Method and device for musical sound production

Description: TECHNICAL FIELD [Detailed description of the invention]

【0001】

[Technical field to which the invention belongs]

The present invention relates to a musical tone generation method and device for simultaneously producing a plurality of channels of musical tones.

【0002】

[Conventional technology]

Sound sources that simultaneously produce multiple channels of musical tones have been known in the past. In such a conventional sound source, musical sounds are independently generated for each sounding channel, and the number of waveform samples that are calculated and generated per unit time is constant for each sounding channel.

【0003】

[Problems to be Solved by the Invention]

However, in the above-mentioned conventional sound source, since the musical sound is independently generated for each sounding channel, the characteristics of the musical sound generated for each channel are different, and it is required according to the musical sound generated for each sounding channel. Although the quality of the musical tones is different, the waveform samples were generated with the same number of samples for all the sounding channels, so that the arithmetic processing of the musical tone generation may be wasted.

【0004】

For example, if you want to generate a musical tone (that is, a high-quality musical tone) that has a frequency component over a wide frequency band, you need to perform a musical tone generation calculation at a high sampling frequency (that is, a large number of samples). In the case of generating a musical tone having only a frequency component in a low frequency band, it is sufficient to perform a musical tone generation calculation at a low sampling frequency (that is, the number of samples is small). Further, depending on the performance song, the quality of each note may be low, so that the number of pronunciations may be large, or the number of pronunciations may be small, so that the quality of the musical tone may be improved. Further, for the pronunciation channel that generates the musical tone of the prominent part such as the lead part in the performance song, it is desired to generate the musical tone of high quality, but the quality is slightly deteriorated in the relatively inconspicuous part such as the backing. There is not much problem in hearing even if the musical sound is generated.

【0005】

In general, a sound source generates a musical tone by a musical tone generation calculation in an arithmetic circuit, and since the arithmetic circuit has a limit in the arithmetic ability per unit time, the number of musical tones that can be generated at the same time (the number of simultaneous sounds) is limited. In the above-mentioned conventional sound source, since the calculation quality of each sounding channel is made uniform, depending on the channel, the musical sound generation calculation may be performed with a quality higher than necessary, and the number of musical sounds that can be sounded at the same time (the number of sounding channels). ) Decreased.

【0006】

The present invention has been made in view of the above problems, and an object of the present invention is to provide a musical tone generation method and device capable of adopting either a form of emphasizing the number of pronunciations or emphasizing quality according to the purpose of use of the user. And.

【0007】

[Means for solving problems]

In order to achieve the above object, the musical tone generation method according to claim 1 is a musical tone generation method including a step of generating a musical tone based on a musical tone type sample generated by a plurality of channels, and provides pitch information indicating the pitch. A step for inputting, a step for inputting control information indicating the number of waveform samples generated per unit time according to the operation amount of the operator operated by the user, and a musical tone type corresponding to the pitch information at predetermined intervals. This is a step of generating a sample for each channel with the number of waveform samples per unit time according to the input control information, and generates phase information according to the pitch information and the control information. It has a step of generating a musical sound type sample according to the phase information, and is characterized in that a musical sound is generated based on the generated musical sound type sample.

【0008】

The musical tone generation method according to claim 2 is a musical tone generation method including a step of generating a musical tone based on a musical sound waveform sample generated by a plurality of channels, in which a step of inputting performance information of a plurality of parts and a step per unit time are provided. A step of setting control information indicating the number of waveform samples to be generated in each part, and the input performance information is assigned to one of the plurality of channels, and the performance information is assigned to the assigned channel. It has a step of generating a musical tone sample according to the number of waveform samples per unit time according to the set control information of the part to which the performance information belongs, and based on the generated musical tone sample. It is characterized by generating musical tones.

【0009】

The musical tone generation method according to claim 3 is a musical tone generation method including a step of generating a musical tone based on a musical sound type sample, in which a step of inputting performance information and control information indicating the number of waveform samples generated per unit time are shown. A musical tone generation operation based on the waveform data stored in the waveform memory is executed according to the step of generating the above-mentioned input performance information and the number of musical tone samples per unit time according to the generated control information. It has a step of generating a sample, and in the step of generating the musical tone type sample, a musical tone generation calculation is performed by selectively using different waveform data in the waveform memory according to the generated control information, and the musical tone is generated. The step of generating a musical tone is characterized in that a musical tone is generated based on the musical tone type sample generated by the musical tone generation calculation.

【0010】

The music sound generation method according to claim 4 includes a step of storing a first waveform data for generating a music sound at a predetermined sampling frequency in a storage means, and a predetermined sampling of the stored first waveform data. It has a step of converting it into waveform data having a sampling frequency different from the frequency and storing it as a second waveform data in the storage means, and a step of generating an ultrasonic sample based on the waveform data stored in the storage means. However, in the step of generating the music type sample, the music sound type sample can be generated with a plurality of different numbers of waveform samples per unit time, and the first waveform data can be generated according to the number of waveform samples per unit time. And, one of the second waveform data is selected, and a music sound type sample is generated based on the selected waveform data.

【0011】

The musical tone generator according to claim 5 is a musical tone generator provided with a musical tone generating means for generating a musical tone based on a musical tone type sample generated in a plurality of channels, and the musical tone generator for inputting pitch information indicating the pitch is input. An information input means, a control information input means for inputting control information indicating the number of waveform samples generated per unit time according to the operation amount of the operator, and a musical tone type sample corresponding to the pitch information at predetermined intervals. Is a musical tone type sample generation means for generating each channel with the number of waveform samples per unit time according to the input control information, and generates phase information according to the pitch information and the control information. However, it has a device that generates a musical tone type sample according to the phase information, and the musical tone generating means is characterized in that a musical tone is generated based on the generated musical tone type sample.

【0012】

The musical tone generator according to claim 6 is a musical tone generator provided with musical tone generating means for generating musical tones based on musical sound sample generated in a plurality of channels, and is a musical tone generating means for inputting performance information of a plurality of parts. A control information setting means for setting control information indicating the number of waveform samples generated per unit time for each part, and the input performance information assigned to any one of the plurality of channels, and the allocation thereof. In the channel, there is a musical tone sample generation means for generating a musical tone sample corresponding to the performance information with a number of waveform samples per unit time according to the set control information of the part to which the performance information belongs. However, the musical tone generating means is characterized in that a musical tone is generated based on the generated musical sound type sample.

【0013】

The musical sound generator according to claim 7 is a musical sound generator including a musical sound generating means for generating a musical sound based on a musical sound type sample, and is a musical sound generating device for inputting performance information and a waveform sample generated per unit time. A musical tone generation calculation based on the waveform data stored in the waveform memory is executed according to the control information generating means for generating the control information indicating the number and the input performance information, and the unit time corresponding to the generated control information is executed. It has a musical sound type sample generation means for generating a musical sound type sample with the number of waveform samples per hit, and the musical sound type sample generation means selectively selects different waveform data in the waveform memory according to the generated control information. It is characterized in that a musical sound generation calculation is performed using the musical sound generation means, and the musical sound generation means generates a musical sound based on a waveform sample generated by the musical sound generation calculation.

【0014】

According to the configuration of the invention according to claims 1 and 5, phase information is generated according to the input pitch and control information, and based on this phase information, the input pitch is performed at predetermined intervals. An information-based musical tone sample is generated for each channel with the number of waveform samples per unit time according to the input control information, and a musical tone is generated based on the generated musical tone sample. Depending on the purpose of use of the user, either the number of pronunciations or the quality can be emphasized.

【0015】

Further, according to the configuration of the invention according to claims 2 and 6, the musical tone type sample of the channel corresponding to the input performance information has a unit time according to the set control information of the part to which the performance information belongs. It is generated with the number of waveform samples per hit, and musical tones are generated based on this generated musical sound sample, so it is possible to generate high-quality musical tones for parts that have a large audible effect, maximizing limited computing power. It can be utilized.

【0016】

Further, according to the configuration of the invention according to claims 3 and 7, a music sound generation calculation based on the waveform data stored in the waveform memory is executed according to the input performance information, and the generated control information is struck. A music sound type sample is generated based on the number of waveform samples per unit time, and in the music sound generation calculation, a music sound generation calculation is performed by selectively using different waveform data in the waveform memory according to the generated control information. Since the music is generated based on the sound-type sample thus generated, the high equivalent sampling frequency channel uses waveform data with a wide band of frequency components (ie, high recording sampling frequency). However, for channels with a low equivalent sampling frequency, waveform data having a narrow band frequency component (that is, having a low recording sampling frequency) can be used, so that the waveform designation in the tone color data does not need to be changed.

【0017】

If the selected waveform data is not changed according to the time density of the generated musical tone sample, folding noise may occur in the generated musical tone sample or the time density may be generated for the reasons described below. It may not be possible to generate a musical tone of the quality corresponding to the above. Here, the time density is a sampling frequency when a musical sound type sample is generated, and is referred to as an equivalent sampling frequency in this specification. According to the sampling theorem, a musical tone sample can reproduce frequency components in a band below half the sampling frequency (hereinafter referred to as "upper limit frequency") of the sampled sampling frequency.

【0018】

When a musical tone is generated using the waveform data stored in the waveform memory, the stored waveform data is converted into a waveform sample having the pitch of the musical tone to be generated under the equivalent sampling frequency (hereinafter, "" It is called "pitch conversion"), and a musical tone type is generated based on the waveform data after pitch conversion. At this time, if the waveform data obtained as a result of pitch conversion of the waveform data contains a frequency component higher than the upper limit frequency corresponding to the equivalent sampling frequency, the frequency component is the waveform data after pitch conversion as loopback noise. Will be mixed in. On the other hand, when the waveform data after pitch conversion is considerably lower than the upper limit frequency, for example, contains only one-third or less of the frequency component, the waveform is generated at the equivalent sampling frequency with a high frequency, but the high frequency range is generated. Only musical tones lacking components are produced, and the quality of the musical tones is not very good.

【0019】

In the present invention, as waveform data used as a material for musical sound generation, waveform data suitable for the generation time density of the musical sound sample is prepared, and the waveform data is selectively used according to the time density.

【0020】

Further, according to the configuration of the invention according to claim 4, similarly to the invention of claim 3, since the waveform data corresponding to the number of waveform samples per unit time of the musical sound type sample is used, the folding noise is generated. Can be reduced. Furthermore, when a waveform sample is generated with two different waveform samples per unit time, the waveform data obtained by converting the waveform data suitable for the number of waveform samples per unit time in one unit is converted into the waveform data in the other unit time. Since it is used to generate a waveform with a large number of waveform samples, it is possible to generate an orthophonic sample that sounds the same tone even with different numbers of waveform samples per unit time.

【0021】

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

【0022】

FIG. 1 is a block diagram showing a configuration of a musical sound generator according to an embodiment of the present invention. The structure of this block diagram is exactly a general-purpose computer running an operating system (OS) such as Windows. Therefore, the present invention can be implemented as software on a general purpose computer. It can also be implemented on an electronic musical instrument having the same configuration as this block diagram.

【0023】

In the figure, the music sound generator of the present embodiment has a CPU 1 that performs various data processing, a keyboard 2 for the user to instruct program execution and input data, and various image information and character information. Display 3 to display, hard disk device 4 to store programs and data executed by CPU 1, ROM 5 to store data input / output control programs for keyboard 2, display 3, hard disk 4, etc., and running programs RAM 6 for storing waveform data and data being calculated, timer 7 for time measurement, MIDI interface 8 for inputting performance data connected to a performance device such as a keyboard, and RAM 6 in response to instructions from CPU 1. DMA (Direct) that directly accesses, for example, reads music sound data at a frequency corresponding to a sampling frequency of 48 kHz and inputs one sample at a time to the DA converter 10 (hereinafter referred to as "read / playback processing"). MemoryAccess) Control unit 9, DA converter (DAC) 10 that converts the waveform data of the digital signal supplied from DMA control unit 9 into analog music signal, and sound system 11 that amplifies the music signal and outputs it from the speaker. And a bus 12 that connects the above components 1 to 9 to each other.

【0024】

Hereinafter, in the present embodiment, the "waveform sample" or "sample" means the sampled individual waveform sample data, and the "waveform data" is the data obtained by summarizing the individual waveform sample data. , The data that is the basis of the music sound generation processing, and the "music sound waveform data" refers to the data generated as a result of the music sound generation processing among the collected data.

【0025】

FIG. 2 is a diagram showing the configuration of the tone color data and the waveform data stored in the RAM 6 and the configuration of the input buffer and the sound source register set on the RAM 6.

【0026】

First, the tone color data PDp (p = 1, ..., 16) in the figure (a) is the waveform specification data WN (p) that specifies the waveform of each range, and the LFO (Low Frequency), respectively. It is composed of Oscillator) control original data (OD), filter envelope generator (FEG) control original data, amplitude envelope generator (AEG) control original data, other original data, and calculation mode CM (p). .. Here, p indicates a part number, and the musical sound generator of the present embodiment is configured so that tone color data of 16 parts can be set. Further, in the present embodiment, the waveform designation data WN (p) is set by the waveform name, and the waveform name is set by the user (see FIG. 9). Further, the calculation mode CM (p) shows data (corresponding to an equivalent sampling frequency) that indirectly specifies how many samples are generated per second when generating a musical tone waveform. In the present embodiment, the calculation mode CM (p) is configured to take an integer value of any one of 0 to 2. Here, the calculation mode CM (p) and the equivalent sampling frequency correspond to CM = 0 at 48 kHz, CM = 1 at 24 kHz, and CM = 2 at 12 kHz.

【0027】

Next, the waveform data WD1, WD2, ... In the figure (b) show the waveform data that is the source of the music sound type data generated by the music sound generation processing, and among these waveform data, the waveform data WD1, WD2 Is waveform data composed of samples for a predetermined time length, and is composed of a one-shot reading waveform or a waveform having an attack portion and a loop portion. Each of the waveform data WD1 and WD2 reads out a plurality of waveform data sampled at a predetermined recording sampling frequency and stored in the hard disk 4 as necessary, and stores them in the waveform data area of the RAM 6. For the waveform data WD1'and WD2', respectively, the waveform data WD1 and WD2 are subject to a predetermined band limitation, downsampled at 1/2 of the original recording sampling frequency (skipping one sample at a time), and then the waveform data area is set. It is stored. The waveform data WD2 applies a predetermined band limitation to the waveform data WD2 and is downsampled at 1/2 of the sampling frequency of the waveform data WD2 , that is, 1/4 of the original recording sampling frequency, to obtain a waveform data area. It is stored in. The details of the process for generating the waveform data WD1', WD2', and WD2'" (waveform LPF process) will be described later with reference to FIG.

【0028】

The input buffer in FIG. 6 (c) is a buffer for storing performance data input via MIDI interface 8, an area for storing data indicating the number of events waiting to be processed, and events corresponding to each event. It consists of an area where data ID1, ID2, ID3, ... are stored. Each event data consists of data indicating the content of the event and data indicating the event occurrence time. The data indicating the time of occurrence is required for CPU1 to process a plurality of events at once.

【0029】

The sound source register shown in FIG. 6D stores control data for each sounding channel in the musical tone generation process, and 32 channels of registers are provided in the present embodiment. The control data of each sounding channel is generated by processing various original data of the tone color data PDp (see (a) above) according to performance data such as touch, and determines the pitch of the musical sound to be generated. The pitch SP shown, the F number FN indicating the amount of advance of the address per sample when reading the waveform data, the waveform specification data that specifies the waveform data to be read, and the LFO generated by processing the LFO control original data. Control data, FEG control data generated by processing the FEG control original data, AEG control data generated by processing the AEG control original data, note-on data indicating whether or not sounding is in progress, and other data. It consists of data and calculation mode CM (i) (i = 1, ..., 32).

【0030】

Here, the F number FN is a numerical value indicating the amount of advance of the read address per sample as described above, and is specifically calculated and set by the following equation.

【0031】

FN = 2^{(SP-OP) / 1200}×2^(CM-WM)However, SP indicates the pitch SP of the musical tone to be generated, and OP indicates the unique pitch (original) of the waveform data when the waveform data specified by the waveform specification data is read out one sample at a time at a sampling frequency of 48 kHz. Pitch) is indicated, CM indicates the calculation mode CM (i), and WM is a waveform mode, that is, a numerical value indirectly assigned to each waveform data and indirectly indicating the recording sampling frequency of the waveform data (this implementation). In the form of, the waveform mode WM is also configured to take an integer value of any one of 0 to 2 as in the calculation mode CM (i)). For example, waveform mode WM = 0 corresponds to a recording sampling frequency of 40k to 48kHz, waveform mode WM = 1 corresponds to 20k to 24kHz, and waveform mode WM = 2 corresponds to 10k to 12kHz. The waveform data of each waveform mode can have a high frequency component up to a frequency that is half the recording sampling frequency. Since the values SP and OP are given in cents, they are divided by "1200".

【0032】

Here, it is this F number FN that controls the mode of pitch conversion described above.

【0033】

Further, a work area for sound processing is secured in the sound source register, and this work area is used when it is necessary to correct the F number FN by, for example, LFO control data or the like.

【0034】

FIG. 3 is a diagram showing a configuration of an output buffer set on RAM6.

【0035】

In the figure, the buffer 0 accumulates 128 waveform samples generated in the sound channel to which the calculation mode 0 (CM (i) = 0) is assigned in the control data stored in the sound source register. Buffer 1 is a buffer for accumulating 64 waveform samples generated in the sound channel to which the calculation mode 1 (CM (i) = 1) is also assigned. 2 is a buffer for accumulating 32 waveform samples generated in the sound channel to which the calculation mode 2 (CM (i) = 2) is also assigned.

【0036】

Further, buffer 1'is a buffer for interpolating 64 waveform samples of buffer 1 to generate 128 waveform samples, and buffer 2'is a buffer for 32 waveform samples of buffer 2. It is a buffer for interpolating and generating 128 waveform samples. The 128 musical tone samples in buffers 1'and 2'generated in this way are accumulated with the 128 musical tone samples in buffer 0 and stored in buffer 0 as 128 waveform samples for reproduction. To. That is, the buffer 0 also functions as a buffer for storing 128 musical sound sample data for reproduction read by the DMA control unit 9.

【0037】

It is generated in buffer 0 to interpolate the musical tone samples (in this embodiment, the musical tone sample data of the buffers 1 and 2) generated at such a low sample rate into 128 musical tone samples. 128 musical samples and 64 and 32 musical samples generated in buffers 1 and 2, respectively, are accumulated without causing wrapping noise, and 128 musical samples for final reproduction. This is to generate. As a method of this interpolation, a well-known method such as linear interpolation may be used.

【0038】

In this embodiment, three types of buffers 0 to 2 are provided as described above, and the buffer size of each buffer is configured to be buffer 0: buffer 1: buffer 2 = 4: 2: 1. Needless to say, the number of buffers and the buffer size of each buffer are not limited to this. For example, two types of buffers 0 and 1 may be provided, and the buffer size of each buffer may be set so that buffer 0: buffer 1 = 3: 1.

【0039】

Next, an outline of the musical tone generation process performed by the musical tone generator of the present embodiment will be described with reference to FIG.

【0040】

For example, when a note-on event of part p is input to the input buffer of FIG. 2 (c), the tone color data PDp is specified by this event, and the waveform name WN (p) and the calculation mode CM ( When p) is specified, CPU1 puts the waveform sample of the music sound corresponding to the event input at the time density (equivalent sampling frequency) of the music sound generator according to the calculation mode CM (p) on the buffer corresponding to the calculation mode. Generates and executes a generation operation such as an interpolation operation on this waveform sample, stores the calculated data in buffer 0 of RAM6, and notifies the DMA control unit 9 of the end of the operation (Fig. 4 (b)). The DMA control unit 9 sequentially reads out the data and performs a reproduction process ((c) in the figure). Therefore, among the performance data input via the MIDI interface 8, the performance data corresponding to the performance input ((a) in the figure) between the previous occurrence time tBC of the clock BC and the current occurrence time tBC is this time. Is the target of the generation operation of. Then, the reading / reproducing process of the waveform data based on the generated data whose calculation is completed at the time tCE is executed by the DMA control unit 9 from the next time tBC, and the musical sound is output. The arrow P shown in the figure simply indicates the correspondence between the generated data and the read / play processing, and indicates that the operation result completed at time tCE is transferred to time tBC. is not it.

【0041】

Since the sampling frequency of the DAC10 of the musical tone generator of the present embodiment is 48 kHz, the generation cycle of the clock BC is 2.7 msec (128 / 48 k), and the maximum delay time from the performance input to the actual musical tone generation is about 5 msec. Therefore, it does not matter to the human sense of hearing. Of course, in the case of automatic performance, there is no problem even if this delay time is longer, so the size of buffer 0 may be increased. Further, the read / play process does not have to be limited to the timing of the clock BC, and may be any timing as long as there is no problem in hearing, such as a predetermined time after the time tBC of the clock BC.

【0042】

The control processing executed by the CPU 1 of the musical tone generator configured as described above will be described below with reference to FIGS. 5 to 13.

【0043】

FIG. 5 is a flowchart of the main routine. For example, when the user turns on the power of the musical tone generator (in the case of a general-purpose personal computer, when the software sound source program is started), the process is started. Further, FIG. 6 is a flowchart of MIDI reception interrupt processing, and this interrupt processing is executed with the highest priority when performance data is input via the MIDI interface 8.

【0044】

First, the process of FIG. 6 will be described. In step S11, the received data is taken in, and in step S12, the received data is written to the input buffer of the RAM 6 together with the time data indicating the reception time.

【0045】

In the main routine of FIG. 5, first, all the sounding channels are turned off, the registers are cleared, and then the initial setting processing such as initializing and starting the reading / reproducing processing of the DMA control unit 9 is performed (step S1), and then the input is performed. Determine if there is received data in the buffer (step S2). As a result, if there is no received data, the process immediately proceeds to step S4, and if there is received data, processing according to the received data, for example, note-on event processing, note-off event processing, pedal processing, etc., is performed, and the process proceeds to step S4.

【0046】

In step S4, it is determined whether or not a switch event such as tone selection has occurred, and if it does not occur, the process immediately proceeds to step S6. If so, each MIDI channel depends on the tone selection switch setting. After performing panel switch event processing such as selecting a tone each time (step S5), the process proceeds to step S6.

【0047】

In step S6, the sound source processing subroutine described later is executed using FIG. 13, and in step S7, other processing is performed and the process returns to step S2. After that, steps S2 to S7 are repeatedly executed.

【0048】

FIG. 7 is a flowchart showing a procedure of note-on-event processing, which is one of the received data processing executed in step S3 of FIG.

【0049】

First, in step S21, the pitch and the part number indicated by the received note-on event data are set to the area SP (hereinafter, this content is referred to as pitch SP) and p (hereinafter, this content is referred to as part, respectively) reserved in the RAM6. It is stored in (referred to as "p"), and its occurrence time (reception time) is stored in the area TM reserved in RAM 6 (hereinafter, this content is referred to as "occurrence time TM").

【0050】

Next, a sound allocation process for determining which channel to write to in the sound source register (Fig. 2 (d)) is performed, and the allocated channel number is assigned to the area i secured in RAM 6 (hereinafter, this content is referred to as "allocated channel i"). Store in (step S22).

【0051】

In the following step S23, the tone color data PDp (Fig. 2 (a)) of part p is searched to acquire the waveform name (waveform designation data) WN (p) and the calculation mode CM (p), and this waveform name WN (p) is obtained. ) Is selected from the waveform data according to the calculation mode CM (p), and the read address on the RAM 6 (see FIG. 2 (b)) is assigned to the waveform specification data of the channel i (Fig. 2 (d)). )) Set as.

【0052】

At this time, there may be a plurality of waveform data having the same waveform name WN (p). For example, the waveform data WD2 and WD2 are downsampled waveform data WD2 as described above, and their waveform names are the same as the waveform names of the waveform data WD2. When there are a plurality of waveform data having the same waveform name in this way, one waveform data is selected from among them according to the calculation mode CM (p).

【0053】

However, even when there are a plurality of waveform data having the same waveform name, there are cases where the waveform data corresponding to the calculation mode CM (p) does not exist. For example, in FIG. 2 (b), there is no waveform data WD1 corresponding to the calculation mode CM (p) = 2. This is because the user did not create the waveform data WD1 by, for example, waveform LPF processing (described later with reference to FIG. 11). In such a case, the waveform data corresponding to the calculation mode CM (p) cannot be selected. Therefore, the waveform data having the same waveform name WN (p) but different waveform mode WM is selected, and the waveform data is read out. The desired musical tone signal is generated by changing the speed (that is, changing the pitch, specifically adjusting the F number FN).

【0054】

In this embodiment, there is a one-to-one correspondence between the calculation mode CM (p) and the waveform mode WM of the waveform suitable for generating musical sounds in the calculation mode. That is, for example, when the calculation mode CM (p) = 0, the waveform data in which the waveform mode WM is 0 is selected. However, the calculation mode CM (p) and the waveform mode WM are different concepts and do not always match. In the present embodiment, the calculation mode CM (p) and the waveform mode WM are simply matched. In addition, it only generated waveform data.

【0055】

In the following step S24, the timbre data PDp of part p (that is, the timbre data indicated by the timbre number TC (p)) is processed according to the pitch SP and the calculation mode CM (p), and the assigned channel i is processed together with the event occurrence time TM. Set to a predetermined area of the sound source register of (step S24). Here, one purpose of processing the timbre data PDp is to control the volume envelope generator (not shown), such as the deformation of the volume EG control data and the cutoff frequency of the timbre filter (not shown). This is to prevent it from changing according to (p). Another purpose is to change the musical sound characteristics such as the shape of the envelope according to the performance information such as the pitch SP, as is performed in a normal electronic musical instrument. The timbre number TC (p) is set by the user as described later with reference to FIG.

【0056】

Next, note-on data is written to the sound source register of the allocated channel i (step S25), and this process is terminated.

【0057】

8 to 11 are flowcharts showing a procedure for processing various events generated when the user presses the panel switch, and each process is one process of the panel switch event process in step S5 (FIG. 5). Here, the panel switch may be one assigned in advance to the keyboard 2 or may be displayed on the display 3. When the panel switch is displayed on the display 3, the cursor may be moved by the up / down keys of the keyboard 2 or a mouse (not shown), and the desired panel switch may be pressed.

【0058】

FIG. 8 is a flowchart showing the procedure of the part tone selection process when the part tone selection switch (not shown) is pressed.

【0059】

In the figure, when the user first inputs a part number, the part number is stored in the area p, and then when the user inputs a tone number, the tone number is stored in the area TC (p) reserved in the RAM 6. Store (step S31). The timbre number stored in the area TC (p) is the timbre number TC (p).

【0060】

Next, the timbre data is prepared (step S32). Specifically, the tone color data indicated by the tone color number TC (p) is searched from the tone color data group stored in advance in the predetermined area of the hard disk 4, and the tone color data area indicated by the part p (FIG. 2 (a)). ) See). Further, referring to the waveform designation data WN (p) in the loaded tone color data, the waveform data designated by the data WN (p) and the data CM (p) is stored in the waveform data storage area of FIG. 2 (b). It is determined whether or not it exists, and if it does not exist, the waveform data is automatically loaded from the hard disk 4 into the waveform data storage area. At that time, the timbre data originally set in part p is saved in the corresponding storage area of the hard disk 4.

【0061】

FIG. 9 is a flowchart showing the procedure of the part mode selection process when the part mode selection switch (not shown) is pressed.

【0062】

First, the part number input by the user is stored in the area p in the same manner as in step S31, and then when the user inputs the calculation mode, the calculation mode (an integer value of 0 to 2). Is stored in the area CM (p) in the tone color data area indicated by part p (step S41).

【0063】

FIG. 10 is a flowchart showing the procedure of the part waveform selection process when the part waveform selection switch (not shown) is pressed.

【0064】

First, the part number input by the user is stored in the area p in the same manner as in step S31, and then when the user inputs the waveform name, the waveform name is stored in the area in the tone color data area indicated by the part p. Store in WN (p) (step S51).

【0065】

Next, the waveform data corresponding to the calculation mode CM (p) is prepared (step S52). Specifically, the waveform data indicated by the waveform name WN (p) in consideration of the calculation mode CM (p) is searched from the waveform data group stored in the predetermined area of the hard disk 4, and the waveform data of the RAM 6 is searched. Load into the area (see Figure 2 (b)).

【0066】

Here, "considering the calculation mode CM (p)" means that there are a plurality of waveform data having the same waveform name WN (p), and the waveform data of the waveform mode WM corresponding to the calculation mode CM (p) is In some cases, it means that only the waveform data of the waveform mode WM is loaded into the waveform data area, and in this way, RAM 6 can be used efficiently. However, not limited to this, all indicated by the waveform name WN (p) regardless of whether or not there are multiple waveform data with the same waveform name WN (p) without considering the calculation mode CM (p). The waveform data of may be loaded into the waveform data area.

【0067】

If the same waveform name is already stored in the waveform data area of RAM6, this process is not necessary, but reloading does not cause any problem.

【0068】

FIG. 11 is a flowchart showing a procedure of waveform LPF processing when a waveform LPF switch (not shown) is pressed.

【0069】

First, when the user inputs the waveform to be processed and the processing content thereof, the waveform to be processed is searched from the waveform data area of RAM 6 (step S61). At this time, if the waveform to be processed is not searched, the waveform data group of the hard disk 4 is searched. The processing content refers to the type of downsampling, the band to be restricted, and the like.

【0070】

Next, it is determined whether or not the waveform searched in this way has a loop portion that is repeatedly read (step S62), and if it does not have a loop portion, band limitation is first performed (step S63). Then, downsampling is performed (step S64), the waveform is completed, and stored in the waveform data area of RAM 6 (step S65). The band is limited in step S63 because if the downsampled waveform data is stored as it is without band limitation, wrapping noise is mixed in the stored waveform data. Therefore, it is necessary to limit the band so that the turnaround noise does not occur in the downsampling, and the frequency band to be limited is set by the user in step S61.

【0071】

FIG. 12 is a diagram for explaining the processes of steps S63 and S64, in which the horizontal axis indicates the frequency and the vertical axis indicates the level. Then, (a) is an original waveform (waveform name: PIANO) sampled at a predetermined recording sampling frequency Fs (that is, waveform mode WM = 0) before performing waveform LPF processing, that is, for music generation calculation with an equivalent sampling frequency of 48 kHz. An example of the frequency spectrum of .0) is shown, (b) shows the frequency spectrum after band limiting processing is applied to the waveform of (a), and (c) shows the original recording sampling of the waveform of (b). The frequency spectrum of the waveform data (waveform mode WM = 1) for music generation calculation with an equivalent sampling frequency of 24 kHz, downsampled at a frequency of 1/2 of the frequency, is shown, and (d) is for the waveform of (c). Further, LPF processing, that is, band limiting processing is performed, and the frequency spectrum of the waveform data (waveform mode WM = 2) for music generation calculation with an equivalent sampling frequency of 12 kHz, which is downsampled at a frequency 1/2 of the recording sampling frequency, is shown. There is. Here, the downsampling of (c) is performed by skipping the waveform data of (b) one by one as described above using FIG. 2 (b). Similarly, the downsampling of (d) is also performed by skipping the band-limited waveform data of (c) one by one. Therefore, in the present embodiment, the original waveform 2^-nOnly (n is a positive integer) can be downsampled at a recording sampling frequency, but by performing the processes of steps S63 and S64 at the same time, downsampling is performed at an arbitrary multiple of the sampling frequency of the original waveform. It is also possible to do.

【0072】

Returning to FIG. 11, if the searched waveform data has a loop portion in the determination in step S62, the waveform is first expanded into waveform data in which the loop portion that repeats multiple times is connected to the attack portion (step S66). Next, after band limitation (step S67), downsampling (step S68) is performed, a new attack part and loop part are cut out from the waveform data generated in this way (step S69), and the waveform is obtained in the same manner as in step S65. Complete and memorize (step S70). The processes of steps S67 and S68 are the same as those of steps S63 and S64, respectively. Here, the above processing is applied to the waveform having the loop portion in the case of the waveform data consisting of the attack portion and the loop portion, because the loop portion has a small number of samples of the waveform data, the original waveform is subjected to the above processing. This is because a low-pass filter of a high order cannot be used when the band is limited as it is, so that sufficient band attenuation characteristics cannot be obtained. According to the above method, even if the waveform data has a loop portion, a high-order low-pass filter can be used, so that LPF processing with less unnecessary noise is possible.

【0073】

In the following step S71, the waveform data stored in the step S65 or S70 is stored (registered) in a predetermined area of the RAM 6 or the hard disk 4, and then the waveform LPF process is terminated.

【0074】

FIG. 13 is a flowchart showing a detailed procedure of the sound source processing subroutine in step S6 of FIG.

【0075】

First, the sound source register (FIG. 2 (d)) is checked in step S81, and it is determined in the following step S82 whether or not there is a new write. If there is no new writing, the process immediately proceeds to step S84, and if there is a new writing, the sound source control preparation process of the sound source channel in which the writing was made is performed (step S83). Specifically, the sound source control preparation process converts the data of channel i (pronunciation channel that has been newly written) into various control data for actually performing waveform calculation, and corresponds to the data of channel i. This is the process of setting the first read address of waveform data.

【0076】

In the following step S84, the calculation time is managed. That is, the time at which the reading of the waveform data currently being reproduced is started is set as the calculation start time tBC (see FIG. 4) of the next waveform data so that the reading of the reproduced waveform data in the reproduction unit (DMA control unit 9) is not interrupted. specify. Further, in step S85, it is determined whether or not the calculation start time tBC has been reached, and if not, the present process is immediately terminated.

【0077】

When the calculation start time is reached, first, channel control is performed to determine the calculation order and the channel to be muted according to the musical tone to be generated by each sounding channel (step S86). The reason for determining the operation order is to process the operation with the highest importance first in consideration of the case where the operation is not completed by the time when the operation should be completed (at this time, the operation is terminated). Next, the data prepared in step S83 is expanded on the time axis to prepare for the waveform calculation (step S87), and the sounding channel number of the calculation sequence No. 1 is set as the parameter i (step S88).

【0078】

Next, in step S89, the value of the calculation mode CM (i) is determined, and when CM (i) = 0, 128 samples of channel i are generated and added to buffer 0 (step S90). When CM (i) = 1, 64 samples of channel i are generated and added to buffer 1 (step S91), and when CM (i) = 2, 32 samples of channel i are generated. Is generated and added to buffer 2 (step S92). Here, the processing of each step S90 to 92 corresponds to the F number FN of each channel i (see FIG. 2 (d)) instead of simply adding the musical sound data (waveform sample data) in the corresponding buffers. Read address update control, read and interpolate waveform sample data according to the read address from the waveform data storage area shown in Fig. 2 (b), and further tone color and volume for the waveform sample obtained by interpolation. It is added after processing such as processing to make a musical sound type sample by processing an envelope or the like. The above processing is repeated for each sample of the musical tone type data generated by one generation processing. For example, in channel i of calculation mode CM (i) = 0, it is repeated until 128 samples of musical sound waveforms are completed. Here, the waveform data to be read is the waveform data designated by the waveform designation data described above. The waveform data is selected and specified corresponding to the waveform name WN (p) of the part p to which the musical tone being generated on the sounding channel i belongs and the calculation mode CM (i). If the recording sampling frequency of the waveform data currently being generated for the musical tone signal is not compatible with the calculation mode (equivalent sampling frequency) set for channel i, the above-mentioned F number FN The value of the F number FN is corrected by the term in the latter half of the calculation formula, in order to generate a musical tone with the specified pitch even in such a case. For details of the processing in steps S90 to 92, see FIG. 13 and Ming in the drawing of Japanese Patent Application No. 7-197923.

【0079】

As described above, even if the waveform of the optimum waveform mode for each calculation mode is not selected, by correcting the F-number FN, a musical tone sample with the specified pitch SP can be generated in any calculation mode. If the value of the F number FN deviates significantly from the reference value "1" due to the correction or the like, a problem arises in the quality of the generated musical tone. For example, if the correction causes the F number FN of that channel to become very large (eg FN> 2), the high frequency component of the waveform data read from the waveform data storage area will exceed half the equivalent sampling frequency. , The folding noise is generated in the waveform sample obtained by interpolation. Conversely, if the correction results in a very small F number FN (eg FN <0.5), the waveform sample obtained by interpolation will contain less than a quarter of the equivalent sampling frequency harmonic content. The quality corresponding to the time density of the harmonic sample generation set for the channel cannot be obtained. Therefore, as the waveform data indicated by the waveform designation data, that is, the waveform data read by the music sound generation of each channel i, the waveform data of the waveform mode suitable for the calculation mode CM (i) of each channel (for example, WM = 1 if CM = 1). It is desirable to select and set the waveform data of.

【0080】

Next, it is determined whether or not channel i is the last channel, that is, whether or not all the calculations of the channels to be waveform-calculated have been completed (step S93). After changing the channel number of (step S94) to channel i, the process returns to step S89 and the above-described processing is repeated. On the other hand, in the determination of step S93, when all the channels to be calculated are completed, the process proceeds to step S95.

【0081】

In step S95, as described in FIG. 3, 64 samples generated in buffer 1 and 32 samples of ultrasonic data generated in buffer 2 are interpolated (oversampled) to obtain 128 samples, respectively. It is stored in buffer 1'and buffer 2'as ultrasonic data, and in step S96, 128 samples of ultrasonic data obtained by adding 128 each of buffers 0, 1', 2'are buffered. Store in 0.

【0082】

In the following step S97, the musical sound form data stored in the buffer 0 is subjected to reverb processing to give a reverberation effect, and in step S98, 128 samples of the buffer 0 are reserved for playback in the playback section, and then the main sound source is used. End the process.

【0083】

As described above, in the present embodiment, a plurality of arithmetic modes for arithmetic processing for generating musical tones are provided, and the user can arbitrarily select from these modes. Therefore, the number of pronunciations is increased according to the purpose of use of the user. It can take either form of emphasis or quality.

【0084】

In addition, since the calculation mode is set for each part, the musical sound of the part having a large audible effect can be generated with high quality, and the limited calculation ability can be fully utilized.

【0085】

Furthermore, from the waveform data of the high recording sampling frequency for the high quality calculation mode (mode 0) of the generated musical tone, the waveform data of the low recording sampling frequency according to the low quality calculation mode (mode 1 or 2) of the generated musical tone is LPF. It is created by processing, and from the waveform data created in this way, the waveform data is selected according to the calculation mode selected by the user to generate a musical tone, so the calculation is performed at a low equivalent sampling frequency. In the channel, it is possible to generate a waveform of the same tone as a musical tone calculated at a high equivalent sampling frequency without generating wrapping noise, although the quality is lowered.

【0086】

In addition, since the waveform data is automatically selected according to the calculation mode selected by the user, the waveform data having a frequency component over a wide band with a high recording sampling frequency is used in the channel having a high equivalent sampling frequency. For channels with a low equivalent sampling frequency, waveform data having a narrow band frequency component with a low recording sampling frequency can be used, so that it is not necessary to change the waveform designation in the tone color data.

【0087】

In the present embodiment, when the calculation mode CM (p) having a lower number of samples is selected, if there is no waveform data having the waveform mode WM corresponding to the calculation mode CM (p), the waveform data is used. Waveform data of waveform mode WM different from the calculation mode CM (p) is selected, and a music signal is generated based on this waveform data, but it is not limited to this, and the selected calculation mode CM (p) is used. It automatically detects whether or not the waveform data of the corresponding waveform mode WM exists, and if it does not exist, it automatically shifts to the waveform LPF processing of FIG. 11 and the waveform data of the calculation mode CM (p). May be created and a music signal may be generated based on the created waveform data.

【0088】

Further, in the present embodiment, the waveform data WDn (n = 1, 2 ...) is subjected to waveform LPF processing so as to obtain waveform data WDn , WDn with a recording sampling frequency lower than that. However, the waveform data WDn and WDn may be prepared by a method other than the waveform LPF processing. For example, if recording is performed at a low recording sampling frequency from the beginning, the waveform data WDn and WDn can be directly obtained. Then, not only the waveform data WDn is stored in the hard disk 4, but also the waveform data WDn and WDn are stored in the hard disk 4. When those waveforms are stored in the hard disk 4, the waveform data WDn and WDn can be read out from the hard disk to RAM 6 without performing waveform LPF processing on the waveform data WDn.

【0089】

In the present embodiment, only one calculation mode CM (p) can be selected for each part, but the present invention is not limited to this, and even within a plurality of channels for sounding the same part, for example, for each channel. The calculation mode may be automatically changed according to the volume, pitch, and the like.

【0090】

Further, in the present embodiment, the invention has been realized as software executed by a CPU, but the invention is not limited to this, and can be realized by a general arithmetic unit that operates in a program such as a DSP (digital signal processor). .. Further, the musical sound generator of the present embodiment may be realized by a general-purpose personal computer or a dedicated device.

【0091】

[Effect of the invention]

As described above, according to the present invention according to claims 1 and 5, phase information is generated according to the input pitch and control information, and based on this phase information, the above-mentioned is performed every predetermined period. A musical tone sample corresponding to the input pitch information is generated for each channel with the number of waveform samples per unit time according to the input control information, and a musical tone is generated based on the generated musical tone sample. Therefore, it is possible to take either a form of emphasizing the number of pronunciations or an emphasis on quality, depending on the purpose of use of the user.

【0092】

Further, according to the configuration of the invention according to claims 2 and 6, the musical tone type sample of the channel corresponding to the input performance information has a unit time according to the set control information of the part to which the performance information belongs. It is generated with the number of waveform samples per hit, and musical tones are generated based on this generated musical sound sample, so it is possible to generate high-quality musical tones for parts that have a large audible effect, maximizing limited computing power. It can be utilized.

【0093】

Further, according to the configuration of the invention according to claims 3 and 7, a music sound generation calculation based on the waveform data stored in the waveform memory is executed according to the input performance information, and the generated control information is struck. A waveform sample is generated based on the number of waveform samples per unit time, and in the music generation calculation, a music generation calculation is performed by selectively using different waveform data in the waveform memory according to the generated control information. Since the musical sound is generated based on the music sound type sample generated in this way, the waveform data having the frequency component over a wide band is used for the high sampling frequency channel, and the narrow band is used for the low sampling frequency channel. Waveform data having the frequency component of can be used, so that it is not necessary to change the designation of the waveform in the tone color data.

【0094】

Further, according to the configuration of the invention according to claim 4, since the waveform data corresponding to the number of waveform samples per unit time of the musical sound type sample is used, the folding noise can be reduced. Furthermore, when a waveform sample is generated with two different waveform samples per unit time, the waveform data obtained by converting the waveform data suitable for the number of waveform samples per unit time in one unit is converted into the waveform data in the other unit time. Since it is used to generate a waveform with a large number of waveform samples, it is possible to generate an orthophonic sample that sounds the same tone even with different numbers of waveform samples per unit time.

[Simple explanation of drawings]

[Figure 1]

It is a block diagram which shows the structure of the musical tone generator which concerns on one Embodiment of this invention.

[Figure 2]

It is a figure which shows the structure of the tone color data and waveform data stored in the RAM of FIG. 1, and the structure of an input buffer and a sound source register set in a RAM.

[Fig. 3]

It is a figure which shows the structure of the output buffer set on the RAM of FIG.

[Fig. 4]

It is a figure for demonstrating the outline of the musical tone generation processing performed by the musical tone generating apparatus of FIG.

[Fig. 5]

It is a flowchart of the main routine executed by the CPU of FIG.

[Fig. 6]

It is a flowchart of MIDI reception interrupt processing.

[Fig. 7]

It is a flowchart of note-on-event processing when data is received via the MIDI interface of FIG.

[Fig. 8]

It is a flowchart which shows the procedure of the part tone color selection process when a part tone color selection switch is pressed.

[Fig. 9]

It is a flowchart which shows the procedure of the part mode selection process when a part mode selection switch is pressed.

[Fig. 10]

It is a flowchart which shows the procedure of the part waveform selection processing when a part waveform selection switch is pressed.

[Fig. 11]

It is a flowchart which shows the procedure of the waveform LPF processing when the waveform LPF switch is pressed.

[Fig. 12]

It is a figure which shows an example of the frequency characteristic of the waveform data generated by the waveform LPF processing of FIG.

[Fig. 13]

It is a flowchart which shows the detailed procedure of the sound source processing subroutine of step S6 of FIG.

[Explanation of symbols]

1 CPU (control information input means, musical tone sample generation means, control information generation means) 2 Keyboard (pitch information input means, performance information input means, controls, control information setting means) 5 ROM 6 RAM (waveform memory, storage means) 8 MIDI interface (performance information input means) 9 Direct memory access control unit (musical tone generating means) 10 DA converter (musical tone generating means)