JP5024136B2 - Sound processing apparatus and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To stably normalize the pitch of an acoustic signal. <P>SOLUTION: A pitch detection unit 46 detects the pitch P of an acoustic signal SIN. A range setting unit 72 sets a normalization range R for each of a plurality of pitches (note numbers). A note specification unit 74 sequentially specifies notes corresponding to the normalization range R to which the pitch P detected by the pitch detection unit 46 belongs. Once the note specification unit 74 specifies a new note, the range setting unit 72 enlarges the normalization range R corresponding to that note, and contracts the normalization range R to the range before the enlargement over time. <P>COPYRIGHT: (C)2010,JPO&amp;INPIT

Description

  The present invention relates to a technique for normalizing a pitch of an acoustic signal to any one of a plurality of pitches.

Various techniques have been proposed for identifying any one of a plurality of pitches (ie, normalizing the pitch of an acoustic signal) from the pitch of an acoustic signal obtained by collecting a human voice or musical instrument sound. Yes. For example, in Patent Document 1, the pitch of an acoustic signal is assigned to each tone of a tuning scale (a set of pitches corresponding to white keys of a keyboard instrument) or 12 scales (a set of pitches corresponding to white keys and black keys). A technique for converting is disclosed.
JP 2003-122351 A

  For pitch normalization, a fixed range is used for each of a plurality of pitches. That is, the pitch corresponding to the range to which the pitch of the acoustic signal belongs is specified as a result of normalization. However, when the acoustic signal is unstable and the pitch fluctuates slightly, the above technique has a problem that the pitch after normalization changes every moment in a short time. For example, if the pitch of the acoustic signal fluctuates in the vicinity of the boundary value between the pitch C and the pitch C #, the normalized pitch will be short from one of the pitch C and the pitch C # to the other. It fluctuates in an unstable manner. In view of the above circumstances, an object of the present invention is to stably normalize the pitch of an acoustic signal.

  In order to solve the above problems, a sound processing apparatus of the present invention includes a pitch detection unit that detects a pitch of an acoustic signal, a range setting unit that sets a normalization range for each of a plurality of pitches, and a pitch detection unit. Pitch specifying means for sequentially specifying pitches corresponding to the normalized range to which the detected pitch belongs, and when the pitch specifying means specifies a new pitch, the range setting means sets the pitch to the pitch. Temporarily expand the corresponding normalization range. According to the above configuration, since the normalization range of the pitch newly specified by the pitch specifying means is temporarily expanded, even if the pitch of the acoustic signal fluctuates unstablely, the pitch specifying means It is possible to suppress the change in pitch specified by.

  In a preferred aspect of the present invention, when the pitch specifying unit specifies a new pitch, the range setting unit expands the normalized range corresponding to the pitch and sets the normalized range to the range before the expansion. Shrink. According to the above aspect, for example, a new pitch is specified as compared with a configuration in which the normalized range after expansion is maintained for a predetermined time (for example, a configuration in which the normalized range R is controlled as shown in FIG. 9). There is an advantage that the possibility of specifying other pitches increases as time elapses. For example, considering the tendency of the pitch of human vocal sounds and musical instrument performance sounds to be particularly unstable at the start of generation, the normalized range after expansion so that the amount of reduction in the normalized range decreases over time (For example, a configuration for controlling the normalization range as shown in FIG. 8) is preferable.

  In a preferred aspect of the present invention, a control means for designating a scale composed of a plurality of constituent sounds is provided, and the pitch specifying means corresponds to the pitch among the plurality of constituent sounds of the scale designated by the control means. Specify the pitch. In the above aspect, for example, it is possible to specify only musically appropriate pitches in the performance or singing of music of a specific scale. Furthermore, when the pitch corresponding to the normalization range to which the pitch detected by the pitch detecting means does not correspond to the constituent sound of the scale, the pitch specifying means adjusts the pitch (for example, the steps of FIGS. 4 and 5). According to the configuration in which step SB3 to step SB8 are performed), even if the pitch of the acoustic signal is inaccurate, a musically appropriate pitch can be specified.

  In a preferred aspect of the present invention, the range setting means does not execute the normalization range expansion when the use of pitch bend is instructed. In the above aspect, since the normalization range is not expanded when the use of pitch bend is instructed, it is also possible to realize a musical expression that changes the pitch.

  The sound processing apparatus according to each of the above aspects is realized by hardware (electronic circuit) such as DSP (Digital Signal Processor) dedicated to processing of input sound, and general-purpose arithmetic such as CPU (Central Processing Unit). This is also realized by cooperation between the processing device and the program. A program according to the present invention includes a pitch detection process for detecting a pitch of an acoustic signal, a pitch specification process for sequentially specifying a pitch corresponding to a normalization range to which a pitch detected by the pitch detection process belongs, and a plurality of pitch detection processes A process for setting a normalization range for each pitch, and when the pitch specifying means specifies a new pitch, a range setting process for temporarily expanding the normalization range corresponding to the pitch To run. According to the program of this invention, the effect | action and effect similar to the sound processing apparatus which concern on each above aspect are show | played. The program of the present invention is provided to a user in a form stored in a computer-readable recording medium and installed in the computer, or provided from a server device in a form of distribution via a communication network and installed in the computer. Is done.

  FIG. 1 is a block diagram of a sound processing apparatus according to an embodiment of the present invention. As shown in FIG. 1, an A / D converter 14 and an output processing device 22 are connected to the sound processing device 100. The A / D converter 14 converts the electric signal supplied from the sound collecting device 12 into a digital acoustic signal SIN and outputs the digital signal to the sound processing apparatus 100. The acoustic signal SIN is a signal that represents the waveform of an input sound (for example, a user's singing sound or musical instrument performance sound) collected by the sound collection device 12.

  The sound processing device 100 generates music data DN from the acoustic signal SIN (that is, encodes the acoustic signal SIN). The music data DN in this embodiment is data that conforms to the MIDI (Musical Instrument Digital Interface) standard. More specifically, the sound processing apparatus 100 specifies a pitch according to the pitch of the acoustic signal SIN (input sound) from among a plurality of pitches. That is, the sound processing apparatus 100 normalizes (quantizes) the pitch of the acoustic signal SIN to a plurality of pitches. The music data DN is a note-on event that designates, for example, a pitch specified from the pitch of the acoustic signal SIN as a note number.

  The output processing device 22 generates an output signal SOUT by executing predetermined processing on the music data DN. The output signal SOUT is a signal representing the sound of a pitch (musical sound or voice) corresponding to the note number specified by the music data DN. For example, a sound source circuit such as a MIDI sound source is suitably employed as the output processing device 22. By supplying the output signal SOUT to the sound emitting device 24 (for example, a speaker device), a playback sound having a pitch corresponding to the note number specified by the music data DN (that is, the pitch of the input sound is quantized to a specific pitch). Played sound).

  As shown in FIG. 1, the sound processing device 100 is realized by a computer system including a control device 32 and a storage device 34. The control device 32 is an arithmetic processing device that functions as a plurality of elements (a feature extraction unit 40, a normalization unit 50, a data generation unit 62, and a control unit 64) by executing a program. The storage device 34 stores a program executed by the control device 32 and various data used by the control device 32. A known recording medium such as a semiconductor storage device or a magnetic storage device is arbitrarily used as the storage device 34.

  The feature extraction unit 40 in FIG. 1 extracts the feature amount of the acoustic signal SIN for each of a plurality of sections (hereinafter referred to as “unit sections”) obtained by dividing the acoustic signal SIN on the time axis. The feature extraction unit 40 of this embodiment includes an average intensity detection unit 42, a peak intensity detection unit 44, and a pitch detection unit 46. The average intensity detector 42 calculates the average value (effective value) of the intensity of the acoustic signal SIN in each unit section as the average intensity QA. The peak intensity detection unit 44 calculates the peak intensity of the acoustic signal SIN in each unit section (the maximum intensity value in the unit section) as the peak intensity QB. The pitch detector 46 calculates the pitch (fundamental frequency) P [Hz] of the acoustic signal SIN for each unit section. For detecting the pitch P, a known technique such as an autocorrelation method or a zero crossing method is arbitrarily employed.

  The normalization unit 50 normalizes each feature amount extracted by the feature extraction unit 40. The normalization unit 50 according to this embodiment includes an average intensity normalization unit 52, a peak intensity normalization unit 54, and a pitch normalization unit 56. The average intensity normalization unit 52 normalizes the average intensity QA of each unit section to any numerical value QnA in n stages (for example, n = 128). The peak intensity normalization unit 54 normalizes the peak intensity QB of each unit section to any one of n stages of peak intensity QnB. The pitch normalization unit 56 specifies a note number (pitch) N corresponding to the pitch P of each unit section for each unit section. The note number N is a numerical value (N = 0 to 127) selected according to the pitch P from n stages. The average intensity QnA, peak intensity QnB, and note number N after normalization by the normalization unit 50 are supplied to the data generation unit 62.

  The data generator 62 generates music data DN from the average intensity QnA, peak intensity QnB, and note number N. A specific example of generating the music data DN will be described later. The control unit 64 controls each element of the control device 32. For example, the control unit 64 selects one of the multiple types of scales (hereinafter, particularly referred to as “selected scale”) in accordance with an instruction from the user, for example. Further, the control unit 64 selects whether or not to use the pitch bend for generating the music data DN in accordance with, for example, an instruction from the user.

  As shown in FIG. 1, the pitch normalization unit 56 includes a range setting unit 72 and a pitch specifying unit 74. The range setting unit 72 sets a predetermined range (hereinafter referred to as “normalized range”) R for each of n stages of pitches (note numbers). Normalization ranges R corresponding to each note number are set so as not to overlap each other.

  The pitch specifying unit 74 specifies one note number N from the pitch P detected by the pitch detecting unit 46. More specifically, the pitch specifying unit 74 first corresponds to the normalized range R to which the pitch P detected by the pitch detecting unit 46 belongs among the n normalized ranges R set by the range setting unit 72. The note number N0 is specified (normalized) for each unit section. Second, the pitch specifying unit 74 specifies (quantizes) the note number N of the constituent sound corresponding to the note number N0 among the plurality of constituent sounds of the selection scale designated by the control unit 64 for each unit section.

  The specification of the note number N0 by the pitch normalization unit 56 will be described below by dividing into the case where the pitch bend is used and the case where the pitch bend is not used. FIG. 2 and FIG. 3 are conceptual diagrams illustrating the change over time of the pitch P (vertical axis) detected by the pitch detection unit 46 and the note number N0 specified by the pitch specifying unit 74. One white circle in FIGS. 2 and 3 means the pitch P extracted from the acoustic signal SIN of one unit section. In FIG. 2, it is assumed that use of pitch bend is instructed, and in FIG. 3, it is assumed that use of pitch bend is not instructed.

[1] When using pitch bend
As shown in FIG. 2, the range setting unit 72 normalizes each of n types of note numbers (only three types of note numbers from pitch B to pitch C # are shown in FIG. 2). Set the range R. For example, the range from the lower limit p0_B to the upper limit p0_C is set as the normalized range R_B of the pitch B, and the range from the lower limit p0_C to the upper limit p0_C # is set as the normalized range R_C of the pitch C. A range from the value p0_C # to the upper limit value p0_D is set as the normalized range R_C # of the pitch C #. When the use of the pitch bend is instructed from the control unit 64, the normalized range R is fixed to a predetermined range (hereinafter referred to as “initial range”) r0 (that is, does not change with time). In FIGS. 2 and 3, the case where the initial range r0 of each pitch has the same size is illustrated for convenience, but in actuality, the width of the initial range r0 is set individually for each pitch. .

  The pitch specifying unit 74 specifies the note number N0 corresponding to the normalized range R including the pitch P of each unit section. For example, the pitch specifying unit 74 specifies the note number N0 of the pitch C for the unit interval where the pitch P belongs to the normalized range R_C, and the pitch C for the unit interval where the pitch P belongs to the normalized range R_C #. Specify note number N0 of #. Therefore, the pitch P fluctuation (pitch bend) is precisely reflected in the time series of the note number N0 specified by the pitch specifying unit 74. For example, when the pitch P fluctuates in the vicinity of the boundary value p0_C # between the normalized range R_C and the normalized range R_C #, the note number N0 specified by the pitch specifying unit 74 is the pitch C and the pitch C #. It frequently changes from one to the other.

[2] When not using pitch bend
When the use of pitch bend is not instructed from the control unit 64, the range setting unit 72 sets the normalization range R to the initial range r0 for each of the n types of note numbers, as shown in FIG. Each time 74 specifies a new note number N0, the normalized range R corresponding to the note number N0 is expanded and then reduced over time to the initial range r0. Further details are as follows.

  As shown in FIG. 3, when the pitch P belonging to the normalized range R_C (initial range r0) starts to be detected in the unit interval at the time point t1, the pitch specifying unit 74 sets a new note number for the unit interval at the time point t1. Specify N0 (pitch C). As described above, when the pitch specifying unit 74 specifies a new note number N0, the range setting unit 72 initializes the normalization range R_C used for normalizing the pitch P of each subsequent unit interval (time t2). Temporarily expand from the range r0. More specifically, the normalization range R_C is expanded to a range from the lower limit value pE_C below the lower limit value p0_C of the initial range r0 to the upper limit value pE_C # above the upper limit value p0_C # of the initial range r0 in the unit interval at the time point t2. Is done. Since the boundary values other than the boundary values of the normalization range R_C (for example, the boundary value p0_B and the boundary value p0_D) do not change, the normalization range R of the pitch (B, C #) adjacent to the pitch C is reduced.

  Further, the range setting unit 72 reduces the normalized range R_C expanded in the unit section at the time point t2 to the initial range r0 with time over a plurality of unit sections. The range setting unit 72 according to this embodiment reduces the enlarged normalized range R_C at a constant rate. That is, the upper limit value of the normalization range R_C decreases linearly from the upper limit value pE_C # at the time point t2 (immediately after the enlargement) over a predetermined time Δ to reach the upper limit value p0_C # of the initial range r0. The lower limit value of the conversion range R_C increases linearly over the time Δ from the lower limit value pE_C at time t2, and reaches the upper limit value p0_C of the initial range r0. The time Δ is a time length corresponding to a plurality of unit sections.

  The expansion of the normalized range R by the range setting unit 72 is performed every time the pitch specifying unit 74 specifies a new note number N0. For example, as shown in FIG. 3, at time t3 after the normalized range R_C is reduced to the initial range r0, the pitch P changes into the normalized range R_C # of the pitch C #, and the pitch specifying unit 74 When the note number N0 of the pitch C # is newly specified, the range setting unit 72 expands the normalized range R_C # corresponding to the pitch C # in the next unit section (time point t4) and then calculates the time Δ. Over time.

  As described above, the normalized range R expanded from the initial range r0 is applied to specify the note number N0 within the period from the unit interval immediately after the specification of the note number N0 to the time Δ. Therefore, when compared with the case where the normalization range R is fixed (for example, when the pitch bend is used as shown in FIG. 2), the note number N0 (note pitch N0 (which is specified by the pitch specifying unit 74) even if the pitch P slightly varies. The pitch C) is more likely not to change. That is, according to this embodiment, the note number N0 specified by the pitch specifying unit 74 can be stabilized.

  Next, FIG. 4 is a flowchart showing a specific operation of the pitch normalization unit 56 when the pitch bend is not used (in the case of FIG. 3). The processing from step SA1 to step SA9 corresponds to the processing for specifying the note number N0 corresponding to the pitch P detected by the pitch detector 46 (that is, normalizing the pitch P), from step SB1 to step SB8. The process of is for specifying the note number N of the constituent sound of the selected scale from the note number N0 specified in the processes from step SA1 to step SA8 (ie, quantizing the note number N0 into each constituent sound of the selected scale). It corresponds to processing.

  When the processing of FIG. 4 is started, the pitch normalization unit 56 converts the pitch P [Hz] detected by the pitch detection unit 46 into a pitch PC having cents [cent] as a unit (step SA1). Further, the pitch specifying unit 74 specifies the note number N0 corresponding to the normalized range R including the converted pitch PC in step SA1 among the n types of normalized ranges R set by the range setting unit 72 ( Step SA2). As illustrated in FIG. 3, the normalization range R is set to the initial range r0 when no input sound is generated.

  Next, the pitch normalization unit 56 determines whether or not the note number N0 specified in step SA2 is continuing over a plurality of unit sections (step SA3). When the result of step SA3 is negative (when a new pitch starts to be generated in the current unit section), the pitch normalization unit 56 executes steps SA4 to SA9 (update of the normalization range R). Without proceeding, the process proceeds to step SB1.

  In step SB1, the pitch specifying unit 74 determines whether or not the note number N0 specified in the processing from step SA2 to step SA9 for the current unit section corresponds to the note number of the constituent sound of the selected scale (that is, the pitch P is set). It is determined whether or not each constituent sound of the selected scale is quantized.

  If the result of step SB1 is affirmative, the pitch specifying unit 74 determines the note number N0 specified for the current unit section as the note number N (step SB2), and then proceeds to step SA10. On the other hand, if the result of step SB1 is negative, the pitch specifying unit 74 determines that the pitch (pitch calculated backward from the note number N0) PN [cent] corresponding to the note number N0 specified for the current unit interval is It is determined whether or not the pitch PC calculated in SA1 is exceeded (step SB3).

  If the result of step SB3 is affirmative, the pitch specifying unit 74 decreases the note number N0 by a unit amount (for example, 1) (step SB4). On the other hand, if the result of step SB3 is negative, the pitch specifying unit 74 increases the note number N0 by the unit amount (step SB5). That is, the pitch specifying unit 74 adjusts the note number N0 so as to approach the note number of the constituent sound of the selected scale.

  Next, the pitch specifying unit 74 determines whether or not the note number N0 adjusted in step SB4 or step SB5 corresponds to the note number of the constituent sound of the selected scale (step SB6). When the result of step SB6 is affirmative, the pitch specifying unit 74 determines the note number N0 after the adjustment in step SB4 or step SB5 as the note number N (step SB7). On the other hand, when the result of step SB6 is negative (that is, when note number N0 does not correspond to the constituent sound of the selected scale even by adjustment in step SB4 or step SB5), pitch specifying unit 74 does not specify note number N. (Step SB8). That is, the input sound of the current unit section in the acoustic signal SIN is ignored. When step SB7 or step SB8 is completed, the process proceeds to step SA10.

  In step SA10, the pitch specifying unit 74 determines whether or not the note number N determined in step SB2 or step SB7 for the current unit section is different from the note number N of the previous unit section (that is, for the current unit section). It is determined whether or not a new note number N has been specified. If the result of step SA10 is affirmative, the range setting unit 72 expands the normalized range R corresponding to the note number N specified for the current unit section, as described above with reference to FIG. 3 (step SA11). ). Furthermore, the range setting unit 72 sets the range expansion flag F stored in the storage device 34 to ON (step SA12), and then proceeds to step SA13. The range expansion flag F is a flag that is set to ON during expansion of the normalization range R.

  On the other hand, when the result of step SA10 is negative (that is, when the note number N has not changed from the previous unit section), the range setting unit 72 performs step SA11 (expansion of the normalized range R) and step SA12 ( The process proceeds to step SA13 without executing (setting of range expansion flag F).

  In step SA13, the pitch normalization unit 56 determines whether or not to normalize the pitch P. When the end of the operation of the sound processing apparatus 100 is instructed by the user, the result of step SA13 becomes affirmative, and the process of FIG. 4 ends. On the other hand, when the result of step SA13 is negative, the pitch normalization unit 56 executes the processing after step SA1 for another unit section subsequent to the current unit section.

  As described above, for the unit section after the start of the input sound, the normalized range R after expansion in step SA11 is applied to normalization of the pitch P in step SA2. When the input sound having the same note number N continues and the result of step SA3 becomes affirmative, the pitch normalization unit 56 determines whether or not the range expansion flag F is set to ON (that is, the normalization range R). (Step SA4).

  The normalization range R is expanded (step SA11) and the range expansion flag F is set (step SA12) in response to the specification of a new note number N by the pitch specifying unit 74, and the sound of the note number N is executed. Is continued even in the current unit section, the results of both step SA3 and step SA4 are affirmative and the processing from step SA5 to step SA9 is executed.

  In step SA5, the range setting unit 72 updates the normalized range R so that it is reduced as compared with the normalized range R at the current stage. For example, the range setting unit 72 sets a numerical value obtained by subtracting a predetermined value from the upper limit value of the normalization range R at the current stage as an upper limit value, and sets a numerical value obtained by adding the predetermined value to the lower limit value of the normalization range R at the current stage. Is calculated as the normalized range R after update.

  Next, the pitch normalization unit 56 determines whether or not the normalized range R after the update in step SA5 has sufficiently approached the initial range r0 (step SA6). More specifically, the pitch normalization unit 56 determines whether the difference value between the upper limit value of the normalized range R after the update and the upper limit value of the initial range r0 (or the difference value of the lower limit value of each range) is less than a predetermined value. Determine whether or not. When the result of step SA6 is affirmative (that is, when the normalized range R after enlargement is reduced until it sufficiently approaches the initial range r0), the range setting unit 72 sets the normalized range R to the initial range r0. At the same time, the range expansion flag F is set to OFF (step SA7), and the process proceeds to step SB1. Therefore, even when the pitch generated in the past unit interval continues in the current unit interval (step SA3: YES), at the stage where a predetermined time Δ has elapsed since the occurrence of the pitch, Since the result of step SA4 is negative, the normalized range R set to the initial range r0 is used for specifying the note number N0 in step SA2.

  When the result of step SA6 is negative (when the normalization range R is larger than the initial range r0), the pitch specifying unit 74 normalizes the pitch PC calculated at step SA1 after the update at step SA5. Renormalization is performed based on the range R (steps SA8 and SA9). More specifically, in step SA8, the pitch specifying unit 74 determines whether or not the pitch PC calculated in step SA1 is within the updated normalized range R (that is, the upper limit value of the updated normalized range R is determined). It is determined whether or not the lower limit value is exceeded. If the result of step SA8 is affirmative, the pitch specifying unit 74 changes the note number N0 specified in step SA2 to the note number N0 continued from the past unit interval (step SA9). The process proceeds to step SB1.

  On the other hand, when the result of step SA8 is negative (when the pitch PC is not included in the normalized range R after update), the pitch normalization unit 56 proceeds to step SB1 without executing step SA9. . That is, the result specified in step SA2 is determined as the note number N0. The above is the specific operation of the pitch normalization unit 56 when the pitch bend is not used.

  FIG. 5 is a flowchart showing a specific operation of the pitch normalization unit 56 when the pitch bend is used. As shown in FIG. 5, when the use of pitch bend is instructed from the control unit 64, the pitch normalization unit 56 specifies the pitch PC [cent] and the note number N0 from the pitch P in the same manner as the processing of FIG. After (Step SA1 and Step SA2), similarly to Step SB1 to Step SB8 in FIG. 4, processing for quantizing the note number N0 into a plurality of constituent sounds of the selected scale is executed. That is, when the use of the pitch bend is instructed, the range setting unit 72 does not execute the normalization range R expansion or the range expansion flag F setting. The above is the specific operation of the pitch normalization unit 56.

  The data generation unit 62 in FIG. 1 generates music data DN by using the note number N specified by the above procedure together with the average intensity QNA and the peak intensity QnB. For example, when the pitch specifying unit 74 specifies a new note number N with the generation of the input sound, the data generating unit 62 generates a note-on event that instructs the note number N to be pronounced as music data DN. . When the pitch specifying unit 74 specifies a new note number N along with the change in the pitch of the input sound, the data generating unit 62 changes the note-off event for instructing to mute the previous note number N and after the change. A note-on event for instructing pronunciation of the note number N is generated as music data DN. When the note number N is not specified by the pitch specifying unit 74 due to the muting of the input sound, the data generating unit 62 generates a note-off event for instructing to mute the previous note number N as the music data DN. To do.

  The velocity (volume) in the note-on event generated by the data generation unit 62 is set according to the normalized peak intensity QnB. FIG. 6 is a graph showing the relationship between peak intensity QnB (horizontal axis) and velocity (vertical axis). As shown in FIG. 6, the storage device 34 stores a plurality of relationships CB (CB1, CB2,...) Regarding the peak intensity QnB and the velocity. The data generation unit 62 selects a relation CB corresponding to an instruction from the user from among a plurality of relations CB held in the storage device 34, and sets the velocity corresponding to the peak intensity QnB in the relation CB as a note-on event. To do. Therefore, it is possible to variously set the volume of the reproduced sound (sound signal SOUT) according to the user's intention.

  The data generator 62 generates an expression event as music data DN every time the average intensity QnA changes. The expression event is data for variably controlling the volume and tone color of the playback sound. An expression value (typically volume) in the expression event is set according to the average intensity QnA after the change. As shown in FIG. 7, a plurality of relationships CA (CA1, CA2,...) Are held in the storage device 34 for the average intensity QnA (horizontal axis) and the expression value (vertical axis). The data generation unit 62 selects a relationship CA corresponding to an instruction from the user from among a plurality of relationships CA held in the storage device 34, and sets an expression value corresponding to the average intensity QnA in the relationship CA in the expression event. To do. Therefore, the volume and tone color of the reproduced sound can be variously set according to the user's intention.

  When the use of pitch bend is instructed from the control unit 64, the data generation unit 62 generates a pitch bend event corresponding to the pitch P as music data DN. The pitch bend event is data for variably controlling the pitch of the reproduced sound. The pitch bend value in the pitch bend event is set according to the change amount of the pitch P (more specifically, the normalized pitch PC). The pitch bend event is generated every time the change amount (pitch bend value) of the pitch P changes.

  As described above, in this embodiment, since the pitch normalization range R newly specified by the pitch specifying unit 74 is temporarily expanded, the pitch P of the acoustic signal SIN (input sound) is unstable. Even if it fluctuates, it is possible to suppress the change in the note number N. Therefore, there is an advantage that stable reproduction sound is generated. On the other hand, when the use of pitch bend is instructed, expansion of the normalization range R is prohibited, so that musical expression (pitch bend) that changes the pitch of the reproduced sound is given priority over the stability of the pitch. Is also possible. That is, there is an advantage that various reproduced sounds reflecting the user's intention can be generated.

  Furthermore, since the note number N corresponding to the pitch P is specified from a plurality of constituent sounds of the selected scale, for example, it is possible to specify only a musically appropriate pitch in performance or singing of a musical piece of a specific scale. it can. In particular, since the note number N0 not corresponding to the constituent sound of the selected scale is adjusted so as to approach the pitch of the constituent sound, even when the pitch P of the acoustic signal SIN is inaccurate or unstable, There is an advantage that a reproduction sound having a musically appropriate pitch can be generated. Even if there is an error in the note number N0 specified from the pitch P to the pitch PC in step SA1 and step SA2, the influence of the error can be suppressed by adjusting the note number N0. It is.

<Modification>
Various modifications are added to the above embodiment. An example of a specific modification is as follows. Two or more aspects may be arbitrarily selected from the following examples and combined.

(1) Modification 1
A method for temporarily expanding the normalization range R is arbitrary. For example, in consideration of the tendency that human vocal sounds and musical instrument performance sounds are particularly unstable immediately after the start of generation, the amount of reduction per unit time of the normalized range R is as shown in FIG. A configuration in which the range setting unit 72 changes the upper limit value and the lower limit value of the normalized range R in a curved manner so that it decreases with time (that is, the normalized range R gradually approaches the initial range r0) is also preferable.

  But the structure which changes the normalization range R continuously is not essential in this invention. For example, as shown in FIG. 9, the normalization range R immediately after the expansion is maintained over the time Δ and the normalization range R is set to the initial range r0 when the time Δ has elapsed since the expansion. The intended effect of suppressing frequent changes in the pitch (note number N) specified by the pitch specifying unit 74 is realized. However, a configuration in which the new pitch normalization range R is gradually reduced as shown in FIGS. 3 and 8 (that is, a configuration in which the normalization range R of other pitches adjacent to the new pitch is gradually increased). ) Has an advantage that the user can easily designate another pitch after designating a new pitch.

(2) Modification 2
A configuration for quantizing the pitch P of the acoustic signal SIN into a plurality of constituent sounds of the selected scale is not essential in the present invention. That is, a configuration in which steps SB1 to SB8 in FIG. 4 and FIG.

(3) Modification 3
In the above embodiment, the note number N specified by the pitch specifying unit 74 is specified by the music data DN. However, the note number N may be further adjusted and used for generating the music data DN. For example, in the configuration in which the control unit 64 specifies the octave shift amount for specifying the octave to the pitch specifying unit 74, the note number N specified in step SB2 or step SB7 is shifted according to the octave shift amount. Specific note number N is identified.

(4) Modification 4
In the above embodiment, the music data DN is output to the output processing device 22, but the method of using the music data DN is arbitrary in the present invention. For example, a configuration in which a time series of music data DN is stored in the storage device 34 (or another storage medium) or a configuration in which the time series of music data DN is transmitted to a communication network is employed.

1 is a block diagram of a sound processing apparatus according to an embodiment of the present invention. It is a conceptual diagram which shows the relationship between the pitch and note number in the case of using a pitch bend. It is a conceptual diagram which shows the relationship between the pitch and note number when not using a pitch bend. It is a flowchart which shows operation | movement of the pitch normalization part when not using a pitch bend. It is a flowchart which shows operation | movement of the pitch normalization part in the case of utilizing a pitch bend. It is a graph which shows the relationship between peak intensity and velocity. It is a graph which shows the relationship between average intensity | strength and an expression value. It is a conceptual diagram which shows the change of the normalization range in a modification. It is a conceptual diagram which shows the change of the normalization range in a modification.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Sound processing apparatus, 12 ... Sound collection apparatus, 14 ... A / D converter, 22 ... Output processing apparatus, 24 ... Sound emission apparatus, 32 ... Control apparatus, 34 ... Memory | storage device, 40 ...... Feature extraction unit 42... Average intensity detection unit 44... Peak intensity detection unit 46... Pitch detection unit 50... Normalization unit 52. Normalization unit 56... Pitch normalization unit 62... Data generation unit 64... Control unit 72. Range setting unit 74 74 Pitch specification unit SIN ... Acoustic signal P Pitch , N ... note number, DN ... music data, SOUT ... output signal, R (R_B, R_C, R_C #) ... normalized range, r0 ... initial range.

Claims (6)

Pitch detection means for detecting the pitch of the acoustic signal;
Range setting means for setting a normalization range for each of a plurality of pitches;
Pitch specifying means for sequentially specifying pitches corresponding to the normalized range to which the pitch detected by the pitch detecting means belongs,
When the pitch specifying unit specifies a new pitch, the range setting unit temporarily expands a normalized range corresponding to the pitch. When the pitch specifying unit specifies a new pitch, the range setting unit expands the normalized range corresponding to the pitch and reduces the normalized range over time to the range before the expansion. 1. Sound processing apparatus. Comprising a control means for designating a scale composed of a plurality of constituent sounds;
The sound processing device according to claim 1 or 2, wherein the pitch specifying means specifies a pitch corresponding to the pitch from a plurality of constituent sounds of a scale designated by the control means. The sound processing device according to claim 3, wherein the pitch specifying unit adjusts the pitch when a pitch corresponding to a normalized range to which the pitch detected by the pitch detection unit belongs does not correspond to a constituent sound of the scale. The sound processing device according to any one of claims 1 to 4, wherein the range setting means does not execute the normalization range expansion when the use of pitch bend is instructed. A pitch detection process for detecting the pitch of the acoustic signal;
A pitch specifying process for sequentially specifying pitches corresponding to the normalized range to which the pitch detected by the pitch detection process belongs;
A process for setting the normalization range for each of a plurality of pitches, and when the pitch specifying means specifies a new pitch, range setting for temporarily expanding the normalization range corresponding to the pitch A program that causes a computer to execute processes.

Images