JP2015148663A - Musical composition processing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To set natural variation amount fully consistent with the actual trends for each note of a musical piece.SOLUTION: The variation amount at the time the pronunciation of the note is set in the variation amount setting unit, depending on the random number λ generated in the probability distribution D of dispersion σ corresponding to the position of each note in the target music. For example, in a variation amount setting unit, the variation amount is set in accordance with the random number λ generated in the probability distribution D of dispersion σ set to a numeric value d1 when the time of the pronunciation of the note is located on the inner side of the beat point range R including the beat point B of the music and the variation amount is set in accordance with the random number λ generated in the probability distribution D of dispersion σ set to a numeric value d2 exceeding the numerical value d1 when the time of the pronunciation of the note is located on the outer side of the beat point range R.

Description

  The present invention relates to a technique for controlling each note of a music piece.

  In actual singing and performance scenes, the time point at which each musical note is pronounced varies slightly from the exact position on the score. From the viewpoint of generating perceptually natural sound in consideration of the above tendency, for example, Patent Document 1 discloses a technique for changing the time of pronunciation of each note of a musical piece from a predetermined position on a musical score back and forth. ing.

JP 2009-258291 A

  By the way, in actual singing and performance, the degree to which the sounding time of each note fluctuates tends to be different depending on the position of the note in the music. For example, notes that are pronounced near the beat point of a song are easier for singers and performers to know when to pronounce, so the degree of variation at the point of pronunciation is suppressed compared to notes that are separated from the beat point. There is a tendency to be. Therefore, in a configuration in which the amount of variation at the time of pronunciation of each note is set regardless of the position of each note in the music, the variation at the time of pronunciation of each note may deviate from the actual tendency. In the above description, for the sake of convenience, attention has been paid to the time point at which each note is pronounced. However, the same situation applies to the time point at which each note is muted and various music information (for example, characteristics of vibrato). In view of the above circumstances, an object of the present invention is to set a natural fluctuation amount that matches an actual tendency for each note of a music piece.

  In order to solve the above-described problems, the music processing device of the present invention, for each note of the music, according to the random number generated in the probability distribution of the degree of spread according to the position of the note in the music, A fluctuation amount setting means for setting the fluctuation amount of. In the above configuration, since the amount of variation at the time of sound generation of the note is set according to the random number generated in the probability distribution of the spread degree according to the position of the note in the music, the sound is generated regardless of the position of each note in the music. Compared to a configuration that sets the amount of variation at the time (for example, Patent Document 1), it reflects the relationship between the amount of variation (error) at the time of pronunciation of each note in actual singing and performance and the position of each note in the song. There is an advantage that a natural variation amount can be set.

  In a preferred aspect of the present invention, the fluctuation amount setting means is a random number generated in the probability distribution of the first distribution degree when the time of pronunciation of one note is located inside the beat range including the beat of the music. The variation amount at the time of sound production of one note is set according to the above, and when the time of sound production of one note is located outside the beat range, it occurs with a probability distribution of the second spread degree exceeding the first spread degree. Sets the amount of fluctuation at the time of the sound generation of one note according to the random number. In the above aspect, since the degree of distribution of the probability distribution is changed depending on whether or not the note generation time is included in the beat range, depending on the distance between the beat point of the music and the sound generation time of each note. It is possible to set a natural variation amount that reflects the tendency that the variation amount (error) at the time of pronunciation varies. In addition, since the spread degree is set in a binary manner, there is an advantage that the process of setting the variation amount is simplified as compared with, for example, a configuration in which the spread degree is continuously changed.

  In a preferred aspect of the present invention, the fluctuation amount setting means sets the fluctuation amount at the time of sounding a note in accordance with a random number generated in a probability distribution that maximizes the probability of random number generation corresponding to the variable correction value. In the above aspect, since the random number (for example, the center of the probability distribution) having the maximum occurrence probability in the probability distribution is variably set according to the correction value, the leading / delaying (previous / rear) of the time point at which each note is generated It is possible to set a natural amount of fluctuation that reflects the tendency of (slipping). In addition, the specific example of the above aspect is later mentioned as 2nd Embodiment, for example.

  In a preferred aspect of the present invention, the fluctuation amount setting means is such that the sounding time point of one note after adjustment according to the fluctuation amount is positioned forward compared to the sounding time point of one note, and one note The amount of fluctuation at the time of sounding of one note is set so that it is located behind the time of sounding of the other notes immediately before. In the above aspect, the range of the variation amount of each note is limited so that the sounding point of the note after adjustment according to the variation amount is located between the sounding point of the note and the sounding point of the immediately preceding note. The Therefore, there is an advantage that it is possible to prevent the disappearance of the corresponding note or the immediately preceding note (destruction of the music content) due to the fluctuation of the time of pronunciation of each note. In addition, the specific example of the above aspect is later mentioned as 3rd Embodiment, for example.

  In a preferred aspect of the present invention, the fluctuation amount setting means sets the second note when the sound generation period of the first note of the music and the sound generation period of the second note immediately after are continuous on the time axis. The amount of fluctuation at the time of muting of the first note is set according to the amount of fluctuation at the time of pronunciation. In the above aspect, since the fluctuation amount at the time of muting of the first note is set according to the fluctuation amount of the second note, the first note and the second note are continuous even after adjustment by applying each fluctuation amount. There is an advantage that the state is maintained. In addition, the specific example of the above aspect is later mentioned as 4th Embodiment, for example.

  In a preferred aspect of the present invention, the fluctuation amount setting means individually sets the fluctuation amount at the time of pronunciation of each note in each of a plurality of different voice parts of the music for each voice part. In the above aspect, the fluctuation amount at the time of pronunciation of each note is set for each of the plurality of voice parts of the music. Therefore, it is possible to disperse the time of pronunciation of each note for each voice part on the time axis. In addition, the difference of the note sequence (melody) corresponding to each voice part of a music is unquestioned.

  Each of the above aspects can also be applied to setting the amount of variation of various music information set for a note. The music processing device according to a preferred aspect of the present invention, for each musical note of the music, the amount of change in the music information of the musical note according to the random number generated in the probability distribution of the spread degree according to the position of the musical note in the musical piece. Fluctuation amount setting means for setting is provided.

  The music processing apparatus according to each of the above aspects is realized by hardware (electronic circuit) such as DSP (Digital Signal Processor), and cooperation between a general-purpose arithmetic processing apparatus such as CPU (Central Processing Unit) and the program. It is also realized by. The program of the present invention can be provided in a form stored in a computer-readable recording medium and installed in the computer. The recording medium is, for example, a non-transitory recording medium, and an optical recording medium (optical disk) such as a CD-ROM is a good example, but a known arbitrary one such as a semiconductor recording medium or a magnetic recording medium This type of recording medium can be included. The program of the present invention can be provided, for example, in the form of distribution via a communication network and installed in a computer. The present invention is also specified as an operation method (music processing method) of the music processing device according to each aspect described above.

It is a block diagram of the speech synthesizer in 1st Embodiment. It is a schematic diagram of music data. It is a schematic diagram of an edit screen. It is a flowchart of a variation | change_quantity setting process. It is explanatory drawing of a variation | change_quantity setting process. It is a flowchart of the variation | change_quantity setting process in 2nd Embodiment. It is explanatory drawing of the variation | change_quantity setting process in 2nd Embodiment. It is explanatory drawing of the variation | change_quantity setting process in 3rd Embodiment. It is explanatory drawing of the variation | change_quantity setting process in 4th Embodiment. It is explanatory drawing of the variation | change_quantity setting process in 4th Embodiment. It is a flowchart of the process which sets the variation | change_quantity at the time of mute of each note in 4th Embodiment.

<First Embodiment>
FIG. 1 is a configuration diagram of a speech synthesizer 100 according to the first embodiment of the present invention. The speech synthesizer 100 is a signal processing device that generates an acoustic signal V of a singing voice of an arbitrary song (hereinafter referred to as “target song”), and includes an arithmetic processing device 10, a storage device 12, a display device 14, and an input device 16. This is realized by a computer system (for example, an information processing device such as a portable information terminal or a personal computer) including the sound emitting device 18.

  The display device 14 (for example, a liquid crystal display panel) displays an image instructed from the arithmetic processing device 10. The input device 16 is an operation device operated by a user for various instructions to the speech synthesizer 100, and includes a plurality of operators operated by the user, for example. A touch panel configured integrally with the display device 14 may be employed as the input device 16. The sound emitting device 18 (for example, a speaker or headphones) reproduces sound corresponding to the sound signal V. The D / A converter that converts the acoustic signal V from digital to analog is not shown for convenience.

  The storage device 12 stores a program executed by the arithmetic processing device 10 and various data used by the arithmetic processing device 10. A known recording medium such as a semiconductor recording medium or a magnetic recording medium or a combination of a plurality of types of recording media is arbitrarily employed as the storage device 12. The storage device 12 of the first embodiment stores a plurality of speech element groups L and music data M of the target music.

  Each speech unit group L is a set (speech library) of a plurality of speech units collected in advance from the uttered sound of a specific speaker. The acoustic characteristics of each speech unit are different for each speech unit group L. Specifically, the speaker of each speech unit is different for each speech unit group L. Each speech element is a phoneme (for example, a vowel or a consonant) that is a minimum unit of linguistic meaning distinction, or a phoneme chain (for example, a diphone or a triphone) that connects a plurality of phonemes.

  The music data M is data for designating a time series of a plurality of notes constituting the target music. As illustrated in FIG. 2, the music data M includes a plurality of music data MP corresponding to different voice parts (parts) of the target music. Consists of including. The music data MP of any one voice part specifies a time series of notes of the voice part of the target music. In the first embodiment, a case (unison) is assumed in which an acoustic signal V of a choral voice sung by a plurality of speakers is sung on a common melody of the target music piece. That is, the note strings (melody) specified by the plurality of music data MP of the music data M are common, and a separate speech segment group L for each voice part is used as a material for speech synthesis. However, the note sequence designated by the music data MP can be made different for each voice part. Further, by using a common speech segment group L for a plurality of voice parts corresponding to each music data MP, it is possible to give a natural thickness to the synthesized speech (doubling effect).

  As illustrated in FIG. 2, the music data MP corresponding to any one voice part specifies the note information QN and the control information QC for each note of the voice part. The note information QN is information for specifying a note, and includes a pitch X1, a pronunciation period X2, and a voice code X3 in the first embodiment. The pitch X1 is a note number based on, for example, the MIDI (Musical Instrument Digital Interface) standard. The sound generation period X2 is a period during which the sound of a note is maintained, and is specified, for example, at the time of sound generation of a note (at the time of sound generation of the note) and the time of mute (at the time of sound generation of the note). Note that the sound generation period X2 can also be specified by the time of sound generation and the duration (tone value). The voice code X3 specifies the pronunciation content of the synthesized voice (that is, the lyrics of the target song).

  On the other hand, the control information QC is information for controlling the musical characteristics of the synthesized speech (for example, the tendency or expression of the speaker's singing). The control information QC of the first embodiment includes a fluctuation amount Δ at the time of note generation. The fluctuation amount Δ is a temporal error (variation) with respect to the sounding point of the sounding period X2 designated by the note information QN of each note (that is, the expected sounding point specified on the score). The fluctuation amount Δ can be set to a positive number or a negative number. For example, when the fluctuation amount Δ is a positive number, the sounding time point changes backward on the time axis, and when the fluctuation amount Δ is a negative number, the sounding time point changes forward on the time axis.

  The arithmetic processing device 10 in FIG. 1 executes a program stored in the storage device 12, thereby performing a plurality of functions (display control unit 22, information management unit 24) for editing the music data M and generating the acoustic signal V. , The voice synthesis unit 26, and the fluctuation amount setting unit 28). A configuration in which each function of the arithmetic processing device 10 is distributed to a plurality of devices, or a configuration in which a dedicated electronic circuit (for example, DSP) realizes a part of the functions of the arithmetic processing device 10 may be employed. The display control unit 22 and the information management unit 24 are realized by music editing software (editor), and the voice synthesis unit 26 is realized by voice synthesis software (voice synthesis engine). The fluctuation amount setting unit 28 is realized by, for example, plug-in software for music editing software or voice synthesis software. However, the separation of software corresponding to each function is arbitrary, and for example, the function of the fluctuation amount setting unit 28 can be included as one function of the music editing software.

  The display control unit 22 displays various images on the display device 14. The display control unit 22 of the first embodiment causes the display device 14 to display the editing screen 40 of FIG. 3 for the user to confirm and edit the content of the target music specified by the music data M. The edit screen 40 is a piano roll type in which a note image 44 representing each note designated by each music data MP of the music data M is arranged in a score area 42 in which a time axis and a pitch axis intersect with each other are set. It is an image. The position of the note image 44 in the direction of the pitch axis is set according to the pitch X1 of the note, and the position and display length of the note image 44 in the direction of the time axis are set according to the sound generation period X2 of the note. . Specifically, as illustrated in FIG. 3, the position of the leading edge (left end) of the note image 44 is obtained by adding a fluctuation amount Δ of the control information QC to the sounding point of the sounding period X2 specified by the note information QN. The position of the trailing edge (right end) of the note image 44 is set in accordance with the mute time point specified by the note information QN. Further, the character of the voice code X3 of each note is added to each note image 44. In the first embodiment, note images 44 of a plurality of voice parts are displayed on the editing screen 40. A configuration in which the display form (color, shape, etc.) of the note image 44 is different for each voice part is preferable so that the user can visually distinguish each note image 44 for each voice part.

  The information management unit 24 in FIG. 1 manages the music data M stored in the storage device 12. Specifically, the information management unit 24 updates each music data MP (pitch X1, sound generation period X2, voice code X3) of the music data M in accordance with an instruction from the user to the input device 16. The information management unit 24 of the first embodiment duplicates the music data MP that represents the time series of the notes that the user has instructed for one voice part, so that a plurality of speech unit groups L corresponding to different speech unit groups L can be obtained. The music data MP (music data M) of the voice part is generated.

  The speech synthesizer 26 generates an acoustic signal V by speech synthesis processing using each speech element group L and music data M stored in the storage device 12. The voice synthesizer 26 of the first embodiment generates a basic signal representing the singing voice specified by each music data MP of the music data M of the target music for each voice of the target music, and the basic signals of a plurality of voices. Is mixed to generate an acoustic signal V. Specifically, the speech synthesizer 26 sequentially selects speech units corresponding to the speech code X3 designated for each note by the music data MP of each voice unit from the speech unit group L of the voice unit, The voice segment is adjusted to the pitch X1, the time length is adjusted according to the pronunciation period X2 and the fluctuation amount Δ, and the individual voice segments are connected to each other on the time axis for each voice part. Generate a base signal. The sound generation time for each note of each voice is adjusted according to the variation Δ specified by the control information QC of the note from the sound generation time of the sound generation period X2 specified by the note information QN of the note. Specifically, the sound generation point designated by the note information QN is changed by the variation Δ. The sound signal V generated by the speech synthesizer 26 is supplied to the sound emitting device 18 so that the chorus sound of the target music is reproduced.

  The fluctuation amount setting unit 28 in FIG. 1 sets a fluctuation amount Δ (control information QC) at the time of sound generation for each note of each voice part of the target music piece. Specifically, the variation Δ is set for each note according to the random number λ. The variation setting unit 28 of the first embodiment sets the variation Δ according to the random number λ generated in the probability distribution D of the spread degree (variance) set variably according to the position of the note in the target music. To do. Focusing on the actual singing, the note whose pronunciation point is located near the beat point of the target music is less affected by the fluctuation of the pronunciation point compared to the note whose pronunciation point is far from the beat point (the error at the point of pronunciation is A tendency to be reduced) is observed. In consideration of the above-mentioned tendency, in the first embodiment, the distribution width at the time of sound generation of the notes located near the beat point of the target music is less than the distribution width at the time of sound generation of the notes apart from the beat point (sound generation time point). The fluctuation amount Δ at the time of pronunciation of each note is set so that the dispersion of the sound is suppressed.

  FIG. 4 is a flowchart of processing (variation amount setting processing) in which the variation amount setting unit 28 sets the variation amount Δ, and FIG. 5 is an explanatory diagram of a specific example of variation amount setting processing. In the variation amount setting process of FIG. 4, an instruction from the user (for setting variation amount Δ) to the input device 16 in a state where the music data M of the target music is generated according to the instruction from the user to the editing screen 40. (Execution instruction for plug-in software). At the stage where the fluctuation amount setting unit 28 is started, the fluctuation amount Δ of each note is set to an initial value (for example, zero). In FIG. 5, the note N1 and the note N2 of the target music are shown in the same format as the editing screen 40 described above, and the beat point B of the target music is shown together for convenience.

  When the variation setting process is started, the variation setting unit 28 selects one piece of music data MP (voice part) from the music data M stored in the storage device 12 (SA1), and each note of the music data MP. One note (hereinafter referred to as “selected note”) designated by the information QN is selected (SA2).

  The fluctuation amount setting unit 28 determines whether or not the sounding point of the selected note specified by the note information QN is located inside the beat range R of the target music (SA3). As illustrated in FIG. 5, the beat point range R is a range on the time axis including the beat point B of the target music. Specifically, a range having a predetermined width centered on the beat point B is set as the beat point range R. As illustrated in FIG. 5, the sounding point of the note N1 of the target music is located inside the beat range R (SA3: YES), and the sounding point of the note N2 is located outside the beat range R (SA3: NO). As understood from the above description, the determination in step SA3 is a process for determining whether or not the time point of the selected note is located near the beat point B of the target music (the position of the selected note in the target music is determined). Process).

  As illustrated in FIG. 5, the fluctuation amount setting unit 28 sets the fluctuation amount Δ of the selected note according to the random number λ generated in the probability distribution D (SA4 to SA7). In the first embodiment, a case where a normal distribution with an average of zero is used as the probability distribution D is illustrated. The setting of the variation Δ according to the probability distribution D will be described in detail below. The probability distribution D is not limited to a normal distribution.

  The fluctuation amount setting unit 28 variably sets the variance (dispersion degree) σ of the probability distribution D according to whether or not the sounding point of the selected note designated by the note information QN is located inside the beat range R. . Specifically, when the sounding time of the selected note is located inside the beat point range R (SA3: YES), that is, when the selected note is close to the beat point B, the fluctuation amount setting unit 28 indicates the probability distribution D. The variance σ is set to a numerical value d1 (first spreading degree) (SA4). On the other hand, when the selected note is located outside the beat range R (SA3: NO), that is, when the selected note is separated from the beat point B, the fluctuation amount setting unit 28 sets the variance σ of the probability distribution D to a numerical value d2 ( (Second spreading degree) is set (SA5). As understood from FIG. 5, the numerical value d2 exceeds the numerical value d1 (d2> d1). For example, a configuration in which the numerical value d2 is calculated by multiplying the predetermined numerical value d1 by a coefficient k (k> 1), or a configuration in which the numerical value d1 is calculated by dividing the predetermined numerical value d2 by the coefficient k is employed.

  As illustrated in FIG. 5, the fluctuation amount setting unit 28 generates a random number (that is, a normal random number) λ based on the probability distribution D of the variance σ set by the above procedure (SA6). A known technique can be arbitrarily adopted to generate the random number λ, but a normal random number λ is generated by applying the Box-Muller method to a uniform random number in which all numerical values are generated with the same probability. Is possible. Since the probability distribution D of the first embodiment is a normal distribution with zero as an average, the random number λ can be set to a positive number or a negative number.

  The fluctuation amount setting unit 28 sets the fluctuation amount Δ at the time of sound generation according to the random number λ (SA7). Specifically, a configuration in which the random number λ is adopted as the variation amount Δ or a configuration in which the variation amount Δ is calculated by a predetermined calculation using the random number λ as a variable may be employed. That is, the variation setting unit 28 of the first embodiment sets the variation Δ at the time of pronunciation of the selected note according to the random number λ generated in the probability distribution D of the variance σ according to the position of the selected note in the target music. . As can be understood from the above description, the note whose pronunciation point specified by the note information QN is located inside the beat range R is more variable than the note whose pronunciation point is located outside the beat range R. The adjusted sounding time points when the amount Δ is applied are dispersed over a wide range. That is, the actual singing tendency that the notes located near the beat point B are less likely to be dispersed (difficult to vary) than the notes separated from the beat point B is reproduced.

  When the fluctuation amount Δ is set for the selected note, the fluctuation amount setting unit 28 sets the fluctuation amount Δ for all the notes specified by the music data MP to be processed at the current stage (that is, all the notes of one voice part). It is determined whether it has been set (SA8). When there is a note for which the variation Δ is not set (SA8: NO), the variation setting unit 28 selects the immediately following note as a selected note (SA2) and then sets the variation Δ (SA3 to SA7). Run. On the other hand, when the setting of the fluctuation amount Δ is completed for all the notes (SA8: YES), the fluctuation amount setting unit 28 determines whether or not the processing has been completed for all the music data MP of the music data M (SA9). When the unprocessed music data MP exists (SA9: NO), the unprocessed music data MP is selected (SA1) and the above processing (SA2 to SA8) is executed. When the process is completed for all the music data MP included in the music data M (SA9: YES), the variation amount setting process ends. As understood from the above description, the setting of the variation amount Δ of each note is executed separately for each music data MP (for each voice part of the target music piece), so that the time point of the note specified by the note information QN is determined. Even when it is common to a plurality of voice parts, the sound generation time point of the note in the acoustic signal V may be different for each voice part. That is, natural synthesized speech in which the sound generation points of each voice part are appropriately dispersed is generated.

  As described above, in the first embodiment, the fluctuation amount Δ at the time of pronunciation of the note is determined according to the random number λ generated in the probability distribution D in which the variance σ is variably set according to the position of the note of the target music. Therefore, when compared with the configuration of Patent Document 1 in which the amount of variation at the time of pronunciation is set regardless of the position of each note in the target music, the amount of variation (error) at the time of pronunciation of each note in actual singing or performance ) And the position of each note in the music piece, there is an advantage that a natural variation Δ can be set.

  For example, in the first embodiment, since the variance σ of the probability distribution D varies depending on whether or not the pronunciation point specified by the note information QN is included in the beat range R, it is separated from the beat point B of the music. It is possible to set a natural variation Δ that reproduces the actual singing tendency that the notes are more easily distributed than the notes near the beat point B. Further, since the variance σ is set to be binary (d1, d2), for example, the amount of variation is set as compared with the configuration in which the variance σ is continuously changed according to the distance between the beat point B and the time point of note production. There is also an advantage that the processing is simplified.

Second Embodiment
A second embodiment of the present invention will be described below. In addition, about the element which an effect | action and function are the same as that of 1st Embodiment in each form illustrated below, the reference | standard referred by description of 1st Embodiment is diverted, and each detailed description is abbreviate | omitted suitably.

  FIG. 6 is a partial flowchart of the fluctuation amount setting process executed by the fluctuation amount setting unit 28 of the second embodiment, and FIG. 7 is an explanatory diagram of the fluctuation amount setting process in the second embodiment. As illustrated in FIG. 6, the fluctuation amount setting unit 28 of the second embodiment variably sets the correction value A for each music data MP (for each voice part) (SA11). Specifically, the correction value A can be set to a positive number or a negative number for each voice part in accordance with an instruction from the user to the input device 16, for example.

  As understood from FIG. 7, the fluctuation amount setting unit 28 controls the center (average value) of the probability distribution D of the variance σ according to the position of the selected note according to the correction value A. Specifically, the fluctuation amount setting unit 28 generates a random number λ based on a probability distribution D of variance σ having the correction value A as an average value (that is, a probability distribution D in which the generation probability of the correction value A is maximum). (SA6). The setting of the variation Δ according to the random number λ (SA7) is the same as in the first embodiment. As understood from the above description, the correction value A corresponds to the offset value of the probability distribution D. In the normal distribution exemplified above, the center of the probability distribution D (the numerical value with the highest occurrence probability) corresponds to the average value, but in other types of probability distributions (for example, lognormal distribution), the occurrence probability is the maximum. The numerical value (mode) and the average value can be different.

  When the correction value A is set to a positive number, the sound generation time of each adjusted note to which the variation Δ is applied is centered on a time point behind the initial sound generation time specified by the note information QN. Distributed. Therefore, the tendency of singing (so-called post-slipping) in which the time of pronunciation of each note is delayed compared to the intended time (sounding time on the score) is reproduced. On the other hand, when the correction value A is set to a negative number, the sound generation time of each adjusted note to which the variation Δ is applied is centered on the time point ahead of the initial sound generation time specified by the note information QN. As distributed. Therefore, the tendency of the singing preceding the time point at which each note is pronounced compared to the intended time point (so-called pre-groove) is reproduced.

  In the second embodiment, the same effect as in the first embodiment is realized. In the second embodiment, since the center position (average value) of the probability distribution D is variably set according to the correction value A, the leading / delaying tendency (previous / rear) at the time of sound generation of each note. ) Can be set. In the second embodiment, since the correction value A is individually set for each voice part of the target music piece, the tendency of leading / delaying at the time of pronunciation of each note is made different for each voice part (for each singer). Is possible.

<Third Embodiment>
FIG. 8 is an explanatory diagram of the operation (variation amount setting process) of the variation amount setting unit 28 in the third embodiment. FIG. 8 illustrates a note N1 included in one voice part of the target musical piece and a note N2 immediately after the note N1.

  As understood from FIG. 8, the fluctuation amount setting unit 28 of the third embodiment is such that the sound generation time point of the adjusted note N2 to which the fluctuation amount Δ is applied is inside a predetermined range (hereinafter referred to as “fluctuation allowable range”) P. The variation amount Δ of the note N2 is limited so as to be located at. The end portion (leading edge) p1 of the fluctuation allowable range P is set, for example, at the sounding time point of the note N1 immediately before the note N2 (the sounding time point after adjustment by applying the fluctuation amount Δ of the note N1). That is, the fluctuation amount setting unit 28 changes the fluctuation amount of the note N2 so that the sounding time point of the adjusted note N2 to which the fluctuation amount Δ is applied is positioned rearward on the time axis compared with the sounding time point of the immediately preceding note N1. Limit Δ.

  On the other hand, the end portion (rear edge) p2 of the fluctuation allowable range P is set, for example, at the time when the note N2 specified by the note information QN is muted. In other words, the fluctuation amount setting unit 28 sets the fluctuation amount Δ of the note N2 so that the sounding time point of the adjusted note N2 to which the fluctuation amount Δ is applied is positioned forward on the time axis as compared to the mute time point of the note N2. Limit.

  According to the above configuration, the variation amount Δ at the time of sound generation of each note is limited so that the sound generation time of each note after adjustment to which the variation amount Δ is applied is located within the variation allowable range P. Therefore, it is possible to adjust the sound generation time point of each note without causing the disappearance of the note of the target music (destruction of the music content).

  In the example of FIG. 8, the end portion p1 of the fluctuation allowable range P is set to the time point when the note N1 is sounded. It is also possible to set the point of time as the end portion p1. According to the above configuration, there is an advantage that a predetermined time length can be secured for the immediately preceding note N1 even when the sounding time point of the note N2 fluctuates excessively forward. Similarly, in the example of FIG. 8, the end portion p2 of the fluctuation allowable range P is set to the time point when the note N2 is muted, but only a predetermined time length (for example, a time length corresponding to a sixteenth note) from the time point when the note N2 is muted. It is also possible to set the front time point as the end portion p2. According to the above configuration, there is an advantage that a predetermined time length can be secured for the note N2 even when the sounding point of the note N2 is fluctuated excessively backward.

<Fourth embodiment>
9 and 10 are explanatory diagrams of the operation (variation amount setting process) of the variation amount setting unit 28 in the fourth embodiment. FIG. 9 and FIG. 10 illustrate the note N1 and the note N2 that are reciprocal in one voice part of the target music piece. In FIG. 9, when the initial sound generation period X2 designated by the note information QN is continuous between the note N1 and the note N2 (when the mute time of the note N1 coincides with the sound generation time of the immediately following note N2). In FIG. 10, it is assumed that the initial sound generation period X2 specified by the note information QN is spaced apart from each other by the note N1 and the note N2.

  The variation amount setting unit 28 of the fourth embodiment sets the variation amount Δ at the time of pronunciation of each note of the target music by the variation amount setting process similar to that of the first embodiment, and also the variation amount ΔE at the time point when each note is muted. Set. Specifically, the fluctuation amount setting unit 28 of the fourth embodiment executes the process of FIG. 11 for each setting of the fluctuation amount Δ when the selected note is sounded (SA7), so that the note immediately before the selected note ( The amount of fluctuation ΔE at the time of mute of “preceding note” is set.

  When the process of FIG. 11 is started, the fluctuation amount setting unit 28 determines whether or not the selected note (note N2) and the preceding note (note N1) are continuous with each other on the time axis (SB1). When the note N1 and the note N2 are continuous with each other (SB1: YES), as shown in FIG. 9, the fluctuation amount setting unit 28 determines the immediately preceding note according to the fluctuation amount Δ set for the time point at which the note N2 is sounded. A fluctuation amount ΔE at the time of mute of N1 is set (SB2). Specifically, the note N1 and the note N2 are continuous with each other even after the adjustment of the sounding point of the note N2 to which the fluctuation amount Δ is applied and the adjustment of the muting point of the note N1 to which the fluctuation amount ΔE is applied. The variation amount ΔE is set to a value equivalent to the variation amount Δ.

  On the other hand, when the note N1 and the note N2 are spaced apart from each other on the time axis (SB1: NO), the fluctuation amount setting unit 28 changes the time when the note N1 is muted as illustrated in FIG. The amount ΔE is set according to a predetermined probability distribution DE (SB3). That is, similarly to the setting of the fluctuation amount Δ at the time of sound generation, the fluctuation amount setting unit 28 sets the fluctuation amount ΔE according to the random number λ generated in the probability distribution DE. However, the variance of the probability distribution DE is fixed to a predetermined value. The variation amount ΔE set for each note by the variation amount setting unit 28 is included in the control information QC of the music data MP together with the variation amount Δ of the note. In addition, when the note N1 and the note N2 are separated from each other, the variance of the probability distribution DE can be set variably.

  In the speech synthesis process for generating the acoustic signal V, the speech synthesizer 26 adjusts the mute time point of the sound generation period X2 designated by the note information QN of each note according to the variation ΔE designated by the control information QC of the note. To do. For example, the speech synthesizer 26 varies the time point of the note specified by the note information QN by the variation ΔE on the time axis.

  In the fourth embodiment, the same effect as in the first embodiment is realized. Further, in the fourth embodiment, when the sound generation period X2 specified by the note information QN is continuous between the note N1 and the note N2, the mute time of the note N1 according to the variation Δ at the time of sound generation of the note N2. The fluctuation amount ΔE is set. Therefore, the relationship (continuous / separated) between the note N1 and the note N2 can be maintained before and after the adjustment. On the other hand, when the sound generation period X2 is separated from each other by the note N1 and the note N2, the fluctuation amount ΔE when the note N1 is muted is set according to the probability distribution DE. Therefore, it is possible to generate natural synthesized speech in which the mute times of each note are appropriately dispersed.

<Modification>
Each of the above forms can be variously modified. Specific modifications are exemplified below. Two or more aspects arbitrarily selected from the following examples can be appropriately combined.

(1) In each of the above-described forms, the variation Δ at the time of sound generation of each note is set according to the random number λ generated in the probability distribution D, but the variance σ is variably set according to the position of the note in the target music. The type of variable (music information) for which the amount of variation is set using the probability distribution D thus set is arbitrary. For example, the fluctuation amount ΔE at the time of mute of each note is generated in a probability distribution D of variance σ according to the position of the note in the target music (for example, whether the mute time is included in the beat range R). It is also possible to set according to the random number λ. Further, as illustrated below, the fluctuation amount of the variable set for each note other than the end point (sound generation time, mute time) of the sound generation period X2 is generated in the probability distribution D of the variance σ according to the position of the note. It is also possible to set according to the random number λ to be performed.

  For example, the amount of change in a variable (hereinafter referred to as “attenuation variable”) for controlling the characteristics of an attenuation section (decay) in which the intensity attenuates with time in the sound of each note of the target music is the same as in the first embodiment. In addition, it is possible to set according to the random number λ of the probability distribution D of the variance σ according to the position of the note in the target music. The attenuation variable is a variable for controlling the temporal change (attenuation amount per unit time) in the attenuation section of the sound of each note. Note that there is a tendency for the attenuation characteristics to be less fluctuated for notes that follow the note immediately after them or for notes that have a correspondingly long note value. Considering the above tendency, for example, according to the random number λ of the probability distribution D for notes that do not continue to the note immediately after in the target music and notes with a note value lower than a predetermined value (for example, notes less than 16th notes) A configuration in which the amount of fluctuation of the attenuation variable is set is preferable. On the other hand, for notes that do not match the conditions (notes that continue immediately after them or notes that have a correspondingly long note value), depending on the configuration in which the variation amount of the attenuation variable is set to zero, or the random number λ of the probability distribution D with a small variance σ Thus, a configuration for setting the fluctuation amount is adopted.

  As in the first embodiment, the variable amount for controlling the musical expression of each note (expression variable) is a random number λ of the probability distribution D of variance σ corresponding to the position of the note in the target music piece. It is also possible to set according to. Examples of facial expression variables include not only volume (velocity, dynamics) but also variables that control pitch bend and variables that control vibrato (for example, vibrato speed and depth).

  As can be understood from the above examples, the variation setting unit 28 in each of the above-described forms has a random number λ generated with a probability distribution D of variance σ corresponding to the position of the note in the target song for each note of the target song. Accordingly, the music information is comprehensively expressed as an element (variation amount setting means) for setting the variation amount of the music information of the note. Can be included.

(2) In each of the above-described embodiments, the variance σ of the probability distribution D is controlled according to whether or not the sound generation time of each note specified by the note information QN is included in the beat point range R, but the probability distribution D The method of setting the variance σ is not limited to the above example. Specifically, the variance σ of the probability distribution D can be changed over time. For example, a configuration is adopted in which the variance σ is changed continuously or step by step with a predetermined number of beats of the target music. It is also possible to change the variance σ regardless of the beat point of the target music piece. For example, a configuration in which the variance σ is increased or decreased from the start point to the end point of the target music, or a configuration in which the variance σ is different between a specific section (for example, a chorus section) of the target music and another section is adopted.

(3) In each of the above embodiments, the case where the acoustic signal V of the singing voice is generated (speech synthesis) is assumed, but the case where the acoustic signal V of the performance sound obtained by playing the target music piece with the musical instrument is generated (automatic performance). Also, the present invention is applied in the same manner as the above-described embodiments. Further, in each of the above-described embodiments, the case where the variation amount Δ of each note is separately set for each of a plurality of voice portions has been illustrated, but also when the variation amount Δ of each note is set for one voice portion. The present invention applies.

(4) In each of the above-described embodiments, the speech synthesizer 100 that executes the editing of the music data M and the generation of the acoustic signal V is exemplified, but the editing of the music data M (the display control unit 22 and the information management unit 24) The present invention is also specified as a music processing apparatus in which the generation of the acoustic signal V (speech synthesizer 26) is omitted. The music processing device is a device (variation amount setting device) that sets a variation amount of music information (for example, a variation amount Δ at the time of pronunciation) for each note of the target musical piece, and the variation amount setting unit 28 exemplified in the above embodiments. It is comprised including. The speech synthesizer 100 of each embodiment described above is understood as an apparatus in which the functions of editing the music data M and generating the acoustic signal V are added to the music processing apparatus.

(5) In each of the above-described embodiments, the range width of the beat point range R is fixed to a predetermined value. However, the range width (time length) of the beat point range R can be variably set. For example, a configuration in which the range width of the beat point range R is variably set according to the genre of the target music and the performance tempo (for example, a configuration in which the range width of the beat point range R is reduced as the performance tempo is faster), or a variance value ( A configuration in which the range width of the beat point range R is variably set according to d1, d2) can also be employed. It is also possible to variably set the range width of the beat point range R in accordance with an instruction from the user to the input device 16.

(6) In each of the above embodiments, the variation Δ at the time of pronunciation of each note is set for all the music data MP (whole voice part). The time point specified by the music data MP (that is, the fluctuation amount Δ is not set (Δ = 0)) can also be used. It is possible to reproduce the tendency that when a singer with accurate pitch and pronunciation time is included in a chorus with a large number of people, the chorus voice has an overall impression.

(7) Variables can be individually set for each of a plurality of voice parts (music data MP). For example, a configuration in which variables such as the numerical value (d1, d2) of the variance σ of the probability distribution D and the range width of the beat point range R are individually set for each voice part is preferable.

(8) The music processing device (speech synthesizer 100) can be realized by a server device that communicates with a terminal device via a communication network such as a mobile communication network or the Internet. Specifically, the music processing device sets the amount of variation Δ at the time of pronunciation for each note by executing the processing exemplified in each of the above-described forms for the music data M received from the terminal device via the communication network. In addition, data specifying the variation amount Δ of each note in time series and music data M obtained by adjusting the sound generation point in the sound generation period X2 specified by each note information QN according to the variation amount Δ are transmitted from the communication network to the terminal device. Send.

DESCRIPTION OF SYMBOLS 100 ... Music processing device, 10 ... Arithmetic processing device, 12 ... Memory | storage device, 14 ... Display device, 16 ... Input device, 18 ... Sound emission device, 22 ... Display control part, 24 ... Information Management unit, 26... Speech synthesis unit, 28... Fluctuation amount setting unit.

Claims (5)

A music processing apparatus comprising: a fluctuation amount setting unit that sets a fluctuation amount at the time of sound generation of each note according to a random number generated with a probability distribution of the degree of dispersion according to the position of the note in the music for each note of the music . The variation amount setting means is configured such that when the sound generation time of one note is located inside the beat point range including the beat point of the music, the one variation is set according to the random number generated in the probability distribution of the first distribution degree. A variation amount at the time of sounding a note is set, and when the sounding time of the one note is located outside the beat point range, a random number generated with a probability distribution of the second spread degree exceeding the first spread degree The music processing device according to claim 1, wherein a variation amount at the time of sounding the one note is set accordingly. The variation amount setting means sets the variation amount at the time of sounding of the note in accordance with a random number generated in the probability distribution that maximizes the probability of random number generation corresponding to a variable correction value. 2. Music processing apparatus of 2. The fluctuation amount setting means is configured such that the time point of the one note after adjustment according to the amount of fluctuation is positioned forward compared to the time point of mute of the one note, and the other note immediately before the one note. The music processing device according to any one of claims 1 to 3, wherein a fluctuation amount at the time of sound generation of the one note is set so as to be located behind the time of sound generation of the note. The variation amount setting means is a variation amount at the time of sounding set for the second note when the sounding period of the first note and the sounding period of the second note immediately after the music are consecutive on the time axis. The music processing device according to any one of claims 1 to 4, wherein a variation amount at the time of mute of the first note is set according to.

Images