JP2000276149A - Method and device for generating music sound and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method and a device for generating music sound and a recording medium, by which a proper delay time is set in accordance with the state of performance information. SOLUTION: A sounding delay time is set at every part of performance information. Then the sounding delay time is set to be short in the part concerning manual performance (for example, a 'saxophone' in the part 3) and natural performance is made possible by a user. In the meantime, the sounding delay time is set to be long concerning the part with respect to automatic performance (the one except the part 3) and waveform data to be required later is previously generated over a comparatively long range so that the increase of a temporary processing load is treated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a musical sound generating method, a musical sound generating apparatus, and a recording medium suitable for use as a so-called software sound source.

[0002]

2. Description of the Related Art In recent years, a so-called software sound source for synthesizing a musical tone waveform using a general-purpose personal computer or the like has been put to practical use. In the case of a software sound source, it is possible to perform an ensemble by generating a musical sound waveform of a plurality of parts in the same manner as a hardware sound source device. On the other hand, a general-purpose personal computer needs to perform various processes in addition to generating a musical tone waveform, and it is difficult to generate a musical tone waveform completely in real time.
Also, in order to effectively utilize the processing capacity of the CPU, it is often advantageous to generate a musical tone waveform within a certain range at once.

Therefore, in a software tone generator, waveform data to be generated in the future is generated first based on performance information and stored in a buffer, and thereafter, the buffer is read out at each sampling period to reproduce the waveform data. That is common. Then, when the waveform data of a portion that has not been reproduced yet becomes less than a predetermined amount, the subsequent waveform data is generated based on the remaining performance information and written into the buffer. Therefore, a certain delay time is required from when the performance information is supplied to the software sound source to when the sound is actually generated. This delay time is called latency.

The performance information supplied to the software sound source may be performance information stored in a file or the like in advance, and performance information input in real time from an external performance operator or the like. In the former case, a sufficient length of latency may be set in advance. However, in the latter case, if the latency is too long, a time lag between the operation of the performance operator by the user and the actual sound generation becomes noticeable, which hinders the performance. on the other hand,
If the latency is too short, a problem often occurs such that the waveform generation processing cannot be performed in time and the sound is interrupted. As described above, when performance information is input in real time from a performance operator or the like, it is necessary to reduce the latency as much as possible within a range that does not hinder reproduction.

[0005]

By the way, in the case where both the performance information stored in a file or the like in advance and the performance information input in real time from an external performance operator or the like are provided to the software sound source, That is, an ensemble of an automatic performance and a manual performance may be performed. In such a case, since the latency is set uniformly in the conventional software sound source, it has been difficult to sufficiently reduce the latency.

This is because, when the waveform generation processing of the automatic performance and the manual performance is concentrated in a short period of time and exceeds the processing capability of the CPU, a problem such as interruption of sound occurs as described above. The present invention has been made in view of the above circumstances, and has as its object to provide a musical sound generation method, a musical sound generation device, and a recording medium that can set an appropriate delay time according to the form of performance information.

[0007]

According to a first aspect of the present invention, there is provided a tone generating method for outputting a tone signal of each part based on performance information of a plurality of parts. The delay time from when the performance information of a part is provided to when the tone signal of each part is output,
It is characterized in that it differs depending on the part. In the musical tone generating method for outputting a musical tone signal based on performance information, the delay time from when the musical performance information is provided to when the musical tone signal is output may be set to be equal to the performance time. It is characterized in that the information is set according to whether the information is performance information relating to automatic performance or performance information relating to manual performance. According to a third aspect of the present invention, the method according to the first or second aspect is performed. According to a fourth aspect of the present invention, a program for executing the method of the first or second aspect is recorded.

[0008]

BEST MODE FOR CARRYING OUT THE INVENTION Configuration of Embodiment A hardware configuration of a tone synthesis system according to an embodiment of the present invention will be described with reference to FIG. Note that the present embodiment is realized on a general-purpose personal computer. In the figure, reference numeral 21 denotes a CPU, which controls each unit via the CPU bus 20 according to a control program described later. A ROM 22 stores an initial program loader and the like. A RAM 23 is loaded with various programs and data described later, and is accessed by the CPU 21. Reference numeral 24 denotes a timer which generates an interrupt to the CPU 21 at predetermined time intervals.

Reference numeral 25 denotes a MIDI interface,
MIDI signals are exchanged with an external MIDI device (not shown). A hard disk 26 stores an operating system, various drivers, various application programs, performance information, and the like. 27
Denotes a removable disk, which is provided with a CD-ROM, an MO drive, etc., and stores the same information as the hard disk 26.

Reference numeral 28 denotes a display, which is constituted by a CRT, a liquid crystal display, or the like, and displays various information to a user. 29 is a keyboard & mouse,
Various information is input to the CPU 21 by a user operation. Reference numeral 30 denotes a waveform interface for inputting / outputting an analog signal waveform.

Here, details of the waveform interface 30 and the RAM 23 will be described with reference to FIG. In the figure, reference numeral 31 denotes an AD converter, which converts an input analog signal into a digital signal. A sampling clock generator 33 generates a clock signal having a predetermined sampling frequency. A first DMA controller 32 samples the output signal of the AD converter 31 in synchronization with the clock signal, and transfers the sampling result to a designated location in the RAM 23 by direct memory access.

A second DMA controller 34 reads digital waveform data stored in the RAM 23 by direct memory access in synchronization with a sampling clock signal output from the sampling clock generator 33. A DA converter 35 converts the read digital waveform data into an analog signal and outputs the analog signal.

In the RAM 23, a waveform table area 36 stores various waveform data templates. Reference numeral 37 denotes an input buffer area in which a MIDI event is temporarily written by the CPU 21. Also, 38
Is an output buffer area, and the second DMA controller 3
4 is stored. The output buffer area 38 is a ring buffer, the read address of which is determined by a cyclically incremented read pointer. Although not shown,
The RAM 23 is also provided with a recording buffer area for writing the waveform data by the first DMA controller 32.

2. Operation of Embodiment 2.1. Process for Generating MIDI Event In the tone synthesis system of this embodiment, a MIDI event is generated each time a MIDI signal is supplied via the MIDI interface 25. In addition, a MIDI_Out message is generated as time passes based on performance information stored in the hard disk 26 or the removable disk 27 or the like in advance. In this specification,
All of them are handled as MIDI events,
Various processes described later are performed based on the DI event. The latter processing is described in detail in, for example, Japanese Patent Application No. 10-133761 filed by the present applicant.

2.2. Setting of sound delay time for each part Before processing for a MIDI event is performed, a sound delay time (latency) is set in advance for each part. FIG. 4 shows an example of the setting. In the figure, the column of “part timbre” shows a part number and a corresponding timbre name.
Further, the column of “sound source M” indicates a software sound source module used for the part.

Here, TG1 is a PCM sound source, TG2 is a physical model sound source, and TG3 is an FM sound source. Next, in the column of "sound generation delay time", the group of parts (A, B, C) classified according to the delay time, and the delay time (10 msec, 450 msec, 800 msec) in each group.
c) is shown. The user can arbitrarily set which part is to be assigned to which group. However, it is generally preferable to assign a part related to the manual performance to the group A having the shortest sound generation delay time.

In the above example, only the "saxes" of part 3 are classified into group A. This is because the MIDI event of Part 3 is MIDI interface 2
5, and the other parts are recorded on the basis of performance information stored in advance on the hard disk 26 or the removable disk 27 or the like.
It is assumed that an IDI event will occur.

All parts in FIG. 4 can perform automatic performance based on performance information stored in advance on the hard disk 26 or the like. However, for one or more parts, turn off the automatic performance and MIDI
The user can perform a manual performance via the interface 25. The setting of the on / off state of the automatic performance is shown in the rightmost column of FIG. As mentioned above,
Since the "sax" of part 3 is assumed to be played manually, the automatic performance is turned off only for that part. The setting state shown in FIG.
It is stored in a predetermined area in M23.

2.3. MIDI event processing When the above MIDI event occurs, the MID shown in FIG.
The I event processing routine is started. In the figure, when the process proceeds to step SP10, the generated MIDI event is taken into a predetermined area in the RAM 23. Next, when the process proceeds to step SP11, it is determined which of the groups A, B and C the MIDI event belongs to.

If it is determined that the MIDI event belongs to the group A, the process proceeds to step SP12, and a predetermined area (input buffer area) in the input buffer area 37 together with timing information indicating the timing at which the MIDI event occurs.
(A)). Similarly, if it is determined that the input buffer belongs to the group B or C, the process proceeds to step SP12 or SP13, respectively, and the corresponding input buffer
The MIDI event is written in (B) or (C) together with the timing information.

Here, input buffers (A), (B) and
FIG. 6 shows the memory map of FIG. Input buffer (A),
(B) and (C) are used as independent ring buffers. In the figure, the write pointers WP (A), WP
(B) and WP (C) are input buffers (A), (B) and
Indicate the write address in (C), MID
Each time an I-event is written, it is cyclically incremented in the corresponding buffer.

2.4. Generation Trigger Event Processing In this embodiment, a generation trigger event (interrupt) is generated at each predetermined waveform generation period, and the generation trigger event processing routine shown in FIG. 7 is started. In the figure, when the process proceeds to step SP20, the generation range of the waveform data in group A is determined. The generation range is determined so that sufficient waveform data can be provided until the next generation trigger event occurs.

Next, when the process proceeds to step SP21,
MIDI event in this generation range is input buffer
(A). Next, when the process proceeds to step SP22, waveform data in this range is generated based on the MIDI event and the timing information, and written into the output buffer area 38.

Here, a memory map of the output buffer area 38 is shown in FIG. In the figure, the write pointer GWP (A),
GWP (B) and GWP (C) indicate the write addresses of the waveform data generated in the musical tone waveform generation process of groups A, B, and C, respectively, and the write pointer GWP (E) indicates the group A, B, and C.
The write address of the waveform data obtained by applying the effect processing to the B and C waveform data is indexed.

In the figure, an area 101 contains waveform data generated in group C, and an area 102 contains groups B,
The mixed waveform data of the waveform data generated in C is
The area 103 stores the mixed waveform data of the waveform data generated in the groups A, B, and C, respectively. Further, among the waveform data stored in the area 103, the waveform data in the area between the write pointer GWP (E) and the read pointer RP is the waveform data on which the effect processing has been performed.

The write pointers GWP (A), GWP (B) and GWP (C) are originally provided for each part, but in FIG. It is assumed that one part is assigned to each of B and C (three parts in total). GRP (A), GRP
(B) and GRP (C) are each target pointer, and the difference from the read pointer RP is determined according to the sound generation delay time in the groups A, B and C. In addition, the target pointer GRP (A),
GRP (B) and GRP (C) and corresponding write pointer GWP
The difference between (A), GWP (B) and GWP (C) represents the degree of delay in the tone waveform generation processing of each group.

Write pointers GWP (A), GWP (B) and GWP
For the most advanced write pointer of (C), the generated waveform data is written as it is at the position indicated by that pointer, and for the other write pointers, the generated waveform data is written in the future. In this case, the waveform data is written into the waveform data stored at the position where the data is to be added.

The target pointers GRP (A), GRP (B) and GRP (C) and the read pointer RP advance by one address (an amount corresponding to one sample of waveform data) for each sampling cycle. Although the processing will be described later, the traveling direction is indicated by an arrow in FIG.

These write, read and target pointers are cyclically incremented throughout the output buffer area 38. That is, after reaching the final address of the output buffer area 38, each pointer
8 will be returned to the top address. Therefore, the waveform data generated in the previous step SP22 is written in the area beyond the write pointer GWP (A) in the figure (the area below the write pointer GWP (A) in FIG. 3).

That is, the generated waveform data is GWP
The data is newly written in the area ahead of (A) or added to the contents already written. Then, the write pointer GWP (A) is moved to the final position where the waveform data has been written (the write pointer GWP (A) moves downward).

Referring back to FIG. 7, when the process proceeds to step SP23, it is determined whether the waveform data of groups B and C have already been generated in the same range as the generation range of the waveform data previously determined for group A. Is determined.
If “NO” is determined here, the process proceeds to step SP2
Proceeding to 4, the MIDI events of the groups B and C in the range are fetched based on the timing information.

Next, when the process proceeds to step SP25, waveform data relating to groups B and C are respectively generated based on the fetched MIDI event, and the waveform data ahead of the write pointers GWP (B) and GWP (C) are generated. The write pointers GWP (B) and GWP (C) are changed to the final positions where the data is written (or added) to the respective areas. If the waveform data of the groups B and C in the generation range has already been generated, the above-described step SP2
4, SP25 is skipped.

As can be seen from FIG. 3, since the write pointers GWP (B) and GWP (C) normally advance considerably beyond the write pointer GWP (A), the determination in step SP23 is almost always "YES". "become. The processing for generating the musical sound waveforms of groups B and / or C is delayed due to an increase in the load due to processing other than the software sound source.
Only when the positions of (B) and GWP (C) are equal to the position of the write pointer GWP (A), "NO" is determined in step SP23.

Next, when the process proceeds to step SP26,
An effect process is performed on the mixed waveform data of the waveform data generated in groups A, B, and C. Specifically, the waveform data in the area between the write pointers GWP (A) and GWP (E) shown in FIG. 3 is read.

Then, effect processing is performed on the read waveform data of each group, and the resulting waveform data is written again in the same area. Next, when the process proceeds to step SP27, the position of the write pointer GWP (E) is changed to the same position as the write pointer GWP (A).

Next, when the process proceeds to step SP28,
It is determined whether the time allocated to the generation trigger event processing routine (FIG. 7) has a margin. If there is enough time, the process proceeds to step SP29, and waveform data relating to groups B and C are generated according to the time allowance. Specifically, in step SP29, the subroutine in FIG. 9 is called.

In FIG. 9, when the process proceeds to step SP50, the generation ranges of the groups B and C are determined according to the time margin. Here, to eliminate the delay in the waveform generation processing of group B or C, respectively,
The generation range is determined within the range allowed by the time margin. That is, when the delay of the group is large, a larger generation range is designated.

Next, when the process proceeds to step SP51, a MIDI event in the generation range of the group B is fetched from the input buffer (B) of the input buffer area 37 based on the timing information. Next, the process proceeds to step SP5.
In step 2, waveform data of group B in the generation range is generated based on the received MIDI event. The generated waveform data is stored in the output buffer area 38.
Is written (newly written or added) in the area ahead of the write pointer GWP (B), and the write pointer GWP (B) is updated to the last written position.

Next, when the process proceeds to step SP53, the MIDI event in the generation range of the group C is fetched from the input buffer (C) of the input buffer area 37 based on the timing information. Next, the process proceeds to step SP5.
In step 4, waveform data of group C in the generation range is generated based on the received MIDI event. The generated waveform data is stored in the output buffer area 38.
Is written (newly written or added) in the area ahead of the write pointer GWP (C), and the write pointer GWP (B) is updated to the final position where the writing was performed.

2.5. DMA Processing When the sampling clock generator 33 generates a clock at a sampling cycle, a DMA interrupt is generated for each clock, and the DMA processing routine shown in FIG. 8 is called. In the figure, when the process proceeds to step SP40, the read pointer RP is incremented by "1". However, if the read pointer RP has already reached the end address of the output buffer area 38, the read pointer RP
Is set to the head address of the output buffer area 38.

That is, when the head address of the output buffer area 38 is set to “0”, the read pointer RP becomes
It is set to “the remainder obtained by dividing the result of adding“ 1 ”to the current RP by the size (SIZE) of the output buffer area 38”. Next, when the process proceeds to step SP41, the word indicated by the updated read pointer RP is read from the output buffer area 38 and supplied to the DA converter 35 via the second DMA controller 34, and the sound system (not shown) Pronounced via

3. Effect of Embodiment In the above generation trigger event processing, if there is enough time for the generation trigger event processing, step SP2
At 9, a relatively long range of group B and C waveform data is generated. Therefore, when the waveform data of the group A is generated, the waveform data of the groups B and C in the same range have already been generated in most cases.
In step SP23, "YES" is determined.

In other words, the waveform data for groups B and C are generated in advance if there is enough time, so that the generation range in which a large amount of processing is required to generate the waveform data for group A is temporarily limited. , The waveform data of group B or C in the same range is more likely to have been created in advance, and it is possible to prevent such a problem that the musical sound is interrupted.

4. Modifications The present invention is not limited to the embodiments described above,
For example, various modifications are possible as follows. (1) In the above embodiment, the description has been made on the assumption that all the programs of the tone synthesis system are installed in a personal computer. However, these programs are stored in a recording medium such as a CD-ROM or a floppy disk and distributed. Or may be distributed to each computer via wireless or network.

(2) In the above embodiment, the user determines the part to be assigned to the group (group A) with the shortest sound generation delay time. However, as shown in FIG.
When any one of a plurality of parts capable of performing an automatic performance is turned off, there is a high possibility that a manual performance will be performed on that part. Therefore, a part for which automatic performance is turned off may be automatically assigned to group A.

[0046]

As explained above, according to the present invention,
Since the delay time from when the performance information is provided to when the musical tone signal is output is set according to the part of the performance information, or whether it is an automatic performance or a manual performance, the form of the performance information is An appropriate delay time can be set accordingly.

[Brief description of the drawings]

FIG. 1 is a block diagram of a musical sound synthesis system according to an embodiment of the present invention.

FIG. 2 shows a RAM 23 and a waveform interface 30.
It is a detailed block diagram of.

FIG. 3 is a memory map of an output buffer area 38;

FIG. 4 is a diagram showing a setting state of a delay time and the like of each part.

FIG. 5 is a flowchart of a MIDI event processing routine.

FIG. 6 is a memory map of an input buffer area 37;

FIG. 7 is a flowchart of a generation trigger event processing routine.

FIG. 8 is a flowchart of a DMA processing routine.

FIG. 9 is a flowchart of a subroutine of a generation trigger event processing routine.

[Explanation of symbols]

20 CPU bus 21 CPU 22 RO
M, 23 RAM, 24 Timer 25 MID
I interface, 26 ... Hard disk, 27 ...
... Removable disk, 28 ... Display, 29 ... Keyboard & mouse, 30 ... Waveform interface, 3
1 AD converter, 32 First DMA controller, 33 Sampling clock generator, 34 Second DMA controller, 35 DA converter, 36
...... Waveform table area 37 Input buffer area 3
8 ... output buffer area.

Claims (4)

[Claims] 1. A tone generating method for outputting a tone signal of each part based on performance information of a plurality of parts, wherein a delay from when the performance information of each part is provided to when a tone signal of each part is output. A tone generation method, wherein the time is made different depending on the part. 2. A tone generating method for outputting a tone signal based on performance information, wherein a delay time from when the performance information is provided to when the tone signal is output is determined, wherein the performance information relates to automatic performance. Or the performance information relating to manual performance. 3. A musical sound generating apparatus for executing the method according to claim 1. 4. A recording medium on which a program for executing the method according to claim 1 or 2 is recorded.