JP3285137B2 - Musical sound generating apparatus and musical sound generating method, and storage medium storing program according to the method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a musical sound generating apparatus and a musical sound generating method for generating musical sounds by executing predetermined musical sound generating software on an arithmetic processing unit, and a storage medium storing a program according to the method. .

[0002]

2. Description of the Related Art Conventionally, there has been known a tone generator which generates a tone by executing predetermined tone generating software on a general-purpose arithmetic processing unit such as a CPU. This is a system called a software sound source or a software sound source. In recent years, software sound sources have also been required to have higher functions, and there is a demand for faster processing.

On the other hand, in recent years, CPUs having instructions capable of executing a plurality of operations in parallel with one instruction have been proposed.
For example, a CPU having an extended instruction set called “MMX” by Intel Corporation is used.

[0004]

According to the conventional concept of parallel processing, when parallel processing is applied to image processing, adjacent pixel data (8 bits) are grouped together and processing for a plurality of pixels is performed in parallel. Was. On the other hand, when audio processing and musical sound generation processing are performed in parallel, a plurality of temporally continuous samples (16 bits) are grouped, and processing such as amplitude control and filter processing is parallelized.

It is also conceivable to use a CPU having an extended instruction set capable of executing a plurality of operations in parallel with one instruction. For example, FIG. 5 is a block diagram showing an algorithm for effect processing of a soft sound source, and FIG.
(A) is a detailed view of the circuit APn of FIG. 5, and (b) of FIG.
3 is a detailed diagram of the circuit CFn. In FIG. 6, there is a portion where two input data are each multiplied by a predetermined coefficient and added. For example, (m4, m5, a5) and (m6, m7, a6) in FIG. 6A, or (m9, m1) in FIG.
0, a7). The operations of these parts can be executed with one instruction by using a CPU having an extended instruction for executing the processing of multiplying and adding the two input data by a predetermined coefficient, respectively, with one instruction. Is possible. However, since such speeding up is basically performed by devising the calculation within one processing algorithm, there is a part that cannot be completely parallelized, and the effect of the parallelization is reduced. Not available to the fullest.

In the processing for generating a musical tone waveform, address generation, envelope generation, filtering, etc.
A process of obtaining a current waveform sample using a past waveform sample is included. This is a process of generating a current address based on an address one sampling cycle earlier in address generation, and a process of generating a current envelope value based on the immediately preceding envelope value in envelope generation. The filtering process is a process of performing a filtering operation based on the values of the past waveform samples and the current input waveform samples to generate and output output waveform samples. As described above, since the current waveform sample is obtained using the past waveform sample, it is difficult to parallelize the processing for obtaining the adjacent waveform samples in time series.

The present invention provides a musical tone generating apparatus and a musical tone generating method capable of performing a waveform generation operation at a higher speed in a software tone generator realized by using a CPU capable of executing a plurality of operations in parallel with one instruction. And a storage medium storing a program according to the method.

[0008]

In order to achieve this object, an invention according to claim 1 is a tone generator for generating a tone waveform by executing predetermined tone generation software on a general-purpose arithmetic processing unit. The arithmetic processing unit is provided with an extended instruction set capable of executing a plurality of operations in parallel with one instruction, and generates waveforms of a plurality of channels by using the extended instruction in n channels (n is 2
The above-mentioned integer) are performed in parallel, and the generated waveform data of the plurality of channels are mixed, and the processing of the part capable of parallel processing in the algorithm of the effect processing applied to the mixed waveform data is performed in parallel. It is characterized by the following.

According to a second aspect of the present invention, there is provided a musical tone generating method for generating a musical tone waveform by executing predetermined musical tone generating software on a general-purpose arithmetic processing unit. Is provided with an extended instruction set capable of executing the operations in parallel, and by using the extended instructions, waveform generation for a plurality of channels is performed in parallel for n channels (n is an integer of 2 or more). The mixed waveform data of the plurality of channels is mixed, and processing of a part which can be processed in parallel in an algorithm of an effect processing applied to the mixed waveform data is performed in parallel.

According to a third aspect of the present invention, there is provided a musical tone generating method for generating musical tones for a plurality of channels, wherein a step of supplying performance information; a step of generating a timing signal at predetermined time intervals; For each occurrence,
And generating waveform data for a plurality of channels in accordance with the performance information. In the generating step, processing for the plurality of channels is performed in parallel for n channels (n is an integer of 2 or more) in parallel. Generating and outputting the waveform data for a plurality of samples, and generating the waveform data by one sample every sampling period.
/ A converter and converting the analog waveform into an analog waveform.

According to a fourth aspect of the present invention, there is provided a musical tone generating method for generating a musical tone waveform by executing predetermined musical tone generating software on a general-purpose arithmetic processing unit. Is provided with an extended instruction set capable of executing the operations in parallel, and the first instruction of the waveform generation processing of a plurality of channels is provided by using the extended instruction.
Is performed in parallel for each of n channels (n is an integer of 2 or more), and the second process is performed in parallel for each of m channels (m is an integer of 2 or more other than n). Mixing the waveform data of the plurality of channels, further performing in parallel the processing of a part capable of parallel processing in an algorithm of an effect processing applied to the mixed waveform data, and outputting the waveform data to which the effect is given. It is characterized by. The first process is, for example, generation of an address of waveform data, and the second process is, for example, a volume control process.

According to a fifth aspect of the present invention, there is provided a storage medium storing a program relating to a musical sound generating method for generating a musical sound waveform by executing predetermined musical sound generating software on a general-purpose arithmetic processing device. The processing device is provided with an extended instruction set that can execute a plurality of operations in parallel with one instruction, and the program uses the extended instruction to generate waveforms of a plurality of channels on n channels (where n is 2). The above-mentioned integer) are performed in parallel, and the generated waveform data of the plurality of channels are mixed, and the processing of the part capable of parallel processing in the algorithm of the effect processing applied to the mixed waveform data is performed in parallel. Characterized in that:

According to a sixth aspect of the present invention, there is provided a storage medium storing a program relating to a musical tone generating method for generating musical tones for a plurality of channels, the program comprising: a step of supplying performance information; Generating a timing signal, and for each occurrence of the timing signal,
And generating waveform data for a plurality of channels in accordance with the performance information. In the generating step, processing for the plurality of channels is performed in parallel for n channels (n is an integer of 2 or more) in parallel. Generating and outputting the waveform data for a plurality of samples, and generating the waveform data by one sample every sampling period.
/ A converter and converting the analog waveform into an analog waveform.

According to a seventh aspect of the present invention, there is provided a storage medium storing a program relating to a musical sound generating method for generating a musical sound waveform by executing predetermined musical sound generating software on a general-purpose arithmetic processing device. The processing device is provided with an extended instruction set capable of executing a plurality of operations in parallel with one instruction, and the program uses the extended instruction to execute a first one of a plurality of channels of waveform generation processing. The processing is performed in parallel for each of n channels (n is an integer of 2 or more), and the second processing is performed in parallel for each of m channels (m is an integer of 2 or more other than n). The waveform data for the plurality of channels is mixed, and the processing of the part capable of parallel processing in the algorithm of the effect processing applied to the mixed waveform data is performed in parallel to provide the effect. And outputs the waveform data.

[0015]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing an electronic musical instrument to which a musical sound generating apparatus and a musical sound generating method according to the present invention are applied. The electronic musical instrument has a central processing unit (CPU) 10
1, read only memory (ROM) 102, random access memory (RAM) 103, drive device 104,
Timer 106, network input / output (I / O) interface 107, keyboard 108, display 10
9, hard disk 110, sampling clock (F
s) A generator 111, a sound I / O 112, a direct memory access (DMA) controller 114, a sound system 115, and a bus line 116 are provided.

The CPU 101 controls the operation of the entire electronic musical instrument. The CPU 101 has an extended instruction set that can execute a plurality of operations in parallel with one instruction. Specifically, it is assumed that the data handled by the 64-bit width register inside the CPU is four 16-bit width data, and an instruction for simultaneously processing the four 16-bit width data;
And 64-bit data into 32-bit data 2
And has an instruction to handle two pieces of the 32-bit data at the same time.

The ROM 102 stores a control program (such as a program for a soft sound source including a soft effector) executed by the CPU 101 and various constant data.
Further, on the ROM 102, waveform data (waveform sample data sampled at a predetermined rate) used when the CPU 101 executes a program of a software tone generator to generate a musical tone is prepared. The control program, various constant data, and the waveform data are stored in the ROM 10
Instead of 2, the one prepared on the RAM 103 can be used. In this case, data such as a control program is supplied from an external storage medium 105 such as a CD-ROM or a network I / O 107, expanded on the RAM 103, or stored on the hard disk 110. RAM
The 103 is provided with work areas such as various registers, a waveform generation buffer, and a reproduction buffer. The drive device 104 is for inputting and outputting various data to and from an external storage medium 105 such as a floppy disk (FD) or a flash card. Hard disk 110
Is a storage device used for recording various data.

The timer 106 generates a timer clock signal for causing the CPU 101 to generate a timer interrupt at predetermined intervals. The network I / O interface 107 is an interface for exchanging various data via an external public line or a LAN (local area network). The keyboard 108 is a typing keyboard used for inputting various information.
The display 109 is a display for displaying various information. The user can perform various setting operations and instruction operations on the electronic musical instrument using the keyboard 108 and the display 109.

The Fs generator 111 has a sound I / O 11
2 to generate a sampling clock having a frequency Fs. The sound I / O 112 is a codec (CODE
C). Sound I / O112
Has an analog / digital (A / D) conversion function and a digital / analog (D / A) conversion function.
An analog tone signal from an external input 113 is input to a D input terminal, and a sound system 115 is input to a D / A output terminal.
Is connected. The sound I / O 112 has two FIFO (First In First Out) stack areas inside. One is an input FIFO for holding digital waveform data input via an A / D input terminal, and the other is an output FIFO for holding digital waveform data output via a D / A output terminal.

A sound I / O 112 from an external input 113
The analog tone signal input to the A / D input terminal is A / D-converted according to a sampling clock having a frequency Fs, and is written to an input FIFO (which may be compressed by an ADPCM method or the like as necessary). If the input FIFO contains waveform data, the sound I / O 112
A request for processing the input waveform data is issued to the A controller 114. The DMA controller 114 stores the data of the input FIFO in the RAM 10 in response to the processing request.
3 to a predetermined area (recording buffer area secured in advance). This DMA controller 114
Is performed by the DMA controller 114 securing the bus line 116 by interrupting the CPU 101 (hardware interrupt) every sampling clock Fs. The CPU 101 is a DMA controller 1
It is not conscious of securing the bus line 116 by 14.

On the other hand, the output FI in the sound I / O 112
When waveform data exists in the FO, the waveform data in the output FIFO is D / D for each sampling clock Fs.
A-converted, sound system 1 via D / A output terminal
15 and the sound is emitted. When the waveform data of the output FIFO is output, the output FIFO has an empty space, so that the sound I / O 112 issues a request to the DMA controller 114 to acquire the waveform data. The CPU 101 generates waveform data to be output in advance and stores it in a reproduction buffer (PB0, PB1 described later) on the RAM 103.
A request for reproducing the waveform is issued to the DMA controller 114. The DMA controller 114 secures the bus line 116 by interrupting the CPU 101 for each sampling clock Fs, and transfers the waveform data of the reproduction buffer on the RAM 103 to the output FIFO of the sound I / O 112.
Transfer to The transfer of waveform data by the DMA controller 114 is transparent to the CPU 101.
The waveform data written to the output FIFO is sent to the sound system 115 for each sampling clock Fs and emitted as described above.

As for the software sound source, the CPU 101
However, when viewed from the side of using the software sound source, the tone generation software is registered as a driver, and after starting the driver, an API related to a predetermined software sound source is executed. (Appl
MIDI (Musical Instruments digital Interface) representing various performance inputs
The procedure is such that an event message is output and various processings relating to musical sound generation can be performed on the software sound source. CP
U101 is a general-purpose arithmetic processing unit, and also performs processing other than the software sound source, such as processing for giving a performance input to the API, that is, processing for outputting a MIDI event to the API. The process in which the CPU 101 provides a performance input to the API includes, for example, outputting a performance input generated in real time in response to an operation of a performance operator (keyboard or the like) (not shown) to the API, a network I / O1.
07, a performance input corresponding to a MIDI event input in real time is output to the API.
3, a sequence of MIDI events is prepared (the data on the external storage medium 105 or the hard disk 110 may be used, or the data input via the network I / O 107 may be used). It is a process of outputting to the API as a performance input.

Referring to FIG. 2, the principle of generating a musical tone of a soft sound source will be described. In FIG. 2, each section of S1 to S4 is
A time frame as a unit for performing reproduction of a predetermined number of samples (2 × 128 samples) is shown. A downward arrow written on the line of “performance input” indicates that a performance input was made at that time. The performance input means that various MIDI events such as note-on, note-off, after-touch, and program change are input to the API relating to the above-described software sound source. In the example of FIG.
This means that there were three performance inputs at 1, two at S2, and one at S3. The software sound source can simultaneously generate a plurality of musical tones for a plurality of channels.
Each tone is controlled by a plurality of channels of software tone registers prepared above. The soft sound source is
When a note-on event is input as a performance input,
The tone generation is assigned to the software tone generator register, and various data for controlling the tone generation of the channel and note-on are written in the software tone generator register corresponding to the assigned channel. When a note-off event is input as a performance input, note-off is written to the software tone generator register corresponding to the corresponding channel. Performance input other than note-on and note-off (for example, changing aftertouch)
Similarly, data corresponding to the performance input is written to the software tone generator register corresponding to the corresponding channel.
The data written to the software tone generator register in a certain time frame is always used for the waveform generation calculation from the next time frame regardless of the type of data.

Rectangle 2 of "Waveform generation by CPU" in FIG.
Numerals 01 to 204 denote sections in which the CPU 101 executes a waveform generation calculation including an effect. In this waveform generation calculation, a tone waveform for a plurality of channels is generated based on data for a plurality of channels set in a software tone generator register. The soft tone generator register is rewritten according to the performance input, while the soft tone register holds data written in the past during the period when there is no performance input. Therefore, in the sections 201 to 204 of each waveform generation, a waveform generation calculation corresponding to the performance input detected in the immediately preceding or further previous frame is executed. Since a frame interrupt occurs at the timing of switching frames, the waveform generation operation (FIG. 7A described later) in each frame is executed with the frame interrupt as a trigger.

For example, with respect to three performance inputs detected in the frame S1, a waveform generation operation is performed in the section 202 triggered by an interruption of the first frame of the next frame S2. The CPU 101 calculates RA
Waveform data (data obtained by accumulating a plurality of channels and adding effects) is generated in a waveform generation buffer on M103. The generated waveform data is stored in the RAM 10
3 in the reproduction buffer area. As this reproduction buffer area, two buffer areas PB0 and PB1 of the same size prepared at successive addresses (two are collectively called a double buffer) are used. Writing to the reproduction buffer area is performed in step 807 of FIG.
Perform in. The buffers PB0 and PB
And 1 are alternately used. For example, the frame S1
Is written in the reproduction buffer area PB0 on the RAM 103, and the waveform data generated in the section 202 of the frame S2 is written in the reproduction buffer area PB0.
1, the waveform data generated in the section 203 of the frame S3 is written in the reproduction buffer area PB0, the waveform data generated in the section 204 of the frame S4 is written in the reproduction buffer area PB1, and so on.
Write waveform data to 0 and PB1.

The reading and reproducing of the waveform data written in the reproducing buffers PB0 and PB1 is performed in the section of the frame next to the frame in which the waveform is generated, triggered by the frame interruption, as shown in "Reading and reproducing" of FIG. That is, the waveform data generated in frame S1 and written in PB0 is the next frame S2, and the waveform data generated in frame S2 is PB0.
The waveform data written in B1 is the next frame S3, the waveform data generated in frame S3 and written in PB0 is the next frame S4, and so on. The waveform data of PB0 and PB1 are alternately read and reproduced. . In the read reproduction, the DMA controller 114 interrupts the CPU 101 for each sampling clock Fs,
This is performed by transferring the waveform data of the reproduction buffer (PB0 or PB1 designated) on M103 to the output FIFO of the sound I / O 112. The frame interruption occurs when a return occurs when reading the playback buffers PB0 and PB1 in a loop (that is, when the playback of PB1 is completed) and when the intermediate point of the loop reading passes (that is, when PB1 is read).
0 at the end of playback). The frame interrupt is a hardware interrupt generated by the sound I / O 112, and indicates a point in time when the reproduction of one frame is completed. That is, the sound I / O 112 determines the number of transfer samples (DMAC 11
4, the output FI of the sound I / O from PB0 and PB1
The number of samples transferred to the FO is counted, and a frame interrupt is generated every time half of the number of samples of the reproduction buffer (the size of PB0 and PB1 combined) is transferred.

The software tone generator can simultaneously generate a plurality of musical tones for a plurality of channels. In particular, in this embodiment, since the CPU 101 for realizing the soft sound source has a function of performing the same processing for a plurality of data with one instruction, the processing of a plurality of channels for which a waveform is to be generated is performed in parallel using this function. The execution speeds up the processing. More specifically, the waveform generation processing for one channel includes an address generation processing, a waveform sample reading processing, an interpolation processing, a filter processing, a volume control & accumulation processing, and the like. I try to do it at the same time.

FIG. 3 is a diagram showing how data for four channels is packed. According to the extended instruction set of the CPU 101, 16 bits × 4 data are collected (packed) and set in a 64-bit register.
Operations such as multiplication, addition, and subtraction can be performed simultaneously with 16 bits × 4 data of a 4-bit register. FIG. 3 is an example of multiplication. In this way, the processing of a plurality of channels is divided into four
The processing for the channels is performed simultaneously. Run simultaneously 4
Channels are defined as ｛4 × (n
-1) +1} to {4 × n} channels.

FIG. 4 shows an example of an algorithm of the tone color filter processing of each tone generation channel. As can be seen from FIG. 4, the timbre filter processing is substantially composed of addition and multiplication. Therefore, if the above-mentioned extended instruction for processing 16 bits × 4 data in parallel is used, tone color filter processing for 4 channels can be executed simultaneously in parallel. The delay circuits d1 and d2 may write 16-bit × 4 64-bit data at a predetermined address and read the data later.

FIGS. 5 and 6 show an example of an algorithm for effect processing. The effect processing is not performed for each channel, but the result of waveform generation and channel accumulation for each channel is input (in FIG. 5, XL, X
R, XX, three systems).
5 and 6 are executed as one processing algorithm. Therefore, in such an effect processing, as described in the conventional example, the speed is increased by parallelizing a part that can be processed in parallel in the calculation in the processing algorithm. For example, (m4,
m5, a5) and (m6, m7, a6) and FIG.
That is, (m9, m10, a7) in (b) is performed by one instruction.

In the case where an effect processing for performing the same processing on the left and right stereo outputs is used instead of the algorithms shown in FIGS. 5 and 6, an extended instruction for simultaneously processing 32 bits × 2 data is used. The effect processing of the left and right outputs of the stereo can be performed simultaneously.

Next, FIGS. 7A and 7B and FIG.
The processing procedure of the CPU 101 of the electronic musical instrument will be described with reference to the flowchart of FIG.

FIG. 7A shows a processing procedure of a main routine related to the software sound source in the control program of the CPU 101. This main routine is registered in the OS (Operating System) as a driver of the software sound source. When generating a tone using a software sound source, the driver, that is, the process of FIG. 7A is first activated, and the API related to the software sound source is enabled. FIG.
In (a), various initial settings are made in step 701. In this initial setting, the reproduction buffers PB0 and PB1 are cleared to 0, and the sound I / O 112 and DMA are cleared.
Instruct C114 to start the processing of alternately reading and reproducing the reproduction buffers PB0 and PB1 described with reference to FIGS. Next, at step 702, it is checked whether there is any activation factor, and if there is an activation factor at step 703, the process proceeds to step 704. If there is no activation factor, the process returns to 702 again. The reception of the activation factor in steps 702 to 704 corresponds to the reception of an instruction to the API related to the software sound source.

In step 704, the type of the activation factor that has occurred is determined, and the process branches to each process. If the activation factor is the input of a MIDI event, step 70
5 is performed, and then step 702 is performed again.
Return to This MIDI event input and step 705
MIDI processing corresponds to the reception of the performance input of FIG. In step 704, the activation factor is the frame interrupt (1
If (frame reproduction is completed), the waveform generation processing of step 706 is performed, and thereafter, the flow returns to step 702 again. The frame interrupt is a hardware interrupt generated every time the sound I / O 112 completes one frame reproduction. The waveform generation processing of step 706 is performed by
This is a process for performing the waveform generation calculations shown in FIGS. In this waveform generation processing, one frame of waveform data (a number of waveform samples corresponding to a half of the combined size of the reproduction buffers PB0 and PB1) is generated at a time, and is alternately written to the reproduction buffer PB0 or PB1. If the activation factor is another request in step 704, a process corresponding to the activation factor is performed in step 707, and the process returns to step 702 again. In particular, sound I / O11
2, instructions for sampling the external input 113, instructions for changing the setting of the algorithm of the soft effector, and instructions for setting weighting coefficients for defining the signal transmission levels of the three series output from the waveform generation to the effect processing. If so, corresponding processing is performed in the other processing of step 707. If the activation factor is a request for terminating the software sound source in step 704, termination processing is performed in step 708, and the processing is terminated.

FIG. 7B is a flowchart showing step 70 in FIG.
5 shows a procedure of a note-on event process executed when a note-on event is input among the MIDI processes of No. 5; First, in step 711, the MIDI channel, note number, and velocity of the input note-on event are set in registers MC, NN, and VE, respectively. Next, at step 712, sound channels are assigned. In step 713, information (note number NN, velocity VE, etc.) necessary for sound generation is set in the software sound source register of the allocated sound channel. Further, in step 714, note-on is written into the software tone generator register of the assigned tone generation channel to issue a tone generation start instruction, and the process ends.

Although not shown, the same applies to MIDI event processing such as note-off and other performance inputs. That is,
If the note is off, note-off is set in the software tone generator register corresponding to the corresponding tone generation channel, and if another performance input is made, the data corresponding to the performance input is written in the soft tone generator register corresponding to the corresponding tone generation channel.

FIG. 8A is a flowchart showing step 70 in FIG.
6 shows the procedure of the waveform generation processing of No. 6. First, in step 801, calculation preparation is performed. This is a process of referring to the software tone generator register to recognize the channel on which the waveform generation calculation is to be performed, and recognizes which of the reproduction buffer PB0 and PB1 should be set for the waveform of one frame generated by the current waveform generation calculation. Preparation for processing, such as clearing all areas of the waveform generation buffer to zero. Next, "1" is set in the work register n in step 802, and the flow advances to step 803.

In step 803, ｛4 × (n-1) +
Waveform samples for one frame are generated for four channels {1} to {4 × n}. This processing will be described in detail with reference to FIG. In step 804, it is determined whether there is a remaining channel to be calculated. If there is, in step 805, the value of the work register n is incremented, and the process returns to step 803. This is repeated to generate waveforms for all channels to be calculated. Since waveform generation is performed for four channels at a time, calculation may be performed for channels that are not sounding.
The volume of the channel is controlled to be zero, so that the output musical tone is not affected. Step 804
When the waveform generation for all the channels to be calculated is completed, the process proceeds to step 806. In step 806, effect processing as shown in FIGS. 5 and 6 is performed.

After the effect processing, in step 807,
The reproduction of the generated waveform for one frame is reserved. this is,
Transfer the generated waveform sample to the playback buffer PB0 or PB
1 is a process of copying to one of the two (the one not currently used for reproduction). Since the process of alternately reading and reproducing the reproduction buffers PB0 and PB1 has already been started in step 701, the sound of the waveform is reproduced in the next frame simply by copying the generated waveform to the one not used for reproduction. Be executed.

FIG. 8B is a diagram showing a step 80 in FIG.
3 shows the procedure of one-frame waveform generation processing for four channels of No. 3. In step 811, the address generation processing for the first two channels of the four channels is performed, and in step 812, the address generation processing for the remaining two channels is performed. The address generated here is a read address of the above-described waveform data (prepared on the ROM 102 in this example). Since the address to be generated is not sufficient with a 16-bit width, the address is generated for each of two channels. Since the parallel processing is performed for each two channels, the CPU 101 uses an extended instruction for performing parallel processing of 32 bits × 2 data.

Next, in step 813, waveform samples are read. In the subsequent interpolation processing, linear interpolation using two samples is performed for each channel.
Read out the channels. In step 814, the 4
Interpolation processing (linear interpolation using two samples) is performed on channels in parallel. In step 815, filter processing (FIG. 4) is performed on the four channels in parallel. In step 816, the processes of controlling the volume and accumulating the four channels are performed in parallel. This is accomplished by multiplying the waveform of each channel by a predetermined level control coefficient and accumulating the result to produce three series of outputs (XL, XR, X in FIG. 5).
X). Since there is an accumulation process, there are some portions where the process of step 816 cannot be executed in parallel. In steps 814 to 816, the CPU 101 uses an extended instruction for parallel processing of 16 bits × 4 data in order to perform parallel processing on four channels.

Next, in step 817, it is determined whether or not the generation of waveform samples for one frame (here, 128 sets of XL, XR, and XX are counted as one set) is completed. If not, the process returns to step 811 to generate the next waveform sample. When one frame has been completed, the process ends.

Next, the processing of the DMA controller 114 during reproduction will be described with reference to the flowchart of FIG. During reproduction, a sample request interrupt (hardware interrupt) is issued from the sound I / O 112 every sampling period. In response, the DMA controller 114 performs the processing in FIG. First, at step 821, one sample of the reproduction buffers PB0 and PB1 is sent to the output FIFO of the sound I / O 112. Output FI
The waveform data written in the FO is D / A-converted at each sampling period, sent to the sound system 115 and emitted as described with reference to FIG. Step 8
The "DMAB" at 21 means that the reproduction buffers PB0 and PB0
It is B1. The reproduction buffers PB0 and PB1 are D
Since it can be regarded as a buffer of MA, it is described as DMAB. Next, in step 822, the pointer p is incremented, and the process ends. p is the reproduction buffer PB0, P
This is a pointer for reading one sample from B1.
As described above, one sample is transferred from the reproduction buffers PB0 and PB1 to the sound I / O 112 at each sampling period while the pointer p is advanced.

The pointer p is incremented by 1 so that PB1 is shifted from the first sample of PB0 to PB1.
Is a pointer for sequentially reading up to the last sample of PB0, but after reading the last sample of PB1, PB0
The pointer value needs to be updated to point to the first sample of the. This process is set to be performed automatically by the DMA controller 114.

According to the above-described embodiment of the present invention, the waveform is generated at a predetermined time (frame) cycle longer than the sampling cycle, and the waveform samples for the predetermined time are collectively generated. The overhead is smaller and the processing time can be reduced as compared with the case where the waveform is generated periodically. If the CPU has a cache memory of a plurality of ways, the waveform data on the ROM 103 and the waveform of one frame being generated can be cached continuously and in parallel for a plurality of channels. Can be.

Further, in the waveform generation processing, the address generation is performed in parallel with a small number of channels by increasing the processing bits, and the interpolation and amplitude control are performed in parallel with the large number of channels by shortening the processing bits. I have to. That is, since the number of parallel channels is changed according to the data to be handled, the efficiency of the operation is high and the processing time is shortened.

In step 804 of FIG. 8A, even if there is an uncalculated channel during sound generation, if the generation operation is not likely to end within the generation period, the process returns to step 803 without returning to step 803. You may make it go to.

In the above embodiment of the present invention, 64
Although the bits are processed in parallel as 16 bits × 4 data or 32 bits × 2 data, the present invention is not limited to this, and the parallel processing can be performed with an arbitrary data width and an arbitrary parallel number.

Furthermore, the time length of one frame corresponds to 128 musical tone waveforms in the embodiment of the present invention, but may be longer or shorter. For example, the length may be 64 samples or 1024 samples.

Further, when processing of an LFO or a pitch envelope for controlling effects such as vibrato and tremolo is added to the embodiment of the present invention, if the number of bits of the effect control waveform to be generated is set to 8 bits, the generation processing is performed. As for the portion, eight channels can be processed in parallel.

The present invention may have the following configuration as an embodiment. A storage medium storing a musical tone generating method for generating musical tones for a plurality of channels, the musical tone generating method comprising: supplying performance information; generating a timing signal at predetermined time intervals; And generating waveform data for a plurality of channels in accordance with the performance information. In the generating step, processing for the plurality of channels is performed in parallel for n channels (n is an integer of 2 or more) in parallel. Generating and outputting waveform data for a plurality of samples to be performed; supplying the generated waveform data to a D / A converter one sample at a time in each sampling period, and converting the generated waveform data into an analog waveform.

It should be noted that various related information and operation programs for realizing the above-described musical tone generation method are stored in a storage device such as a hard disk, and the CPU reads them out to a RAM and uses them. -R
OM (Compact Disc-Read Only Memory)
If data stored on a portable storage medium such as a hard disk, a floppy disk, or a magneto-optical disk can be transferred to a storage device such as a hard disk, related information and operation programs can be added (installed, etc.) or updated (version This is useful when uploading. Of course, the data may be directly transferred from the portable storage medium to the RAM.

Further, instead of via a portable storage medium,
Via the communication interface (network I / O)
Related information on a storage device such as a hard disk or an operation program may be downloaded from the communication network side. The communication interface is LA
N (local area network), the Internet,
It is connected to a communication network such as a telephone line, and is connected to a server computer via the communication network. If the storage device (such as a hard disk) of the present device serving as a client does not store related information or an operation program, the device transmits the related information or the operation program to the server computer via the communication interface and the communication network. Send a command requesting Upon receiving this command, the server computer distributes the requested relationship information, operation program, and the like to the apparatus via the communication network. Then, the apparatus receives the distributed related information, the operation program, and the like via the communication interface and stores the information in the storage device, thereby completing the download.

Further, the present apparatus may be realized by a commercially available personal computer in which related information, an operation program, and the like are installed. Of course, also in this case, as a method of distributing data such as related information and an operation program, a method of storing the information in advance in a nonvolatile memory such as a ROM, a method of storing and distributing the data in a portable storage medium, and a method of communicating A method of distributing via an interface is applicable.

[0056]

As described above, according to the present invention, in a soft sound source using an arithmetic processing unit having an extended instruction set capable of executing a plurality of operations in parallel with one instruction, a plurality of the same algorithm parts are provided. Since the channels are collectively processed in parallel, the waveform generation operation can be performed at a higher speed. In addition, compared to the case of using an extended instruction that devises parallel processing by devising in the processing algorithm of each channel, parallelization is performed on multiple channels, so a program that easily parallelizes multiple channels from an algorithm for one channel can be created. Can be made, and the processing speed is greatly improved. Further, according to the present invention, a program according to the above-described musical tone generating method is provided in the form of a storage medium.

[Brief description of the drawings]

FIG. 1 is a block diagram of an electronic musical instrument to which a musical sound generating apparatus and a musical sound generating method according to the present invention are applied.

FIG. 2 is a diagram for explaining the principle of generating musical tones of a soft sound source;

FIG. 3 is a diagram showing how data for four channels is packed.

FIG. 4 is a diagram illustrating an example of an algorithm of a tone color filter process of each sounding channel;

FIG. 5 is a diagram showing an example of an algorithm for effect processing;

FIG. 6 is a detailed view of APn and CFn in FIG. 5;

FIG. 7 is a flowchart of a main routine and a note-on event routine.

FIG. 8 is a flowchart of a waveform generation processing routine, a one-channel waveform generation processing routine for four channels, and a DMAC processing routine;

[Explanation of symbols]

101 central processing unit (CPU), 102 read-only memory (ROM), 103 random access memory (RAM), 104 drive device, 106 timer
107: network input / output (I / O) interface, 108: keyboard, 109: display, 11
0: Hard disk, 111: Sampling clock (Fs) generator, 112: Sound I / O, 114: D
MA (Direct Memory Access) Controller, 11
5 ... Sound system, 116 ... Bus line.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G10H 7/02 G10H 1/02

Claims (7)

(57) [Claims] 1. A musical sound generator for generating a musical tone waveform by executing predetermined musical sound generation software on a general-purpose arithmetic processing device, wherein the arithmetic processing device executes a plurality of operations in parallel with one instruction. A set of possible extended instructions, and using the extended instructions, perform waveform generation of a plurality of channels in parallel for n channels (n is an integer of 2 or more),
A musical sound generating apparatus comprising: mixing the generated waveform data for a plurality of channels; and performing, in parallel, processing of a part that can be processed in parallel in an algorithm of an effect processing performed on the mixed waveform data. 2. A musical tone generating method for generating a musical tone waveform by executing predetermined musical tone generating software on a general-purpose arithmetic processing unit, wherein the arithmetic processing unit executes a plurality of operations in parallel with one instruction. A set of possible extended instructions, and using the extended instructions, perform waveform generation of a plurality of channels in parallel for n channels (n is an integer of 2 or more),
A musical sound generating method comprising: mixing the generated waveform data for a plurality of channels; and performing, in parallel, processing of a part that can be processed in parallel in an algorithm of an effect processing applied to the mixed waveform data. 3. A musical tone generating method for generating musical tones for a plurality of channels, comprising: providing performance information; generating a timing signal at predetermined time intervals; And generating waveform data for a plurality of channels in accordance with the information. In the generating step, processing for the plurality of channels is performed in parallel for n channels (n is an integer of 2 or more) in parallel, and Generating and outputting sampled waveform data; and supplying the generated waveform data to a D / A converter one sample at a time for each sampling period to convert the sampled waveform data into an analog waveform. Musical sound generation method. 4. A musical sound generating method for generating a musical tone waveform by executing predetermined musical sound generation software on a general-purpose arithmetic processing device, wherein the arithmetic processing device executes a plurality of operations in parallel with one instruction. A possible extended instruction set is provided. The extended instruction is used to perform the first processing of the multi-channel waveform generation processing on n channels (n is an integer of 2 or more).
The second processing is performed in parallel for each minute, and the second processing is performed for m channels (m
Is performed in parallel by 2 or more integers other than n), and the finally generated waveform data of the plurality of channels are mixed, and the parallel processing in the effect processing algorithm applied to the mixed waveform data is possible. A tone generating method for performing processing of various parts in parallel and outputting waveform data with an effect. 5. A storage medium storing a program relating to a musical sound generation method for generating a musical sound waveform by executing predetermined musical sound generation software on a general-purpose arithmetic processing device, wherein the arithmetic processing device is provided with one instruction. And an extended instruction set capable of executing a plurality of operations in parallel. The program uses the extended instruction to generate waveforms of a plurality of channels by n channels (n is an integer of 2 or more). The waveform data for the plurality of channels generated are mixed in parallel, and processing of a part capable of parallel processing in an algorithm of an effect processing applied to the mixed waveform data is performed in parallel. Storage medium. 6. A storage medium storing a program relating to a musical sound generating method for generating musical sounds for a plurality of channels, the program comprising: a step of supplying performance information; and a step of generating a timing signal at predetermined time intervals. Generating waveform data for a plurality of channels corresponding to the performance information each time the timing signal is generated. In the generating step, processing for the plurality of channels is performed in parallel on n channels (where n is 2 or more). The waveform data of a plurality of continuous samples are generated and output. The generated waveform data is supplied to the D / A converter one sample at a time for each sampling period. Converting to a waveform. 7. A storage medium storing a program relating to a musical sound generation method for generating a musical sound waveform by executing predetermined musical sound generation software on a general-purpose arithmetic processing device, wherein the arithmetic processing device is provided with one instruction. And an extended instruction set capable of executing a plurality of operations in parallel with the program. The program uses the extended instruction to execute the first processing among n channels (n Is performed in parallel by 2 or more integers), the second processing is performed in parallel by m channels (m is an integer of 2 or more other than n), and finally generated waveforms for the plurality of channels are generated. The data is mixed, and the parallel processing of the effect processing algorithm applied to the mixed waveform data is performed in parallel, and the effected waveform data is output. Storage medium characterized Rukoto.