acquired the input / output relationship between a reference note sequence consisting of a plurality of reference notes and each reference note and reference attribute information for creating musical score, the musical score creation device.

2. The musical score creation device according to claim 1 , further comprising a generating unit that generates musical score information indicating a musical score on which the estimated notes and the attribute information are written.

3. The musical score creation device according to claim 1 , wherein the estimation unit estimates a key signature as the attribute information.

4. The musical score creation device according to claim 1 , wherein the estimation unit estimates division and joining of note values ​​as the attribute information.

5. 5. The musical score creation device according to claim 1, wherein the estimation unit estimates a clef as the attribute information.

6. The musical score creation device according to claim 1 , wherein the estimation unit estimates a voice part as the attribute information.

7. 7. The musical score creation device according to claim 1, further comprising a first determination unit that determines accidentals based on the estimated notes and the attribute information.

8. The musical score creation device according to any one of claims 1 to 7, further comprising a second determination unit that determines a time signature based on the estimated notes and the attribute information.

9. a reception unit for receiving an input note token sequence, which is data for performance including information on notes, parts, and beats; an estimation unit that converts the image score into a score element token sequence including note drawing, attribute, and measure information, creates a training note token sequence from the score element token sequence, and estimates the score token sequence from the input note token sequence using a trained model that has been trained to input the training note token sequence and output score tokens; and a creation unit that creates an image score from the score token string.

10. A first acquisition unit that acquires a reference note sequence including a plurality of reference notes; A second acquisition unit that acquires each reference note and reference attribute information for creating a musical score; A training device comprising: a construction unit that constructs a trained model that has acquired the input / output relationship between the reference note sequence and each of the reference notes and the reference attribute information.

11. Accepts a sequence of notes, Using the trained model, estimate each note and its attribute information to create a musical score, The trained model is a machine learning model that has learned the input / output relationship between a reference note sequence consisting of a plurality of reference notes and each reference note and reference attribute information for creating a musical score. A computer implemented method for creating musical scores.

12. Obtaining a reference note sequence consisting of a plurality of reference notes; Obtaining each reference note and reference attribute information for creating a musical score; constructing a trained model that has learned the input / output relationship between the reference sequence of notes and each of the reference notes and the reference attribute information; A computer-implemented training method.

13. A program for causing a computer to execute a method for creating musical scores, comprising: receiving a sequence of notes comprising a plurality of notes; A process of estimating each note and attribute information for creating a musical score using the trained model. Executing the computer, The trained model is a machine learning model that has acquired the input / output relationship between a reference note sequence consisting of a plurality of reference notes and each reference note and reference attribute information for creating musical score, said program.

14. A program for causing a computer to execute a training method, comprising: A process of obtaining a reference note sequence consisting of a plurality of reference notes; A process of obtaining each reference note and reference attribute information for creating a musical score; A process of constructing a trained model that has learned the input / output relationship between the reference note sequence, each of the reference notes, and the reference attribute information; A program to be executed by the computer.

Images