JPH09311681A - Musical sound producing and processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To efficiently produce musical sound on a widely used computer in real time by carrying out a processing suitable for a widely used computer by carrying out respectively specified musical sound data producing processing and a musical sound data output processing. SOLUTION: At the time when musical sounds are produced in real time on a widely used computer, a program is so composed as to carry out processing suitable for a widely used computer. The program is composed of a control task, a sound source task, and an output task. In the musical sound producing processing, data of the musical sounds at a plurality of sampling times included in a prescribed period for respective prescribed periods longer than the sampling cycles produced collectively and processing to record while ordering time series is carried out. In the musical sound data output processing, the data at respective sampling times recorded by the sound data producing processing is successively read out following the time order for every sampling cycle and sent out.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a tone generation processing method, and more particularly to a tone generation processing method for constructing a synthesizer by software on a general-purpose computer such as a personal computer or a workstation.

[0002]

BACKGROUND OF THE INVENTION Conventionally, in order to generate musical tones, it has been necessary to use dedicated hardware for generating musical tones, so-called electronic musical instruments.

However, in recent years, a general-purpose computer has been equipped with a high-performance CPU and a D / A converter (digital-analog converter) for outputting data processed by the CPU as an analog signal to the outside. Accordingly, the possibility of generating musical sounds in real time on a general-purpose computer without using dedicated hardware has been studied.

More specifically, since tone data must be accurately output to the D / A converter at each sampling period, a process for generating tone data at each sampling period in accordance with a timer may be performed. Are being considered.

In order to realize this processing, it is conceivable to perform processing for generating musical tone data in an interrupt routine in response to a timer interrupt generated at each sampling period. It has been pointed out that the processing is complicated and it is difficult to perform such complicated processing in an interrupt routine.

Alternatively, in order to realize the above-described processing, it is conceivable to perform processing for generating musical tone data by a task started at each sampling cycle. It has been pointed out that it is necessary to perform a process of switching tasks at a timing, which increases the overhead for task switching and shortens the time during which the CPU can actually perform processing.

Generally, one piece of software can be used on various general-purpose computers. However, the capabilities of CPUs mounted on these various general-purpose computers are various and low-capacity. CPU
If you create software according to the
There is a problem in that the computer equipped with U cannot use up the capacity of the CPU.

Further, in a general-purpose computer, a method of executing a plurality of pieces of software at the same time is generally used. In addition to the process of generating musical tones, other processes other than the process of generating musical tones are simultaneously performed. When the processing is performed, the load on the CPU changes every moment according to the processing status of other processing. For this reason, when software is created in accordance with the state where the load on the CPU is the heaviest, there is a problem that the capacity of the CPU cannot be used when the load is light.

Further, generally, at least a part of the process of generating a musical tone by the CPU is repeatedly performed at a constant time interval, but an instruction indicating note-on (start of sounding) / note-off (end of sounding) is issued. , It occurs asynchronously with this repetition interval according to the desired timing of the user. That is, in the process of passing the data to the process of generating the musical tone in accordance with the instruction indicating the note-on / note-off, and in the process of generating the musical tone to which the data is passed, the process of generating the musical tone based on the passed data. However, since the instruction indicating note-on / note-off is asynchronous with the above-mentioned repetition interval, during the process of generating a musical sound, the instruction indicating note-on / note-off is issued in response to the instruction indicating note-on / note-off. If data is rewritten, or if a process of generating a musical tone is performed while rewriting the data in response to an instruction indicating note-on / note-off, there is a problem that noise is generated in the musical tone being generated. It was pointed out that there was a need to improve the data transfer method.

The present invention has been made in view of the above-mentioned various problems, and an object thereof is to:
By performing processing suitable for a general-purpose computer, it is intended to solve the various problems described above, and to provide a musical sound generation processing method capable of efficiently generating a musical sound on a general-purpose computer in real time. is there.

[0011]

In order to achieve the above-mentioned object, the tone generation processing method according to the present invention is such that the tone data at a plurality of sampling points included in the above-mentioned predetermined time is provided at every predetermined time longer than the sampling period. To generate the sound data at once, and to store the music data in a time-ordered manner. Musical tone data output processing for sequentially reading and outputting according to such ordering.

The tone data generation process includes a process for directly handling the tone data and a process for obtaining a control amount for controlling the processing mode of the process for directly handling the tone data. The process for obtaining the amount is the first process for obtaining the next control amount after the predetermined time by calculating only once at the predetermined time, and the previous control amount and the next time for the control amount calculated by the first process. And a second process for interpolating between the control amount and the control amount to obtain the control amount at each sampling time point. The process for directly handling the tone data is performed at each sampling time point obtained by the second process. The musical tone data at each sampling time can be processed based on the control amount.

In the first process, the next values of the plurality of types of control amounts are individually obtained, the values obtained individually are combined to obtain the next control amount, and the second process is performed. It is possible to interpolate between the previous control amount and the next control amount synthesized by the first processing.

The tone generation processing method according to the present invention is
A process of generating a musical sound in a predetermined mode set in advance in the processing means, a process of detecting a load amount of the processing device when the process of generating the musical sound is performed, and a load detected by the detecting process. The processing for determining the musical tone generation mode based on the amount is performed.

Here, the process for determining the musical tone generation mode is based on the allowable load amount of the process for generating the musical tone by the processing unit and the load amount of the processing unit detected by the detecting process. , The generation mode of the musical sound can be determined.

Further, in the tone generation processing method according to the present invention, the processing means automatically generates the tone generation processing in a predetermined mode and the load amount of the processing means when the tone generation processing is performed. And a process of automatically changing the musical tone generation mode on the basis of the load amount detected by the above-mentioned detecting process.

Here, in the detecting process, the time spent for the process of generating the musical tone is measured, and the load amount is calculated based on the measured time and the execution time interval of the process of generating the musical tone. It can be detected.

The musical tone generation mode determined by the musical tone generation mode determination process is at least one of the number of simultaneous musical tones, the effect imparting mode of the effect imparting means, and the sampling frequency. can do.

The musical tone generation mode determined by the musical tone generation mode determination process is at least two of the number of simultaneous musical tone generation, the effect imparting mode of the effect imparting means, and the sampling frequency. The process of prioritizing any of the musical tone generation modes according to a user's instruction and determining the musical tone generation mode is based on the load amount detected by the process of detecting the non-priority musical tone generation mode. Can be decided.

Further, the tone generation processing method according to the present invention includes a process of supplying tone control information in response to receipt of a tone control instruction from the outside, and a process of supplying new tone control information by the above-mentioned supplying process. , The process of holding the newly supplied musical tone control information while holding the musical tone control information supplied last time by the supplying process, and the process of supplying the musical tone control information independently at a predetermined time, to the musical tone control information. In the case where new musical tone control information is supplied by the above-mentioned supplying process between the previously generated process and the presently generated process, in the process of generating a musical sound based on , The musical tone is generated based on the new musical tone control information held by the holding process, and the holding process is changed from the previous generation process to the holding process. If new musical tone control information is not supplied by the above-mentioned supplying process before the process of generating the musical tone, the musical tone is generated based on the previously-supplied musical tone control information held by the holding process. The processing is performed.

Here, in the holding process, a pointer indicating which of the two storage areas holding the previously supplied musical tone control information and the newly supplied musical tone control information is used, and the above-mentioned supplying And a flag indicating that new musical tone control information has been supplied by the processing
When new musical tone control information is supplied by the supplying process, the new musical tone control information supplied by the supplying process is stored in the storage area designated based on the content of the pointer, If the flag is raised and the contents of the flag are referred to every predetermined time, and if the flag indicates that new musical tone control information has been supplied, the flag is cleared and the contents of the pointer are rewritten, The process of generating a musical tone may be a process of generating a musical tone based on the musical tone control information stored in the storage area designated based on the content of the pointer.

When the two storage areas are provided as described above, the storage area for storing the new musical tone control information supplied by the supply processing and the musical tone control information for generating the musical tone are referred to. Storage areas are mutually exclusive. That is, when only one pointer is provided as the above-mentioned pointer, the new musical tone control information supplied by the above-mentioned supply processing is stored in one storage area pointed to by this pointer, and
A tone is generated based on the tone control information stored in the other storage area. Alternatively, as the pointer, a pointer pointing to a storage area for storing the new musical tone control information supplied by the supplying process and a pointer pointing to a storage area to which the musical tone control information is referred in generating a musical tone are separately provided. If it is installed in
The storage areas indicated by these pointers are opposite storage areas.

[0023]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the musical sound generation processing method of the present invention will be described in detail below with reference to the accompanying drawings.

FIG. 1 is a block diagram showing a hardware configuration of a typical general-purpose computer capable of mounting a program according to the musical sound generation processing method of the present invention.

The general-purpose computer shown in FIG. 1 has an extremely general configuration, and a C
PU 12, memory 14, disk controller 16
, A video card 18, an I / O card 20, and a sound card 22 are connected, a hard disk 24 is connected to the disk controller 16, a display device 26 is connected to the video card 18,
The I / O card 20 has a keyboard 28 and a mouse 30.
Is connected. Then, in this computer, a sound signal is output from the sound card 22.

In this general-purpose computer, in addition to the program by the tone generation processing method of the present invention, a sequencer (automatic performance) program is also started at the same time, and the program of the tone generation processing method of the present invention is executed from the sequencer program. MIDI information is passed to
In the program according to the tone generation processing method of the present invention, a tone signal is generated and output according to the delivered MIDI information.

FIG. 2 shows the overall structure of a program according to the musical sound generation processing method of the present invention, which is composed of a control task, a sound source task and an output task.

That is, the control task executes the control program, exchanges MIDI information with the program of the sequencer, and passes the event flag, the sound generation parameter, the steady processing parameter, and the effect parameter to the tone generator task side. Receives voice release information from the side.

The tone generator task performs processing relating to a plurality of voices (tone generation channels) and effects, stores the tone signal generated in the wave buffer, and delivers it to the output task. That is, the processing of each of the plurality of voices is performed, the output of each voice is delivered to the effect according to the effect feed amount of the sounding parameter described later, and the output of each voice to which the effect is not added and each voice to which the effect is added. Will be finally mixed and stored in the wave buffer and delivered to the output task.

In the output task, wave output processing is performed using a wave buffer, and a tone signal is output to the outside. Then, the wave buffer used in the output task is returned to the sound source task.

As described above, the information exchanged between the control task and the sound source task includes an event flag, a sounding parameter, a steady processing parameter, an effect parameter, and voice release information. Hereinafter, such information will be described.

(1) Event Flags Event flags are prepared for the number of voices, and the sound source task receives information (flags) indicating the start and end of sounding of each voice from the control task according to the event flags. The event flag is stored in a shared memory accessible from both the control task and the sound source task.

(2) Sound Generation Parameter As will be described later, the sound source task searches the contents of the event flag every time it is activated, and when the sound generation start flag is found, the sound generation task starts sounding according to the sound generation parameter. The information shown in (a) to (i) below is included for each voice. (A) Waveform reading related parameters (waveform start address, loop related address, pitch) (b) Pitch LFO related parameters (waveform, cycle, delay, etc.) (c) Filter LFO related parameters (waveform, cycle, delay, etc.) (d) ) Amplifier LFO related parameters (waveform, period, delay, etc.) (e) Pitch envelope related parameters (attack,
(Each time and level of decay, sustain, release, etc.) (f) TVF (Time Variable Filter)
er: time-varying filter) Envelope-related parameters (each time and level such as attack, decay, sustain, and release) (g) TVA (Time Variable Ampl)
ifier: time-varying amplifier) Envelope-related parameters (each time and level such as attack, decay, sustain, release, etc.) (h) pan (i) effect feed amount The sound generation parameters including these information are used for the control task and the sound source task. It is stored in a shared memory accessible from both. The memory space for storing the sound generation parameters is provided two by two for each voice. In other words, when the sound source task finds the sound generation start flag in the event flag, the sound source task switches the memory space referred to by the sound source task from the previously referenced surface to another surface, and also refers to the previous memory space. Hand over the face to the control task. Therefore, even when setting the sounding parameter of the same voice number, the control task refers to the sound source task side at the time of sound generation by using another new memory space different from the currently processed surface passed from the sound source task. The sound parameter can be transferred to the sound source task without breaking the parameter area of the currently processed surface. Conversely, it is possible to prevent the sound source task from referring to the parameter area being rewritten on the control task side.

(3) Steady Processing Parameter The steady processing parameter includes the following information (a) to (c) for each voice. (A) pitch offset (b) filter offset (c) amplifier offset These steady state processing parameters are stored in a shared memory accessible by both the control task and the sound source task,
It is for providing information obtained from operators such as pitch vendors and wheels.

These steady process parameters are read every time the sound source task is activated, and are synthesized with the envelope, LFO, etc. generated inside the sound source task during sound generation, and the synthesized data is sounded. It is called a service parameter.

The tone generation service parameters control the waveform reading pitch, the TVF cutoff frequency, and the TVA amplitude.

(4) Effect Parameter The effect parameter is information for effect control stored in the shared memory accessible from both the control task and the tone generator task, and is passed to the effect. This effect parameter is read each time the sound source task is started, and if there is a parameter different from the previously read content, the parameter is reflected in the effect.

The effect parameters differ depending on the specific contents of the effect, and since the specific contents of the effect are outside the scope of the present invention, detailed description thereof will be omitted.

(5) Voice release information The voice release information is returned from the sound source task to the control task, and indicates the number of the voice whose sound has ended. The end of sound generation of each voice is determined based on the envelope of the TVA. This voice release information is passed to the control task by an event instead of using the memory space.

Here, while the event flag, the sound generation parameter, the steady processing parameter and the effect parameter are transferred using the shared memory, the difference in the communication method between the tasks is that the voice release information is transferred by the event. This is because the sound source task is activated by the timer, so asynchronous event information cannot be received from the control task side. By the way, the above-mentioned control task, sound source task and output task are provided with a priority (priority), and a task whose processing is not preferably interrupted by another task or the like has a higher priority. Has been made. In this embodiment, the output task has the highest priority, the sound source task has the second highest priority, and the control task has the lowest priority.

The control task, the sound source task and the output task will be described in detail below with reference to FIGS. 2 and 3.

(1) Control Task The control task is to obtain MIDI information and control the tone generator task, and is activated by receiving MIDI information from the sequencer program and voice release information from the tone generator task.

Here, the control task constantly manages, for each voice, voice information indicating whether the voice is sounding or vacant. When receiving note-on as MIDI information, the control task refers to the voice information. Specify an appropriate empty voice, set the sounding parameters necessary for the note-on for the voice so that the note-on will be sounded using the voice, and then start sounding in the event flag. Is set, and the voice information of the specified voice is updated. Then, in the sound source task, a new sound is started with the designated voice according to the processing of the control task described above.

When the note-on is received as the MIDI information, the control task refers to the voice information and if there is no empty voice, selects the voice with a low sounding level or the earliest. Necessary for the note-on for the voice so that the note-on is selected by selecting a voice that seems unnecessary according to a predetermined selection criterion, such as selecting a voice that has started sounding, and using the voice to sound the note-on. The process of setting the sound generation parameter and setting the start of sound generation in the event flag after that. Then, in the sound source task, according to the processing of the control task, the current sound generation of the selected voice is muted quickly, and then a new sound generation is started by the selected voice.

As described above, in the control task,
When note-on is received as MIDI information, voice assignment, so-called assignment processing, is performed.

On the other hand, when receiving the note-off as MIDI information, the control task sets the event flag to end the sounding so as to end the sounding of the voice being sounded corresponding to the note-off. Then, in the sound source task, the currently sounding voice corresponding to the note-off is muted according to the processing of the control task described above.

As will be described later, the processing of voices during the tone generator task is not performed for voices that are not sounding. Therefore, the calculation amount of the sound source task increases or decreases according to the number of sounding voices. Therefore, in the present invention, the maximum number of sounds that can be generated can be set in advance according to the capability of the computer and can be changed as appropriate.

When the control task receives a message for changing the values of the steady process parameter and the effect parameter as MIDI information, the control task performs a process of delivering the corresponding steady process parameter and effect parameter to the sound source task.

Furthermore, when the voice release information is passed from the sound source task to the control task, the control task updates the voice information of the specified voice according to the passed voice release information.

(2) Sound Source Task The sound source task is started by a timer every 10 ms (milliseconds), and waveform reading for 10 ms, TVF, TVA,
Perform pan, envelope, LFO, and effect processing collectively in as short a time as possible. As described above, since the sound source task performs processing for 10 ms, if the sampling frequency is set to 32 KHz, processing for 320 samples will be performed.

Each time the sound source task is activated, it searches the contents of the event flag, and when it finds the sounding start flag, it starts sounding according to the sounding parameters. At this time, a memory space for storing sound generation parameters is provided for each voice in two planes. When the sound source task finds a sound generation start flag in the event flag, the sound source task refers to a memory to which the sound source task thereafter refers. The space is switched from the previously referenced surface to another surface, and the previously referenced surface is passed to the control task. Therefore, even when setting the sounding parameter of the same voice number, the control task refers to the sound source task side at the time of sound generation by using another new memory space different from the currently processed surface passed from the sound source task. The sound parameter can be transferred to the sound source task without breaking the parameter area of the currently processed surface.

Here, the processing in each voice will be described. In the tone generator task, the tone generation parameters and the steady process parameters stored in the shared memory are fetched each time the tone generator task is activated, and the envelope and LFO are fetched.
And a calculation for generating a pronunciation service parameter by reflecting the steady processing parameter. The generated sound service parameter controls the waveform read pitch, the TVF cutoff frequency, and the TVA amplitude (amplification degree), and delivers the voice output to the effect according to the effect feed amount of the sound generation parameter. The output of the voice to which no effect is added is stereo output with the pan pot being controlled by the pan of the sound generation parameter (FIG. 3A).

Then, the output of each voice to which the effect controlled by the effect parameter is added and the output of each voice to which the effect is not added are mixed and stereo-outputted (FIG. 3 (b)). Stored in the wave buffer and delivered to the output task.

FIG. 4 is a flow chart showing the process of generating the pronunciation service parameter, and the process is sequentially performed for each voice from the first voice.

First, it is checked whether the event flag indicates the start of sound generation (step S402). If the event flag does not indicate the start of sound generation, it is checked whether the event flag indicates the end of sound generation (step S4).
04).

If the event flag does not indicate the end of sound generation in step S404, it is checked whether voice active, which is a flag indicating that sound is being generated, is set to "1" to indicate that sound is being generated ( Step S406).

On the other hand, when the event flag indicates the end of sounding in step S404, the envelope is released in response to this (step S408), and the calculated value n1 of the sounding service parameter in the previous processing is set to n.
After being set to 0 (step S410), an operation reflecting the envelope, LFO, and steady processing parameter is performed to generate a pronunciation service parameter and set to the calculated value n1 (step S412). That is, the values of the envelope and LFO are changed in accordance with the elapse of 10 ms from the previous processing, and the new envelope, LFO, and steady processing parameters are combined (added or multiplied) to generate a sounding service parameter. Note that the tone generation service parameter is independently calculated for each of the waveform reading pitch, TVF, and TVA.

In step S406, voiceAc
Even when "1" is set in the "live" and the sound is being generated, after setting the calculated value n1 of the sound generation service parameter in the previous process to n0 (step S410),
An operation reflecting the envelope, LFO, and steady processing parameter is performed to generate a pronunciation service parameter and set to the calculated value n1 (step S412). Note that the sounding service parameters include the pitch for reading the waveform, the TV
F and TVA are calculated independently. By the way, in step S402, when the event flag indicates that the sounding is started, the pointers indicating the sounding parameter areas used by the control task and the sound source task are updated to indicate the sounding parameter areas on the other side (step S41).
4), it is checked whether "1" is set in voiceActive, indicating that sound is being generated (step S416).

In step S416, voiceActiv is set.
If e is set to "1" and the voice is being sounded, the currently sounding voice is rapidly muted (step S418), and then the operation in which the envelope, the LFO and the steady processing parameter are reflected is performed. A pronunciation service parameter at the start of pronunciation is generated and set to the calculated value n0 (step S420). Note that the tone generation service parameter is independently calculated for each of the waveform reading pitch, TVF, and TVA.

On the other hand, in step S416, voiceAc
If "1" is not set in the "live" and the sound is not being generated, the process proceeds to step S420 as it is, and an operation reflecting the envelope, the LFO and the steady processing parameter is performed to generate a sounding service parameter at the start of sounding. Set to the calculated value n0.

As described above, in step S420, the operation reflecting the envelope, the LFO, and the steady processing parameter is performed to generate the sounding service parameter at the start of sounding and set it to the calculated value n0, and then voiceA
After setting "1" to cactive to indicate that sound is being generated (step S422), 10 m from the start of sound generation
According to the passage of time of s, an operation reflecting the envelope, the LFO and the steady processing parameter is performed to generate a pronunciation service parameter and set it to the calculated value n1 (step S412). Note that the tone generation service parameter is independently calculated for each of the waveform reading pitch, TVF, and TVA.

Then, in step S412, an operation reflecting the envelope, the LFO, and the steady processing parameter is performed to generate the pronunciation service parameter and set it to the calculated value n1, so that it is used when performing the interpolation processing as necessary. Difference d between the calculated value n1 of the current pronunciation service parameter and the calculated value n0 of the previous pronunciation service parameter
Is calculated (step S424), and it is checked whether the envelope of TVA has reached the end (step S42).
6).

If the envelope of TVA has not reached the end in step S426, the event flag is cleared as it is (step S428), and step S426.
If the envelope of the TVA is at the end,
The voice release information is returned to the control task (step S43
0), "0" is set in voiceActive to indicate that no sound is being generated (step S432), and then the event flag is cleared (step S428).

Also, in step S406, voiceAc
If "1" is not set in the "live" and no sound is being generated, the event flag is cleared as it is (step S428).

When the event flag is cleared (step S428), it is then checked whether all voices have been processed (step S434). Step S
If the processing of all voices is completed by the check in 434, the processing of this flowchart is ended, but if the processing of all voices is not completed by the check in step S434, return to step S402 and repeat the subsequent processing. To process the next voice. That is, the processing of this flowchart is repeated until the processing of all voices is completed.

Note that the waveform reading, TVF, TVA, pan, and feed amount processing to the effect, which is performed in response to the tone generation service parameter, is performed for 10 ms in total each time the sound source task is activated, and thereafter. To
The effect processing is performed collectively for 10 ms.
FIG. 5 shows a flowchart of this process. That is, in the flowchart of FIG. 5, processing is sequentially performed for each voice from the first voice.

First, it is checked whether voiceActive is set to "1" to indicate that sound is being generated (step S502). In step S502, voiceActive is set to "0".
If no sound is being produced, the process jumps to step S514 to check whether all voices have been processed, and proceeds to the next voice. That is, in this flowchart, the processing of the unvoiced voice is not performed.

On the other hand, in step S502, voiceAc
If "1" is set in the "live" and the sound is being generated, the waveform is read so that the sound is generated at the pitch indicated by the sound generation service parameter (step S504). Here, the pronunciation service parameters are linearly interpolated to the required accuracy using the difference d obtained in step S424 of the flowchart shown in FIG. That is, when the sampling frequency is 32 KHz, n0 + d × k / 320 is obtained by using the pronunciation service parameter set to the calculated value n0 calculated in step S410 or step S420 of the flowchart shown in FIG. 4 and the difference d calculated in step S424. Calculate the interpolation value. Note that 320 is the number of samples for 10 ms when the sampling frequency is 32 KHz, and k indicates the number of the sample among the 320 samples, which is' 0 'in the first processing, and' 1 ',' 2 ',...

Next, the output of the waveform read in step S504 is filtered by the TVF (step S5).
06). Here, the cutoff frequency of the TVF is controlled by the sounding service parameter, and the sounding service parameter is set in step S424 of the flowchart shown in FIG.
Using the difference d obtained in the above, linear interpolation is performed to the required accuracy and used.

Further, the amplitude of the filtered output in step S506 is controlled by the TVA in accordance with the tone generation service parameter (step S50).
8). Here, the pronunciation service parameters are used by linearly interpolating to the required accuracy using the difference d obtained in step S424 of the flowchart shown in FIG.

Next, the amount of feed to the pan and the effect is controlled by the sound generation parameter, and as shown in FIG.
The addition and mixing processing with the tone signals of other voices shown in (b) is performed (step S510), and then it is checked whether the processing for 10 ms has been completed (step S512),
If processing for 10 ms has not been completed, step S5
Returning to step 04, the processing from step S504 is performed on the next sample. That is, the processing of steps S504 to S510 is processing for each sample and is 10 ms.
The processing for the number of samples per minute is performed.

On the other hand, when the processing for 10 ms is completed, it is checked whether the processing for all voices is completed (step S514). By this check, when the processing of all voices is completed, the effect processing is collectively performed for 10 ms (step S516), and the wave buffer storing the 10 ms of musical tone signals subjected to the above processing is passed to the output task. (Step S518), the process of this flowchart ends. Here, the wave buffer storing the tone signal subjected to the above-described processing is passed to the output task, and at the same time, a new wave buffer is received, and the new wave buffer is used for the processing of the next steps S502 to S516.

On the other hand, if the processing of the voice has not been completed, the process returns to step S502 and the subsequent processing is repeated to process the next voice. That is, the processing of this flowchart is repeated until the processing of all voices is completed.

(3) Output Task The output task receives the contents stored in the wave buffer received from the sound source task from the audio data output device (the sound card 22 in the computer shown in FIG. 1).
It is. ) To perform a DMA transfer. This DMA transfer is started by an interrupt from the audio data output device.

Here, the audio data output device such as the sound card 22 (hereinafter referred to as “output device”) has a built-in buffer for temporarily storing the output audio data, and the buffer is temporarily provided at each sampling cycle. The stored audio data is output for each sampling. When the amount of audio data remaining in the internal buffer becomes smaller than a predetermined amount, the output task is interrupted, and in response to the interrupt, the output task DMA-transfers audio data of a predetermined sample required by the output device to the output device. Is what you do.

Incidentally, the interruption to the above-mentioned output task usually takes a cycle of about 10 ms.

That is, in the computer shown in FIG. 1, the output task is activated when the interrupt from the sound card 22 and the wave buffer from the sound source task are received.

Then, the output task is 1 from the sound source task.
When a wave buffer for 0 ms is received, an empty wave buffer is returned to the sound source task. There are four or more wave buffers, and one rotation is always performed by an output task for output to an output device, one is used by a sound source task for processing, and the rest are free rotations.

As described above, in the tone processing method according to the present invention, the envelope and LFO calculations are periodically performed in a unit of time longer than the sampling frequency, and when the calculation result is used, the desired accuracy is obtained. The calculation result is interpolated so that

As a result, a complicated condition judgment is required for calculation, and an envelope or LFO having a relatively large number of calculation steps.
It is possible to reduce the calculation of.

Specifically, in the above processing, the calculation reflecting the information given by the envelope, the LFO and the operator is performed only once during the processing of the sound source task. In other words, the sound source task is 1
Given that it is activated once every 0 ms, these operations are only done once every 10 ms. Then, a part using such a calculation result (waveform reading pitch, TVF, TV
In (A etc.), the calculation result is interpolated and used as necessary.

Since the interpolation is performed on the combined value after combining the information given by the envelope, LFO, operator, etc., the envelope, LF
The amount of processing is reduced as compared with the case where interpolation is performed for each information such as O and operators.

Further, in the tone processing method according to the present invention, calculations such as waveform reading, interpolation, envelope, and effect are not processed at sampling frequency intervals, but are collectively processed in a long time span. I am trying to do it.

Specifically, the waveform is read out every 10 ms and processed for 10 ms collectively. At this time, the processing is completed in the shortest possible time so as not to affect other processing.

As described above, by performing a certain process in a lumped time, it becomes possible to reduce the overhead for shifting from one process to another.

Further, by dividing the processing itself into smaller ones, the processing routines are made smaller, and by looping the processing routines, the processing routines easily fit in the cache memory on the CPU, and there is a high possibility that they will be processed at high speed. It will be.

Furthermore, priority is given to each process so that important processes are not interrupted by other processes and are executed without interruption. Specifically, the output task has the highest priority, the sound source task has the next highest priority, and the control task has the lowest priority, so that the output task is not blocked by other processing. ing. Therefore, the musical tone being generated by the output task is not interrupted by other processing.

Further, in the tone processing method according to the present invention, the processing of voices which do not sound is not performed.

As a result, when the number of pronunciations is small, CP
The processing of U12 can be reduced. That is, the processor resources can be released to other programs.

Further, in the tone processing method according to the present invention, the communication between the tone source task and the control task for controlling the tone source task is realized by sharing the memory space.

That is, as shown in FIG. 6, the memory space shared by the sound source task and the control task is provided in two planes (memory space A and memory space B in FIG. 6), and these are used alternately. Has been done.

Specifically, the control task side prepares parameters necessary for sound generation in the sound generation parameter area of the memory space A, and informs the sound source task side by setting a flag that the sound generation parameters have been prepared. When the sound source task side captures the sound generation parameter area in the memory space A, the sound generation parameter area on the other side (memory space B) is transferred to the control task side. The control task side
If a new voice is to be generated with the same voice, parameters are prepared for the newly provided voice parameter area of the memory space B.

As a result, even if the sound source task side is using the sounding parameter while sounding, the control task side can set a new sounding parameter without being concerned about it.

When only one pointer is provided as a method for managing the two-sided memory space, the control task stores a new parameter in one memory space pointed to by this pointer, and the sound source task Refer to the parameter stored in the other memory space. Alternatively, if a pointer pointing to a memory space in which the control task writes parameters and a pointer pointing to a memory space in which the tone generator task refers to parameters are provided separately, the memory spaces pointed to by these pointers are set to be opposite to each other. do it.

Further, in the tone processing method according to the present invention, the number of soundable voices, the presence / absence of effects, or the sampling frequency may be determined depending on the load on the CPU 12. As a result, it is possible to construct a configuration for musical sound generation processing according to the capability of the CPU 12.

That is, at the time of starting the program according to the tone generation processing method of the present invention, or at any timing specified by the user while the program according to the tone generation processing method of the present invention is being executed, an appropriate waveform is generated. By pronouncing a predetermined number of voices and measuring the CPU consumption rate of the sound source task to determine the number of voiceable voices and the presence or absence of effects or the sampling frequency.

An internal timer may be used to measure the CPU consumption rate of the sound source task. This internal timer is a counter that is only incremented periodically (for example, every 1 ms) by an interrupt from the hardware timer.

In the sound source task, at the start and end of the task, that is, before the sounding process and at the end of the sounding process, the value of the internal timer is read and the difference between them is obtained, so that it is consumed in the sounding process by the sound source task. Time can be calculated.
Since the sound source task is called every 10 ms, the CPU consumption rate of the sound source task is determined by how much time the processing of the sound source task consumes.

It is determined in advance how much the CPU consumption rate of the sound source task is allowed, or the user can arbitrarily set it. Depending on the result of comparison between the actually measured CPU consumption rate and the permissible CPU consumption rate, the number of soundable voices and the presence or absence of effects or the sampling frequency are determined.

That is, the actual CPU consumption rate is C
If it is larger than the PU consumption rate, for example, the number of soundable voices is reduced below a predetermined number at the time of measurement, and the number of voices is adjusted so that the actual CPU consumption rate becomes equal to the allowable CPU consumption rate. CPU with actual CPU consumption rate allowed
If it is smaller than the consumption rate, for example, the number of soundable voices is increased more than a predetermined number at the time of measurement, and the number of voices is adjusted so that the actual CPU consumption rate becomes equal to the allowable CPU consumption rate.

By changing the number of soundable voices, presence / absence of effects, and sampling frequency, the CP
Since it is possible to obtain in advance how much the U consumption rate changes, it is possible to determine these sounding modes at once from the actual CPU consumption rate and the allowable CPU consumption rate. It is also possible to repeat the sound generation while changing it so that the actual CPU consumption rate and the allowable CPU consumption rate match.

At this time, the user can select whether the sound quality (sampling frequency or presence / absence of effect) is emphasized or the number of voices is emphasized. In this case, focused items are fixed, items that are not emphasized are changed according to the CPU consumption rate, or items that are emphasized are fixed for the time being and are not emphasized. Change the item within the allowable range according to the CPU consumption rate, and if the actual CPU consumption rate still exceeds the allowable CPU consumption rate, change the focused item as well. it can.

The CPU consumption rate actually measured may be displayed on the display to inform the user.

The process of changing the number of possible voices may be performed as follows. That is, in the control task,
For each voice, the permission information indicating whether the use of the voice is permitted or not is managed. Then, the control task receives the note-on as MIDI information,
When selecting the voice to which the note-on is assigned, refer to the permission information and select from the voices permitted to use.

Therefore, if the use of all voices is permitted, the maximum number of sounds can be produced. To reduce the number of voices that can be pronounced, rewrite the permission information to reduce the number of voices used.

Therefore, since the voices which are not permitted to be used are not pronounced and the voice source task does not process the voices which are not pronounced, the CPU consumption rate is lowered by reducing the number of voices which can be pronounced. You can

As described above, in the control task for performing the assigning process, the permission information indicating whether or not the use of the voice is permitted is managed, and the voice-producible voice is simply processed by rewriting the permission information. You can adjust the number.

The sampling frequency can be changed by adjusting the output cycle of audio data of the output device (sound card in the computer of FIG. 1) and the number of processing samples in one processing of the sound source task. .

Further, the presence / absence of the effect addition can be realized by making it possible to select whether or not to perform the processing of step S516 of FIG. 5 of the sound source processing by switching.

As described above, any timing designated by the user at the time of starting the program according to the tone generation processing method of the present invention or during execution of the program according to the tone generation processing method of the present invention. Then, sound a certain number of voices in an appropriate waveform, and use the CP of the sound source task.
If the number of voiceable voices and the presence / absence of effects or the sampling frequency are determined by measuring the U consumption rate, the processing load amount (voice volume Number, sampling frequency, presence of effects)
Keep moving at. That is, the processing load amount is constant without being changed until it is set again.

However, when a computer is actually used, programs other than the program according to the musical sound generation processing method of the present invention may be simultaneously run in parallel. In this case, since each program is assigned to each time slot and processed by time sharing, another program interrupts during execution of each task of the program according to the musical sound generation processing method of the present invention. May be executed.

For example, the processing time required for one sound source task varies depending on the operating states of other programs. Therefore, as described above, if the processing load of the program according to the musical sound generation processing method of the present invention is constant,
It becomes impossible to follow the variation of the processing load of the CPU due to other programs. Therefore, in some cases, the sound source task may not be processed in time and the sound may be interrupted, or the CP may be interrupted.
Even if the processing capacity of U12 can be afforded, it will not be able to be utilized effectively.

In order to eliminate such a drawback, the CPU consumption rate of the sound source task is constantly monitored, and the number of voice-producible voices is dynamically increased or decreased so as to follow the CPU consumption rate of the sound source task set by the user. You can do it like this.

Specifically, each time the sound source task is activated, the CPU consumption rate is measured from the difference in the values of the internal timers at the start and end of the sound source task, and the voice-producible voice is generated accordingly. Determine the number and presence of effects or sampling frequency.

When dynamically increasing or decreasing the number of voices that can be sounded, the number of voices that can be sounded changes during sound generation. In this case, the following processing is performed. .

That is, in the case of reducing the number of voices, if there is a voice which is not sounding among the voices which are currently permitted to be used, the use permission information of the reduced voices is prohibited. If there are no voices that have not been pronounced among the voices that are currently allowed to be used, or if all voices that have not been pronounced are not allowed to be used and there is still a need to reduce the number of voices, Select the voices you want to reduce from the existing voices, mute this voice rapidly, and disallow further use. To select a voice, a voice with a low sounding level, a voice that started sounding the earliest, or the like can be selected according to the same selection criteria as in the assigning process.

Alternatively, instead of muting the voice that is sounding rapidly, instead of muting the voice that is sounding, wait until the voice is automatically muted by note-off, and disallow the use. May be.

On the other hand, when the number of voices is increased, it is sufficient to permit the use of only the increased number of voices that are not permitted to be used.

[0119]

As described above, since the tone generation processing method according to the present invention is configured to perform a process suitable for a general-purpose computer when a tone is generated in real time on the general-purpose computer, It is possible to solve various problems pointed out in the past, and it is possible to effectively generate a musical sound in real time on a general-purpose computer.

[Brief description of drawings]

FIG. 1 is a block diagram showing a hardware configuration of a typical general-purpose computer in which a program according to the musical sound generation processing method of the present invention can be installed.

FIG. 2 is a conceptual diagram showing the overall structure of a program according to the musical sound generation processing method of the present invention.

FIG. 3 is a conceptual diagram showing processing in a sound source task,
(A) shows processing for one voice, and (b) shows mixing of voices and effects.

FIG. 4 is a flowchart showing a process of generating a pronunciation service parameter in a sound source task.

FIG. 5 is a flowchart showing a waveform reading process, TVF, TVA, pan, and feed amount processing to an effect, which is performed by receiving a sounding service parameter in a sound source task.

FIG. 6 is an explanatory diagram showing a state in which two memory spaces shared by a sound source task and a control task are provided and they are used alternately.

[Explanation of symbols]

 10 bus 12 CPU 14 memory 16 disk controller 18 video card 20 I / O card 22 sound card 24 hard disk 26 display device 28 keyboard 30 mouse

Claims (11)

[Claims] 1. A musical tone that performs a process of collectively generating data of musical tones at a plurality of sampling points included in the predetermined time at a predetermined time longer than a sampling cycle, and temporally ordering and storing the data. A data generation process and a tone data output process for sequentially reading and outputting the tone data at each sampling point stored by the tone data generation process in accordance with the temporal ordering for each sampling cycle. Music tone generation processing method. 2. The musical tone data generation process includes a process of directly handling musical tone data and a process of obtaining a control amount for controlling a processing mode of a process of directly handling the musical tone data. The process for obtaining is the first process for calculating the next control amount after the predetermined time by calculating only once for each predetermined time, and the previous control amount and the next control amount calculated by the first process. A second process of interpolating between the control amount and the control amount at each sampling time point, and the process of directly handling the tone data is performed at each sampling time point obtained by the second process. 2. The musical tone generation processing method according to claim 1, wherein the musical tone data at each sampling time is processed based on the control amount. 3. The first process individually obtains the next value of a plurality of types of control amounts, combines the individually obtained values to obtain the next control amount, and the second process is 3. The musical sound generation processing method according to claim 2, further comprising: interpolating between a previous control amount and a next control amount synthesized by the first process. 4. A process of generating a musical sound in a predetermined mode set in advance in the processing means, a process of detecting a load amount of the processing means when the process of generating the musical sound is performed, and the detecting process. And a process for determining a musical tone generation mode on the basis of the load amount detected by the musical tone generation processing method. 5. The process of determining the musical tone generation mode,
5. The musical tone generation mode is determined based on an allowable load amount of a process of generating the musical sound by the processing unit and a load amount of the processing unit detected by the detecting process. Music tone generation processing method. 6. A process of generating a musical sound in a predetermined mode in a processing means, a process of automatically detecting the load amount of the processing means when the process of generating the musical sound is performed, and the detection. And a process of automatically changing a musical tone generation mode on the basis of a load amount detected by the process. 7. The process for detecting measures the time spent for the process of generating the musical sound, and based on the measured time and the execution time interval of the process of generating the musical sound,
7. The tone generation processing method according to claim 4, wherein the load amount is detected. 8. The musical tone generation mode determined by the musical tone generation mode determination process is at least one of the number of musical tones generated simultaneously, the effect imparting mode of the effect imparting means, and the sampling frequency. 7. The musical sound generation processing method according to claim 4, 5, or 6. 9. The musical tone generation mode determined by the musical tone generation mode determination process is at least two of the number of simultaneous musical tone generation, the effect imparting mode of the effect imparting means, and the sampling frequency. The process of prioritizing any of the musical tone generation modes in accordance with a user's instruction and determining the musical tone generation mode is based on the load amount detected by the process of detecting the non-prioritized musical tone generation mode. 5. The method according to claim 4, wherein
7. The musical sound generation processing method according to any one of 5 and 6. 10. A process of supplying musical tone control information in response to receipt of a musical tone control instruction from the outside, and when new musical tone control information is supplied by the supplying process, it is supplied last by the supplying process. The process of holding the newly supplied musical tone control information while holding the musical tone control information and the process of generating a musical tone based on the musical tone control information at a predetermined time at a timing independent of the process of supplying the musical tone control information. In the case where new musical tone control information is supplied by the supplying process during the period from the previous generating process to the current generating process, it is retained by the retaining process. The musical tone is generated based on the new musical tone control information that is present, and the supply is performed between the previously generated process and the presently generated process with respect to the holding process. If new musical tone control information is not supplied by the processing described above, the musical tone generation processing is performed based on the previously supplied musical tone control information held by the holding processing. Generation processing method. 11. The holding process comprises: a pointer indicating which of two storage areas for holding the previously supplied musical tone control information and the newly supplied musical tone control information is used; And a flag indicating that new musical tone control information is supplied by the processing, and when new musical tone control information is supplied by the supplying processing, a storage area designated based on the content of the pointer In addition, the new musical tone control information supplied by the supplying process is stored, the flag is raised, the content of the flag is referred to at every predetermined time, and the new musical tone control information is supplied to the flag. If it indicates that the flag is turned off, the content of the pointer is rewritten, and the process of generating the musical sound is performed based on the content of the pointer. Claim 1, characterized in that to generate the musical tone based on the musical tone control information stored in the storage area to be
The tone generation processing method described in 0.