JPH10312189A - Musical sound generation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a musical sound generation method capable of decentralizing and reducing a processing load by hourly averaging the processing load for generating the waveform data form the music data. SOLUTION: The music data includes the performance event data and durative information. The processing A of S1 developes the music data to sound source parameters to write them in a P buffer. This S1 is executed when an unused area exists in the P buffer. The processing B of S2 inputs the sound source parameters added with time information of a time frame (F) in a time frame (F-1) whenever reproduction of one time frame (e.g. F-2) is ended to write them in a W buffer. A DMAC(direct memory access controller) controlled by a CODEC driver reads out the waveform data stored in the W buffer one sample each at every sampling period to output a waveform through a sound system.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a musical tone generating method for generating musical tones by receiving music data such as MIDI data. For example, the present invention can be applied to a software sound source that plays back while receiving a MIDI data stream via a network.

[0002]

2. Description of the Related Art There is known a software sound source which realizes a sound source by software without using special hardware.
In some cases, music data prepared in advance is reproduced using such a software sound source, but recently, music data may be loaded from a network and reproduced.

An instruction for requesting music data is sent from a computer to a server through a network such as the Internet, and the server delivers the music data to the computer in response to the command. As a specific example, a case where the computer requests song data embedded as a tag based on the hypertext language (HTML) in a homepage of the World Wide Web (WWW) when loading the homepage, or a file transfer from the computer There is a case where a MIDI file prepared on an FTP server based on a protocol (FTP) is requested by clicking a specific part on the screen. The song data includes a standard MIDI file (SMF) and a multimedia file such as a karaoke file in which image data and lyrics data are combined with the SMF.

In a computer, software such as a web browser is executed to receive music data. Netscape (Netsca)
peCommunications Corporation
ion, trademark) and Internet Explorer (Mi
Microsoft Corporation (trademark) is common. The web browser has a function of loading a music data file from the server by the above-described method.

[0005] As a specific example, "MIDIPLUG"
(Yamaha Corporation, trademark) "Crescendo" (L
live Up Date Corporation (trademark) and "Karaku" (Yamaha Corporation, trademark).
"MIDIPLUG" has the main tone of XG standard,
This is a software sound source that starts playing when MIDI data is completely received. "Crescendo" is the MID
While receiving the I data, the player performs MIDI output to an external MIDI sound source by local hardware or software when the received data accumulates to some extent. These are plug-in software incorporated in the web browser for browsing the World Wide Web (WWW) homepage, and are additional functions to the web browser by other software. It may be a standard feature. "Koraku" is a communication karaoke software package that displays lyrics and performance in synchronization.

FIG. 14 is a diagram for explaining the operation of a conventional tone generation method. In the figure, reference numeral 121 denotes MIDI data input timing, reference numeral 12 denotes a sound source parameter creation period, and reference numeral 14 denotes a waveform data generation period. In the conventional tone generation method using a soft sound source, the sound source driver uses a time frame (F-
At the MIDI data input timing 121 in the period 2), M
IDI data is input, sound source parameters are generated in a sound source parameter creation period 12, and in a waveform data generation period 14 in a period of the next time frame (F-1),
Waveform data including the newly generated sound source parameters is generated. This waveform data is stored in the next time frame (F).
Will be played back. Generation of sound source parameters is performed by MID
Each time I data is input, generation of waveform data is interrupted.

Therefore, when performance events such as note-on, note-off, pitch bend, etc. are concentrated in a specific time frame, for example, time frame (F-2), the load on the CPU is extremely increased in this time frame, and the sound source is generated. There is a problem that the calculation is affected, and the generation of a waveform cannot be completely performed, such as a decrease in the number of simultaneous sounds.

[0008]

SUMMARY OF THE INVENTION The present invention has been made in order to solve the above-mentioned problems, and a substantial load can be obtained by averaging the load of processing for generating waveform data from music data over time. It is an object of the present invention to provide a method for generating a musical tone that reduces the processing load. It is another object of the present invention to provide a musical sound generation method capable of starting the reproduction of a song without waiting for the completion of the loading of all the song data when loading the song data from the network and reproducing the waveform data. It is the purpose.

[0009]

According to the first aspect of the present invention, in the musical sound generating method, the supplied music data is sequentially converted into parameters and written into the first memory.
And generating waveform data based on the parameters stored in the first memory and writing the generated data to the second memory, and allocating an area in which the parameters used for generating the waveform data have been written to the first memory. A second step of releasing the waveform data stored in the memory; and a third step of reproducing the waveform data stored in the second memory to generate a musical tone. The waveform data is generated in accordance with the progress of data reproduction, and the first step is to execute conversion to the parameter when an unused area exists in the first memory.

Therefore, the process of generating waveform data from music data can be divided into a portion that processes according to the progress of the reproduction of the waveform data and a parameter conversion process that can be executed independently of this. Because of this, the parameter conversion process can be executed earlier in the range of the storage capacity of the first memory for the buffer for this process, and the load of the process of reproducing the waveform data from the music data is time-consuming. And the substantial processing load is reduced.

According to a second aspect of the present invention, in the musical sound generating method, a first step of receiving music data supplied through a communication channel and writing the music data in a first memory;
A second step of converting the music data stored in the memory into parameters and writing the converted data into a second memory; generating waveform data based on the parameters stored in the second memory and writing the generated waveform data into a third memory; A third step of releasing, in a second memory, an area in which the parameters used for generating the waveform data have been written, and reproducing the waveform data stored in the third memory to generate a musical tone. A fourth step of generating the waveform data in accordance with the progress of the reproduction of the waveform data in the fourth step; and a second step of storing an unused area in the second memory. Exists, and when there is music data that has not yet been converted into a parameter in the first memory, the conversion to the parameter is executed.

Therefore, the same effect as that of the first aspect of the invention can be obtained, and the parameter conversion process can be performed when there is music data that has not been converted into parameters in the first memory. The reproduction of the song can be started without waiting for the loading of all the song data to be completed.

According to the third aspect of the present invention, in the musical sound generating method, a supplying step of supplying music data;
A parameter conversion step of extracting performance data and performance timing of a performance event from the supplied music data, converting the performance data into tone generator parameters, and writing the performance data together with the performance timing in a memory; Generating a musical tone waveform based on the sound source parameters stored in the memory, wherein the performance timing stored with the sound source parameters in the memory coincides with a time at which the musical tone waveform is to be generated, A tone waveform generation step of executing the tone waveform generation based on the tone generator parameters stored in the tone generator register while updating the content of the tone generator register with the tone generator parameter corresponding to the performance timing; and Musical sound waveform to the above performance timing A musical tone generating step of reproducing the musical tone in response to the tone waveform generating step, wherein the parameter conversion step has a processing priority set lower than that of the musical tone waveform generating step, and is preceded independently of the musical tone waveform generating step. It can be executed.

Therefore, the process of generating waveform data from music data can be divided into a portion that processes according to the progress of the reproduction of the waveform data and a parameter conversion process that can be executed independently of this. As a result, the parameter conversion process can be executed in advance, and the load of the process of reproducing the waveform data from the music data is averaged over time, thereby reducing the substantial processing load.

According to a fourth aspect of the present invention, in the musical tone generating method according to the third aspect, the performance timing of the performance event for which the parameter conversion step has not been executed is determined by the musical tone waveform generating step. When the musical tone waveform is within the range of the predetermined time interval, the musical tone waveform generating step executes the parameter conversion step for the performance event before the musical tone waveform is generated.

Therefore, if there is a performance event in which the parameter conversion step has not been executed yet within a predetermined time interval in which a musical sound waveform is to be generated, the musical sound waveform generation step is executed after the parameter conversion step is executed. can do.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is an explanatory diagram showing a flow of processing of a musical sound generating method according to a first embodiment of the present invention.
FIG. 2 is a timing chart of the musical sound generation method according to the first embodiment of the present invention. In the figure, the same parts as those in FIG. 14 are denoted by the same reference numerals, and description thereof will be omitted. 11 is music data, and 13 is performance timing. In this embodiment, music data of a music file for automatic performance prepared locally in advance is reproduced.

The music data 11 shown in FIG. 2 is stored in the data area. The music data 11 includes, for each performance event, data representing the performance event and data on the performance timing of the performance event.
The performance timing data may be time information indicating an absolute time from the beginning of the song or the like, or duration information indicating a relative time interval from the immediately preceding performance event to the current performance event. As a standard file for exchanging performance data between different sequencers, SMF
(Standard MIDI File) is known. In this SMF, the performance timing is represented using duration information. Hereinafter, a case where the performance timing is represented using the duration information will be described.

In the process A of S1, music data 11 (MIDI) is sequentially read from the data area, developed into sound source parameters (PARM), and written into a P buffer (parameter buffer). However, S1 is executed when there is an unused area (empty) in the P buffer, and the sound source parameter (PARM) is written in the P buffer in advance. This execution is usually performed before the actual performance timing 13 as shown in FIG. In FIG. 2, the performance timing 13 is shown two frames before the time frame in which the waveform data is actually reproduced, in accordance with the operation explanatory diagram of the conventional tone generation method shown in FIG.

As an example of the parameters, in the case of the PCM tone generator, there are a start address and an end address of the waveform memory corresponding to the selected timbre data, an envelope parameter, and the like. The above-mentioned duration information is converted into time information data indicating the performance timing 13 shown in FIG. 2, and a sound source parameter to which the time information is added is written in the P buffer. However, the sound source parameters may be written by adding some time interval information (or the duration information itself included in the music data 11). In any case, sound source parameters are written with time information indicating performance timing, such as time information or time interval information, and are used to know the performance timing when generating a waveform.

The process B of S2 is executed in accordance with the completion of the waveform reproduction of one time frame, and is started at each start time of the time frame as shown in FIG. In this embodiment, every time the reproduction of one time frame (for example, F-2) is completed, waveform data to be reproduced in the next time frame (F) after the time frame to be reproduced next is generated. It has become. If the time information given to the sound source parameter falls within the range of the time frame for generating the waveform data, the sound source parameter must be written to the sound source register during the generation of the waveform data. That is, in the time frame (F-1) immediately before the time frame (F) for reproducing the waveform data, the sound source parameter to which the time information of the time frame (F) is added is input to the sound source register, and Used for generation.

Then, in this time frame (F-1), waveform data (WAVE) is generated using the sound source parameters of the sound source register, and the W buffer (waveform buffer) is generated.
Write to. By executing this step S2, the used sound source parameters do not need to be stored in the P buffer, the area in which the used sound source parameters are stored is released, and the unused area (empty) is stored in the P buffer. Can be.

In S3, the reproduction device performs processing. The playback device includes a CODEC, a CODEC driver, a sound system, and the like. CO here
The DEC is an LSI for a voice interface, and is a semiconductor chip having an A / D converter, a D / A converter, a sampling cycle generator, a waveform compression / expansion circuit, a DMAC (Direct Access Memory Controller) and the like inside. Yes, although it has functions such as A / D conversion and D / A conversion,
It cannot record or play back waveform data. CO
The DEC driver is software that controls or uses the above-described CODEC under other software such as a software sound source to record an input waveform on the RAM and reproduce waveform data on the RAM. . CODE
The C driver may use a DMAC prepared outside the CODEC LSI. In that case, this DM
AC is also included in the playback device.

The program for generating the waveform data transmits the waveform data to the CODEC driver.
It entrusts a waveform reproduction process using EC, that is, a process of transferring waveform data to the D / A converter one sample at a time in each sampling period. The DMAC controlled by the CODEC driver reads out the waveform data stored in the W buffer one sample at a time in each sampling cycle, and outputs the waveform through the sound system.

As an example, the CODEC driver is a DMDEC.
2 buffer memories for one frame as A buffer
We have prepared. While the CODEC driver is reproducing the waveform data stored in one of the buffer memories, the waveform data for one frame stored in the W buffer is written to the other buffer memory. When the reproduction of the waveform data from one buffer memory is completed, the CODEC driver continues to reproduce the other buffer memory, and in the meantime, writes the waveform data stored in the W buffer to one buffer memory. By repeating this, C
ODEC continuously reproduces waveform data. Note that C
The ODEC driver uses 2 buffer memories for one frame.
Instead of preparing two W buffers as described above, the waveform data stored in the two W buffers are alternately reproduced.
While the waveform data on the buffer is being reproduced, the waveform data to be reproduced next may be generated on the other buffer.

Further, in the step S1, a sound generation can be assigned. In the case where sound assignment is performed at this stage, it is necessary for the CPU to recognize how many channels are used and sound is being generated at the time of note-on. At the time of note-off, it is only necessary to find the number of the sounding channel corresponding to the pitch data of note-off, and write the note-off into the tone generator register of this channel number.

In this embodiment, a process B for generating waveform data from music data in accordance with the progress of the reproduction of the waveform data, and a process which can be executed independently of the process B A. Specifically, the cooperation between the processing A and the processing B is realized by adding time information to sound source parameters using time data called duration information included in the music data. Since the processing A and the processing B can be divided, the processing A can be executed earlier in the range of the storage capacity of the P buffer, and the load of the processing of reproducing the waveform data from the music data is temporally averaged. Be transformed into

The storage capacity of the P-buffer described above does not need to be increased more than necessary. The storage capacity of the P buffer is
Since the optimum capacity is determined according to the processing capacity of the CPU and the like, the type of CPU used may be detected to automatically set the storage capacity to be allocated to the P buffer.
Alternatively, the storage capacity allocated by the user may be set arbitrarily.

In the above description, the process from the reading of the music data to the reproduction of the waveform data has been described in accordance with the flow of the processing. However, as the operation actually executed, the reproduction process of the waveform data by the CODEC is “main”. And is prioritized,
In accordance with the reproduction process of the waveform data, the generation of the waveform data of the process B is executed in a “sub” manner.

FIG. 3 is a schematic diagram of a hardware configuration of a personal computer having a tone generating function. 21
Is a CPU bus, 22 is a hard disk, 23 is an FDD,
A removable disk such as a CD-ROM or MO; 24, a display such as a CRT or LCD; 25, a keyboard and mouse; 26, a CODEC; 27, a sound system;
28 is a MIDI interface, 29 is a timer, 30 is a CPU, 31 is a ROM, 32 is a RAM, and 33 is a network interface. A configuration that can be used in common with the first embodiment shown in FIGS. 1 and 2 and a second embodiment described later with reference to FIG. 10 and the like is shown.

A computer in which a CODEC 26 and a sound system 27 are attached to the basic configuration of an ordinary personal computer, and a CODEC driver having a waveform reproducing function is incorporated in an operating system for controlling the basic operation of the computer. Then, a soft sound source can be executed. The program for executing the software sound source is, for example, the hard disk 22
And loaded into the RAM 32 and executed. The music data is stored in advance in the removable disk 23 or supplied from an external server or the like via the network interface 33 and stored in the hard disk 22 in advance.

The P buffer and the W buffer shown in FIG.
It is provided on the RAM 32. The CODEC 26 fetches data from the RAM 32 by using a DMAC (Direct Memory Access Controller) installed inside or separately. A MIDI interface 28 is required when connecting to an external MIDI device, and a network interface 33 is required when connecting to an external server. Although the CPU 30 is a single CPU, a multi-CPU
It may be.

FIG. 4 is a main flowchart of the musical sound generation method according to the first embodiment of the present invention. This flow is started when the MIDI file name of a desired music piece is selected on the display device by clicking or the like, and the software sound source program is started. In S41, the tone generator registers for all tone generation channels are set to a toneless state. In addition, in order to play back a soft sound source, a playback device C
Initialize the ODEC driver, CODEC 26 and sound system 27, and start the waveform playback software. S42
In, the activation factor check is performed. In S43, if there is no activation factor, the process returns to S42, and if there is an activation factor, the process proceeds to S44 to analyze the activation factor.

In S44, when an unused area (free space) equal to or more than a predetermined amount is created in the P buffer, the process proceeds to processing A in S45, in which MIDI data to be processed next is extracted from the music data, and is set in the sound source parameter (PARM). Converted,
Write to P buffer. When the reproduction of the waveform of one time frame is completed, the process proceeds to processing B of S46, where the waveform of the next time frame is generated based on the sound source parameters with performance timing information stored in the P buffer, and W
Write to buffer. When there is another request such as a tone color selection operation or an algorithm selection operation, the process proceeds to S47 to set the sound source operation according to the request. If there is a request for terminating the software sound source, the process proceeds to S48, and the program for the software sound source is terminated. C shown in FIG.
The PU 30 constantly checks these activation factors, determines the activation factors in S44, and starts specific processing.

The most important of these processes is a process B for generating a waveform. In addition, a change in the tone color or the like does not cause much delay in the processing due to the audibility. In consideration of these points, the priority of the processing in the multitask processing is >>>. In the P buffer and the W buffer, the area in which the used data that has been processed and stored in the data stored in each buffer is released as an unused area.

Here, a modified example of the above-mentioned activation factor of the process B will be described. In the above description, the time when the waveform reproduction of one time frame is completed is set as the activation factor.
Instead, there is a CODEC buffer area that is read by the CODEC driver. However, when this area is vacated to some extent and an unused area equal to or more than a predetermined amount is created, the processing B can be activated.

Alternatively, a waveform generation trigger for interrupting at a medium level at a time interval equal to or less than one time frame is frequently issued, and when the interrupt is successfully applied, the processing B is activated and the waveform generation trigger can be generated first. It is possible to recover the delay of waveform generation even if there is a missing waveform. The sound source can be operated stably even in an environment where there are many fluctuations in how the interrupt is applied.

Although the CODEC buffer is usually a buffer for two frames in many cases, the buffer size may be larger. In any of the modifications, in the process B activated by the activation factor, the CODE at that time is
What is necessary is just to generate a predetermined amount of waveform according to the size of the unused area of the C buffer. In the above-described modified example of the activation factor, the W buffer is not necessarily required.

FIG. 5 is a flowchart of the process A shown in FIG. In S51, the performance data and performance timing of the next performance event are extracted from the music data. Specifically, by indicating the performance event that has already been read from the song data with a pointer,
Take out the following: The performance data is MIDI data. The performance timing is duration information included in the MIDI data or time information obtained from the duration information.

In S52, the envelope of each channel (ch) is calculated up to the performance timing, and written into the E buffer (envelope buffer). The E buffer is completely different from the P buffer and the W buffer shown in FIG. The conventional hardware tone generator has a function of reading an envelope, and assigns a sounding channel by checking the envelope level of each sounding channel. If the processing routine is used as it is for a software sound source, an envelope must first be generated. Even when a performance event is not occurring, the level of the envelope is required during sounding.
For example, one envelope level is created for each time frame and stored in the E-buffer, and the envelope level in the time frame is generated by interpolation processing.

In step S53, the performance data is converted into tone generator parameters, and in the case of new tone generation, tone generation is assigned here based on the above-mentioned envelope level. In S54, the sound source parameters are written into the P buffer together with the above-described performance timing.

It is also possible to omit S52 so as not to create an envelope when generating sound source parameters. In this case, at the stage of sound source parameter creation,
A sounding channel is not assigned but is undecided, and a sounding channel is assigned at the time of generating a waveform in processing B described later.
However, in this case, if the performance events overlap, the processing becomes heavy. Therefore, it is desirable to assign the sounding channel at the stage of generating the sound source parameters. Alternatively, although sound channel assignment is performed at the stage of sound source parameter creation, sound assignment is performed without using an envelope level, for example, with late arrival priority, that is, sound input later is preferentially performed. Is also possible.

FIG. 6 is a flowchart of the process B shown in FIG. Process B is started when a waveform is generated. In S61, the waveform generation amount is determined. FIG.
As shown in (1), when an interrupt occurs every time frame, the waveform generation amount is fixed to one frame. In the case of the modification of the activation factor in FIG. 4 described above, the waveform generation amount is determined according to the size of the unused area of the CODEC buffer.

In S62, it is determined whether or not there is an unprocessed performance event in the range for generating the waveform. This determination is made by checking a pointer indicating a performance event that has already been read from the music data and for which the process A has been executed, and determines whether or not the performance timing of the next performance event is later than the range in which the waveform is to be generated. Do with. Normally, the performance events occur at intervals of several ms to several tens of ms for a time frame. Therefore, the performance events within the range in which the waveform is to be generated should have finished the processing A in advance. . However, in the music data, there is a possibility that a performance event that has not finished the processing A remains in the range for generating the waveform.

FIG. 7 is a timing chart of the first example in the case where there is an unprocessed event in the range for generating a waveform. In the figure, the same parts as those in FIGS. 14 and 2 are denoted by the same reference numerals, and description thereof will be omitted. As shown in FIG. 7, the time frame (F)
When it is determined in the time frame (F-1) that the performance event whose waveform needs to be reproduced in the time frame (F-1), the process proceeds to S63 in FIG. This is executed in the sound source parameter creation period 12 of F-1), and the process returns to S62. S
If it is determined in 62 that there is no unprocessed performance event in the range in which the waveform is to be generated, the process proceeds to S64.
S63 may be repeatedly executed.

In S64, when the performance timing of the sound source parameters stored in the P buffer is included in the time frame in which the waveform is to be generated, the generation position is changed during the generation of the waveform data for one frame. At the time when the timing matches, the tone generator register is updated with the tone generator parameters stored in the P buffer, and the waveform data is generated based on the tone generator parameters stored in the tone generator register. For example, in the case of note-on, the sound source parameters required for sound generation are transferred to one channel of the sound source register, and the register indicating note-on / note-off is turned on to make the sound-emitting state. In the case of note-off, The register indicating note-on / note-off is simply rewritten off. At the same time, based on the sound source parameters stored in the sound source register, a waveform corresponding to the generation amount, for example, one frame, is generated and written into the W buffer.

When generating a waveform, the processing A shown in FIG.
The envelope written in the E buffer in S52 is also used. When no envelope is created in S52 of the process A shown in FIG. 5, the waveform is generated while creating the envelope here. When the envelope of the performance event included in the time frame for generating the waveform data has not yet been created in the E buffer even though it is determined in S62 that there is no unprocessed performance event in the range for generating the waveform data. There is. That is, process A is
When the performance event included in the time frame for generating the waveform data has been completed, but the processing A of the next performance event has not been processed, that is, the processed waveform among the envelope waveforms required for generating the waveform data has been processed. The envelope has already been created in process A until the final event of
This is the case when the subsequent part has not been created. In the interpolation processing, an envelope of the performance event included in this time frame cannot be created. In such a case, in step S64, it is necessary to create an envelope for the time frame for generating the waveform data. S
At 65, the waveform data for one frame generated and written in the W buffer is passed to the CODEC driver.

Here, a modified example of S62 will be described. S6
In step 2, process A determines whether at least one performance event after the time frame for which a waveform is to be generated has been processed, and if the process has not been completed, step S63.
The processing proceeds to at least one
The process proceeds to 64. In this way, the envelope is created by the process A up to the performance event 13 after the time frame for which the waveform is to be generated.
It is not necessary to create an envelope in the waveform generation of.

FIG. 8 is a timing chart of the second example in the case where there is an unprocessed event in the range for generating a waveform. In the figure, the same parts as those in FIGS. 14 and 2 are denoted by the same reference numerals, and description thereof will be omitted. As shown in FIG. 8, the time frame (F-
When it is determined in 1) that the unprocessed performance event that has not been converted to the sound source parameter is to generate a waveform in a time frame (F + 2) subsequent to the range of the time frame (F) in which the waveform is generated, In S62 shown in FIG. 6, the process proceeds to S64. On the contrary,
In the modified example of S62, the process proceeds to S63, in which the music data of the unprocessed performance event is converted into sound source parameters, and then the process B is executed. In this manner, the waveform processing can be performed after the processing A of the performance event at the performance timing after the waveform generation range is always executed.

FIG. 9 is a flowchart of the DMAC. 4 shows a flow of processing of a waveform reproduction program by the CODEC shown in FIG. 3. In S71, in response to a sample request interrupt (hardware interrupt) in which CODEC occurs in the sampling cycle, CODEC
DMA buffer (DMAB), which is a waveform buffer of
The waveform data is sent to the CODEC one sample at a time. S
In 72, the value of the transfer sample number p is increased by 1 and the process of S71 is repeatedly executed.

The number p of transfer samples is represented by, for example, 8 bits. CODEC counts the number of transfer samples p,
Each time half the number of samples of the size of the DMA buffer is transferred, ie, the number of transferred samples p = 127 and p
= 255, it is determined that the waveform reproduction of one frame has been completed, and a hardware interrupt is executed. This hardware interrupt is the start trigger of the process B, that is, the activation factor in S44 shown in FIG.

In the waveform generation processing of the processing B, samples for one frame corresponding to half of the DMA buffer are collectively generated on the W buffer, and the waveform data of the W buffer is transferred to the DMA buffer. The size of the above-mentioned DMA buffer is for two time frames, but this size can be arbitrarily changed by changing the address setting of the DMA controller (DMAC).

FIG. 10 is an explanatory diagram showing a processing flow of a musical sound generating method according to the second embodiment of the present invention. This shows the case of stream reproduction. In S81,
The multimedia data (MMD) is received from the network, and is written in the reception buffer on the RAM 32 shown in FIG. The reception buffer area is secured to have a size in which, for example, one piece of multimedia data can be written. In addition, multimedia data is
In addition to the MIDI data, the data includes lyrics data and image data synchronized with the MIDI data.

In S82, music data (MIDI) composed of MIDI data is extracted from the multimedia data (MMD) read from the reception buffer, and written to the M buffer (MIDI buffer). The extracted MID
Like the music data described with reference to FIG. 1, the I data also includes duration information indicating a time interval between performance events. The duration information is converted into time information data indicating performance timing, and music data (MIDI) is stored in the M buffer in addition to the MIDI data of the performance event. However, S82 is
If there is an unused area (empty) in the M buffer and there is unprocessed multimedia data (MMD) in the reception buffer, the processing is executed.

Steps S83 and thereafter are almost the same as S1 to S3 in the flow of the music file reproduction process shown in FIG. 1, but there are some differences. The process A in S83 is executed if there is an unused area in the P buffer and music data (MIDI) exists in the M buffer. Also,
By executing this step, the music data used is M
There is no need to store the data in the buffer, and the area in which the used sound source parameters are stored is released.
Unused area (empty) is created in the buffer. As for S84, as will be described later with reference to FIG. 13, the processing when there is an unprocessed performance event in the waveform generation range is different. Like the tone generator parameters created in S1 shown in FIG. 1, the tone generator parameters created in S83 are tone generator parameters with time information indicating performance timing.

In this embodiment, a process B for generating waveform data from received data in accordance with the progress of the reproduction of the waveform data and a process A which can be executed independently of the process B And song data extraction processing. Specifically, the cooperation between these processes is performed by using time data called duration information included in the music data (MIDI) in the received multimedia data (MMD), music data (MIDI), This is realized by adding time information to the sound source parameter (PARM).

Since the process B can be separated from the song data extraction process and the process A, the song data extraction process and the parameter conversion process can be executed earlier in the range of the storage capacity of the M buffer and the P buffer. The load of the process of reproducing the waveform data from the received data is averaged over time.

Further, when music data exists in the M buffer, a process of converting the data into parameters can be performed, so that the reproduction of the music can be started without waiting for the completion of the loading of all the music data. it can. It is to be noted that the receiving buffer is set to have a relatively large capacity so that, as described above, for example, a size for writing multimedia data for one song is secured, and the received multimedia data is left without being rewritten. In this case, the data remaining on the reception buffer can be reproduced again. Alternatively, using the reception buffer as a write buffer, the received multimedia data is used as is in FIG.
Hard disk 22 and removable disk 23 shown in
May be written. When the M buffer, instead of the reception buffer, is set to have a certain capacity and the music data is left without being rewritten, only the music data can be reproduced again.

FIG. 11 is a main flowchart of the musical sound generating method according to the second embodiment of the present invention. A program that implements this flow is implemented, for example, as plug-in software incorporated in a web browser. Song data like standard MIDI file (SMF)
Alternatively, "tag" data indicating the location where multimedia data (MMD) including such music data is stored in the server is embedded in the homepage, and download is instructed according to the "tag" data. At this time, this program is activated and performs not only downloading but also reproduction of multimedia data.

In step S91, the tone generator registers for all tone generation channels are set to the silent state, and CODE is set.
Initialize C and start the waveform playback software. In S92, the activation factor check is performed. In S93, if there is no activation factor, the process returns to S92. If there is an activation factor, the process proceeds to S94, where the analysis of the activation factor is performed.

In S94, when multimedia data (MMD) is received from the network,
The process proceeds to S95, performs a reception process, and writes the multimedia data (MMD) to the reception buffer. If there is an unused area (empty) in the M buffer and there is multimedia data (MMD) in the reception buffer, S
The process proceeds to step 96, where multimedia data (MMD) is read from the reception buffer, and music data (MIDI) extraction processing is performed. There is an unused area in the P buffer, and
When music data (MIDI) is present in the M buffer,
The process proceeds to process A of step 97, where music data (MIDI) to be processed next is extracted from the M buffer, and sound source parameters (PA
RM) and write it to the P buffer.

When the reproduction of the waveform of one time frame is completed, the process proceeds to the processing B of S98, and the waveform of the next time frame is generated based on the sound source parameters with performance timing information read from the P buffer. When there are other requests such as switching operation such as tone color selection, multimedia processing such as replacement of lyrics and image display and scrolling, the process proceeds to S99 to perform various settings according to these requests. If there is a request for terminating the sound source software, the process proceeds to S100 to terminate the program of the software sound source. CPU 30 shown in FIG.
Constantly checks these activation factors.
In 4, the activation factor is determined and specific processing is started.

At the start of downloading, first, the reception processing and the music data extraction processing are started, and after the music data for a predetermined amount or the predetermined time has accumulated in the M buffer, the processing with the first processing A is started. The waveform generation processing of the software sound source B is started. Alternatively, the above-described waveform generation processing may be started after a predetermined amount or a predetermined amount of multimedia data has accumulated in the reception buffer. Furthermore, the processing A is started before the processing B at the same time as the reception processing and the music data extraction processing, and the processing B is started after the sound source parameters for a predetermined amount or a predetermined time have accumulated in the P buffer. You may do so. In this way, the waveform generation is started after a certain amount of data is accumulated in each buffer, so that the operation immediately after the start of reproduction can be performed stably.
Since the required amount or the predetermined time varies depending on the transfer rate of the network, line quality, and the like, the user can arbitrarily set the predetermined amount or the predetermined time. As for the processing priority, the multimedia task processing is performed in the priority order of >>> multimedia processing >>>>> switching operation >>. In the M buffer, the P buffer, and the W buffer, among the data stored in each buffer, an area in which used data that has been processed and stored is released as an unused area.

FIG. 12 is a flowchart of the music data extraction processing shown in FIG. In S111, the next set of multimedia data (MM
D), and in step S112, the type of the multimedia data (MMD) is determined, and the music data (MID) is determined.
If it is I), the process proceeds to S113, and if it is other multimedia data, the process proceeds to S114. S
At 113, the set of music data is written into an M buffer. At S114, the set of multimedia data is stored in a buffer corresponding to the type, for example, a G buffer for image data, a K buffer for lyrics data, etc. , And the stored data are subjected to display processing in the other processing of S99 shown in FIG. 11 at the reproduction timing of the data.

The lyrics data and image data included in the multimedia data include time information such as a time stamp and a time code indicating the time at which the lyrics and image data are displayed on the display. Is written to the buffer corresponding to
Using this time information and the reproduction timing of the waveform data generated by the software sound source, that is, the time information indicating the performance timing of the performance event, the reproduction of the waveform data and the display of the lyrics data and the image data are accurately synchronized. Can be easily performed. Also, when the performance timing of the performance event is added to the sound source parameter and stored in the parameter buffer with the time interval information being unchanged, if the time interval information is always obtained by adding the time interval information, synchronization is performed in the same manner. Can be done.

FIG. 13 is a flowchart of the process B shown in FIG. In S121, the waveform generation amount is determined, and in S122, it is determined whether or not there is an unprocessed performance event in the waveform generation range. This determination is made by checking a pointer indicating a performance event that has already been read from the music data and for which the process A has been executed, and determines whether or not the performance timing of the next performance event is later than the range in which the waveform is to be generated. Do with. If there is no unprocessed performance event, the process proceeds to S123, and if there is, the process proceeds to S124.

In S123, if the performance timing of the sound source parameter stored in the P buffer is included in the time frame in which the waveform is to be generated, the generation position is changed during the generation of the waveform data for one frame. At the time when the timing matches, the tone generator register is updated with the tone generator parameters stored in the P buffer, and the waveform data is generated based on the tone generator parameters stored in the tone generator register. Then, the generated amount, for example, the waveform of one frame is written to the W buffer. In S125, the generated and written waveform data of one frame is written to the W buffer.
Pass it to the ODEC driver.

Step S124 is a process to be performed when sound source parameters necessary for generating a waveform are not prepared, and it is determined whether or not unprocessed music data exists in a range where a waveform is to be generated.
A pointer indicating a performance event from which the music data has already been read and the music data has been extracted from the received data is checked, and whether or not the performance timing of the performance event next to this pointer is later than the range in which the waveform is to be generated is determined. If there is no unprocessed music data, the process proceeds to S126; otherwise, the process proceeds to S127. In S126, the processing A of the unprocessed performance event is executed, and in S12
The process returns to 2.

Step S127 is a process when music data is not prepared. It is determined whether or not there is unprocessed reception data in the reception buffer by checking that the last reception data after performing the music data extraction process is the latest reception data. It is determined whether there is any received data. If there is unprocessed received data, the process proceeds to S128; otherwise, the process proceeds to S129.
In S128, music data extraction processing is executed, and S1 is executed.
The process returns to 24.

Step S129 is error processing when necessary received data has not arrived. In the error processing, a waveform is generated based only on the current value of the tone generator register, and the process proceeds to S125 so that the sound that has already been sounded is continued. Alternatively, control may be performed so that the sound that is already sounding is gradually attenuated.

As the processing in S122, the same processing as the modification of S62 in the processing B in the case of reproducing the music data shown in FIG. 6 can be performed. In other words, the waveform data generation processing can be performed after the processing A is executed for at least one of the performance events at a time later than the waveform generation range. Similarly, in S124, the parameter conversion and the waveform data generation process are performed after the music data extraction process is performed for at least one of the performance events at the performance timings after the waveform generation range. Can be changed.

In this modified example, the envelope generation is performed in the processing A for the time frame in which the current waveform is generated, as in the case of the waveform generation in S64 in the processing B for the music data reproduction shown in FIG. Since it can be guaranteed that the process has been completed, there is no need to create an envelope in S123. Conversely, if S122 is not changed,
In S123, it may be necessary to create an envelope.

The second embodiment described with reference to FIGS. 10 to 13 is not limited to the case where data is received from a network, and for example, the case where multimedia data is transferred from an external storage medium. The present invention can also be applied to the case where the reproduction is performed while performing the reproduction, or the case where the reproduction is performed while receiving the data by wireless transmission or optical transmission.

In the above description, an error is minimized in the musical sound generation processing.
2. A flow of returning to the process A in the process B is provided as in S63. However, when using in an environment in which an exceptional situation is unlikely to occur, such as when the CPU power is large, it is possible to omit the flow of returning such processing. S122, S124, S126,
S127 and S128 are also flows returning the same processing.
Further, in the reproduction of the music file described with reference to FIGS. 1 to 9, no error processing flow is provided, but in the case of the stream reproduction described with reference to FIGS. 10 to 13, S129 of FIG. The flow can be changed so as to perform error processing such as

In the above description, the case where the duration information is used as the data of the performance timing included in the music data has been described, but time information indicating the absolute time from the beginning of the music may be used instead. . In this case, without converting the duration information into time information data, the sound source parameters to which the time information is added are written in the P buffer. Although the description has been made on the assumption that a general-purpose personal computer is used, the present invention can also be realized by using a computer dedicated to a sound source. Although an example in which the present invention is applied to a soft sound source has been described, the present invention can be applied to a case using hardware for a sound source. The music is
Not only musical sounds but also audible sounds in a broad sense such as voices and onomatopoeic sounds. The present invention can be used not only as a sound source of an electronic musical instrument but also as a tone generation method for entertainment devices such as games and karaoke, and various home electric appliances such as televisions.

[0076]

As is clear from the above description, there is an effect that a tone generating method in which the load is dispersed and the substantial processing load is reduced can be realized. Since the buffer size can be reduced as compared with the case where the waveform data is generated prior to the playback timing of the playback device, the RA used
There is an effect that the capacity of M can be suppressed. In addition, when the music data is loaded from the network and the waveform data is reproduced, the reproduction of the music can be started without waiting for the completion of the loading of all the music data.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing a processing flow of a musical sound generation method according to a first embodiment of the present invention.

FIG. 2 is a timing chart of the musical sound generation method according to the first embodiment of the present invention.

FIG. 3 is a schematic diagram of a hardware configuration of a personal computer having a tone generation function.

FIG. 4 is a main flowchart of a musical sound generation method according to the first embodiment of the present invention.

FIG. 5 is a flowchart of a process A shown in FIG. 4;

FIG. 6 is a flowchart of a process B shown in FIG. 4;

FIG. 7 is a timing chart of a first example when there is an unprocessed event in a range where a waveform is generated.

FIG. 8 is a timing chart of a second example when there is an unprocessed event in a range where a waveform is generated.

FIG. 9 is a flowchart of a DMAC.

FIG. 10 is an explanatory diagram showing a processing flow of a musical sound generation method according to a second embodiment of the present invention.

FIG. 11 is a main flowchart of a musical sound generation method according to the second embodiment of the present invention.

FIG. 12 is a flowchart of music data extraction processing shown in FIG. 11;

FIG. 13 is a flowchart of a process B shown in FIG. 11;

FIG. 14 is an explanatory diagram of an operation of a conventional tone generating method.

[Explanation of symbols]

11 ... music data, 12 ... sound source parameter creation period, 13
... Performance timing, 14 ... Waveform data generation period, 121
… MIDI data input timing

Claims (4)

[Claims] A first step of sequentially converting supplied music data into parameters and writing the parameters to a first memory;
Generating waveform data based on the parameters stored in the memory and writing the generated data to the second memory, and releasing, in the first memory, the area in which the parameters used for generating the waveform data have been written in the first memory And a third step of reproducing the waveform data stored in the second memory to generate a musical tone. The second step includes a step of reproducing the waveform data in the third step. And generating the waveform data in response to the conversion of the parameter when the unused area exists in the first memory. 2. A first step of receiving song data supplied through a communication channel and writing the song data into a first memory, and converting the song data stored in the first memory into parameters and writing the parameters into a second memory. A second step of generating waveform data based on the parameters stored in the second memory and writing the generated waveform data to a third memory, and setting an area in which the parameters used for generating the waveform data have been written to a second memory; A third step of freeing in the memory of
A fourth step of reproducing the waveform data stored in the third memory to generate a musical tone; and the third step includes the step of reproducing the waveform data in accordance with the progress of the reproduction of the waveform data in the fourth step. Generating data, the second step comprising:
Wherein the unused memory area exists in the memory and music data which has not yet been converted to the parameter exists in the first memory, the conversion to the parameter is executed. 3. A supply step of supplying music data, and from the supplied music data, performance data and performance timing of a performance event are extracted, and the performance data is converted into tone generator parameters and written into a memory together with the performance timing. A conversion step and a step of generating a musical tone waveform based on the sound source parameters stored in a sound source register, which is activated at predetermined time intervals, and wherein the performance timing stored in the memory together with the sound source parameters corresponds to the musical tone. At the point in time when the waveform is to be generated, the content of the tone generator register is updated with the tone generator parameter corresponding to the performance timing, and the tone waveform of the musical tone waveform is updated based on the tone generator parameter stored in the tone generator register. A musical tone waveform generating step for performing the generation; A tone generation step of reproducing the generated musical tone waveform in accordance with the performance timing, wherein the parameter conversion step is such that processing priority is set lower than the musical tone waveform generating step, and A tone generation method characterized in that it can be executed prior to the tone waveform generating step. 4. When the performance timing of the performance event for which the parameter conversion step has not been performed is within the range of the predetermined time interval in which the musical tone waveform generating step attempts to generate the musical tone waveform, 4. The musical sound generating method according to claim 3, wherein the parameter converting step is performed on the performance event before the waveform generating step generates the musical sound waveform.