JPS63261398A - Musical sound generator for sampling musical instrument - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to sampling musical instruments, and particularly to musical tone generation techniques in sampling musical instruments.

[Background of the Invention] In conventional sampling instruments, audio signals are converted into A/D
By sampling with a converter etc., it is converted into digital waveform data format.

This is stored in a waveform memory composed of RAM or the like. The musical tone generation circuit (sound source) directly uses the waveform data stored in the waveform memory to reproduce musical tones. That is, the memory of waveform data (PCM data) is an element of the sound source.

In the case of this type of sampling instrument, it is generally required that the input sampling frequency a (sampling frequency for sound) and the output sampling frequency (sampling frequency for reproduction) match in order to eliminate aliasing noise and the like. Efforts to obtain components (frequency components of musical tones) higher than the highest frequency component allowed by the sampling frequency without increasing the sampling frequency (for example, making full use of complex circuits and control techniques such as skipping and interpolation techniques) Efforts have also been made, but nothing satisfactory has yet appeared.In either case, H-F (fidelity) deteriorates significantly. Especially in the high range, the unnaturalness becomes noticeable. Additionally, due to limitations in frequency division technology, the timbre of a certain pitch becomes distorted compared to the timbre of the original pitch (the timbre of the recorded sound), which also degrades sound quality.

In order to alleviate these problems, sampling instruments that can perform multi-point sampling are also known. One extreme of multipoint is to sample and record the pitches of all keys on a keyboard.

In this case, a storage capacity that is a multiple equal to the total number of keys on the keyboard (for example, 88 keys requires a storage capacity that is 88 times as large). This means that the capacity of the waveform memory must be increased by the number of multipoints.

Furthermore, in the case of a sampling instrument in which part of the sound source is waveform data (PCM data), there is a problem in processability. In other words, it is extremely difficult to process the waveform data read from the waveform memory as desired, and even if the processing is incomplete, complex control and circuit configurations are required. When adding a timbre change function, response function, or loop function, satisfactory functions cannot be realized even though a complex configuration is required for the addition.

[Object of the Invention] Therefore, an object of the present invention is to provide a musical tone generation device for a sampling musical instrument that has a small storage capacity and improves the above-mentioned problem of sound quality.

[Summary of the Invention] In order to achieve the above object, the present invention extracts time-variable spectrum data (a set of frequencies and an envelope for each frequency) that characterizes the sound quality of the acoustic signal from the waveform data of the acoustic signal obtained by sampling. The main point is to extract the set) and generate musical tones using a sine wave synthesis method based on the extracted data.

[Principle and Development of the Invention] As described above, the sampling instrument according to the present invention separates the sampling function for recording and the musical tone generation function for musical tone generation, and uses a sine wave synthesis type as the musical tone generation function. Using a musical sound generation means (sound source), in order to match the sampling function for recording and the musical sound generation function, time-variable spectrum data that characterizes the acoustic signal is extracted from the waveform data of the acoustic signal given by the sampling function. The present invention is characterized in that a time-variable spectrum extraction means is provided, and musical tones are generated using a sine wave synthesis method based on the extraction results.

In general, it is easy to make the highest musical tone frequency that can be generated by a sine wave synthesis type sound source sufficiently higher than the highest frequency that can be generated by a sampling instrument.The present invention first takes this point into account.

Therefore, if the sampled sound can be changed into a data structure that is compatible with a sine wave synthesis type sound source, it is possible to prevent the problem of deterioration in sound quality, particularly in the quality of musical tones generated in the high range. This is the second point of interest.

Then, the data structure can be changed to a suitable one by extracting a set of frequencies that characterize the original sound and a set of envelopes (time-variable amplitude information) for each frequency from the waveform data of the original sound using a time-variable spectrum extraction means. achieved.

Therefore, according to the present invention, the capacity of the memory that is the basis for generating sound can be saved. Furthermore, the sound quality is less susceptible to deterioration depending on the sound range, and the sound quality problems of conventional sampling instruments can be effectively solved. Furthermore, data processing can be easily performed on spectrum data rather than sound waveform data itself.

In a preferred configuration, the time-variable spectrum extraction means include means for converting the set of frequencies into a set of frequency ratios relative to a reference frequency (particularly preferably the fundamental frequency of the sampled sound), while The musical sound generating means includes means for generating sine wave signals of each frequency calibrated according to the above-mentioned frequency ratios in response to pitch designation information given from the performance input device of the sampling instrument.

For example, if a frequency component of 440 H and a frequency component of 88087 are obtained by spectrum extraction,
440Hz is converted to a frequency ratio of 1,880) (z = 2. Then, when pitch designation information corresponding to 880H7 is given, the sine wave generating means with frequency ratio l is 88
A sine wave signal of 0H2 is generated, and a sine wave generating means with a frequency ratio of 2 generates a sine wave signal of 1760H1.

Furthermore, in a preferred configuration example, the time-variable spectrum extracting means includes means for performing fast Fourier transform (FFT transformation) on the waveform data of the original sound for each predetermined section, thereby forming an envelope ( The envelope for each frequency is obtained. 1 more
The time-variable spectrum extraction means includes an envelope approximation means for approximating the envelope thus extracted by a limited number of polygonal lines. Monkeys expressed through data. Such a compressed data structure further facilitates processing on each envelope.

[Example] Hereinafter, one embodiment of the present invention will be described with reference to the drawings.

FIG. 1 shows the overall configuration of a sampling instrument according to this embodiment. A signal of a 0M tone (see FIG. 3, for example) is converted into corresponding digital data in an A/D converter l, and is stored in a waveform memory 2. . Therefore, in waveform memory 2,
The original sound is stored in the form of a digital amplitude data sequence (waveform data).

The stored waveform data of the original sound is transferred section by section by the CPU 3 to the FFT unit 4 (DSP, that is, a fast Fourier transformer composed of a digital signal processor chip, etc.).

The FFT unit 4 executes the FFT calculation and sends the calculation result back to the CPU 3. Based on the transferred data, the CPU 3 determines the frequency ratio of each harmonic component and converts the envelope of each frequency into 8 steps. approximates the envelope of Data memory 5 is used in the processing process of CPU 3. It can also be used as a memory for storing each set of timbre data expressed in the form of time-variable spectrum data.

Elements 6, 7.8 are sine wave synthesis type sound sources. Data defining the tone to be generated is transferred from the CPU 3 to the sine wave generation work memory 7 via the sine wave generation base controller 6 and set. .

When performing, upon receiving key press data from the keyboard 12 (an example of a performance input device), the CPU 3 generates a key code (an example of pitch designation information) corresponding to the fundamental frequency and sends it to the sine wave generator controller 6. do. Upon receiving this, the sine wave generator controller 6 instructs the sine wave generator group to generate sound,
Each sine wave generator generates a sine wave signal at a frequency relative to the fundamental frequency based on the assigned frequency ratio, and also sequentially generates an 8-step envelope signal according to the assigned approximate envelope. Envelope control of a sine wave signal. Each enveloped sine wave signal is accumulated in an internal accumulator (not shown) to form a musical tone signal.

The musical tone signal output from the sine wave generator group 8 is converted into a corresponding analog signal by a D/A converter 9, amplified by an amplifier 10, and converted into an acoustic signal by a speaker 11 to emit sound.

Note that 13 is a data input device for changing tone colors and the like used in the edit mode, and 14 is a spectrum envelope display device 2 used for monitoring and the like.

Extraction of time-variable spectral data Next, extraction of time-variable spectral data will be explained in detail.

First, regarding the FFT unit 4, in this example, the FFT unit 4 has an input of 1024 points (the maximum length of data section 1), and calculates a Fourier series of 512 points. Furthermore, in order to remove the effects of aliasing, 112 points in the high range are discarded and amplitude data is output with a resolution of 400 lines (outputting the value of each of the 400 frequency components for each section). If you try to obtain up to 10KHz component, 1 resolution is 10000Hz/
400 and 25H2. Therefore, the sampling frequency for recording, that is, the sampling frequency in A/D converter 1 is:

25Hz x (400+112)X2=2560OH
2, resulting in a sampling period of approximately 394 seconds.

CPU3 converts 25H2 which is the conversion output of FFT unit 4.
Continuously hold every 400 pieces of amplitude data and select the desired component according to an appropriate selection logic. Then, a ratio is added to the higher frequency components.Also, even if the frequency ratio is 1.0 or more, components whose amplitude does not exceed a certain level are treated as non-existent or are considered to be noise and are omitted. When the Fourier transform is completed for all sections of the waveform data, if the power of a certain component is overwhelmingly lower than the power of other components and there is no peak that would affect the sound quality, then Assume that there is no component a,
Eventually, the frequency components are reduced to a number that can be handled by a set of sine generators.

55 to 511 in FIG. 286 show the flow of the extraction and selection processing exemplified above. Further, FIG. 4 illustrates an example of conversion by the FFT unit 4 and subsequent selection by the CPU 3 in a table. In this example, the frequency component of 75H2 satisfies the peak condition of 25 (>20), but since the final power is low at 7, it is removed as not existing. 100 that satisfies all conditions
H2 is the fundamental tone. In Figures 4 and 6, "peak" is the peak in the entire section of the waveform data, "final power" is the average power for the entire section, and "amplitude" is the peak in each section ( is the amplitude in each window).

Note that the above-mentioned selection logic and fundamental tone decision logic are only examples, and other methods can be used.

The envelope (time variable amplitude data) of each selected harmonic component (including the ZSi wood wave component) is
It is approximated by a broken line envelope of steps (Fig. 6, S
12) Examples of this envelope approximation processing technique before and after approximation are shown in FIG.
(filed on January 6) can be used. Basically, if there are 8 steps, there will be 8 broken lines (monotonic function waveform)
The envelope that most closely approximates the original envelope may be determined. Alternatively, the envelope may be displayed on a monitor such as a CRT, and the user may input an envelope that appears to be the closest.

Data such as each tone color stored in the edit data memory 5 can be edited using the data input device 13. Here, the objects of the user's edits are (a) change of timbre, (a) presence or absence of response, or change of response characteristics, and (c) presence or absence of key follow, or change of key follow characteristics.

The processing executed by the equipment side is (A) selection or updating of the sine wave generator (B) changing or updating envelope data (C) changing or updating response data (D) changing or updating key follow data.

The flow of the operation is shown in S13 to S22 of FIG. Since this description is self-explanatory, detailed explanation will be omitted.

FIG. 2 shows one sine wave generator of the sine wave generator group 8 of FIG. 1 as a representative.

Of the “frequency aNω”, N is a frequency ratio, and ω is a fundamental frequency related to key depressions on the keyboard 12. 8-1 is a sine wave generating element, and for the input of Nω, 5 inch (
A sine wave signal of Nωt) is generated. "E" is the frequency ratio N
, "T" represents response data, and "K" represents key follow data. These three inputs E of envelope converter 8-2
, T, and K to generate an envelope waveform signal denoted by C(t). C(t) and 5in(Nωt) are multiplier 8-3
and the result is sent to the accumulator (Σ). “E
The data indicated by ``,'', ``T'', ``K'', and ``N'' are data that is the extract of the musical tone to be generated, which is stored in the sine wave generation work memory 7.

That is, when generating musical tones due to key presses, the sine wave generator controller 6 determines information to be given to each sine wave generator group 8 from information sent from the CPU 3 (pitch data, touch data, etc.). That is, the response data value T, the key follow data value K, the frequency ratio N, and its envelope E are retrieved from the sine wave generation work memory 7 and sent to the related sine wave generator group 8. Each sine wave generator generates sine wave waveform data (with a frequency calibrated by the frequency ratio), key follow, response, and waveform data c (t) that reflects the envelope according to the given information, and combines both. Multiply. The output of each sine wave generator is accumulated in an accumulator to form the final tone signal.

In this way, each sine wave generator is associated with the frequency component of the desired musical tone on a one-to-one basis, so that musical tones with unnatural tones depending on the musical range are not generated.Furthermore, as an internal expression, It uses envelope, key follow, and response functions in a compressed data structure, and it is easy to change these functions, making it easy to change the tone quality of musical tones. In other words, the data can be easily processed.

In the above embodiment, it has been explained that the sine wave generator group 8 is composed of a plurality of sine wave generators, but the meaning of "plurality" includes a plurality of functional units, and is not limited to a hardware meaning.

In addition, in the flow shown in Fig. 6, FFT processing is performed after the input of the original sound is completed, but if the FFT unit 4 is sufficiently fast, the data in the buffer can be processed by FFT processing at appropriate times.
The waveform memory 2 for storing all the waveform data of the original sound becomes unnecessary.

[Effects of the Invention] As is clear from the above explanation, the present invention separates the recording sampling function and the musical sound generation function, and uses a sine wave synthesis type musical sound generation means (sound source) to generate the musical sound. . Then, time-variable spectrum data is extracted from the waveform data obtained by sampling the recording, and based on this extraction result, a musical tone is generated from a sine wave synthesis type musical tone generating means.

Therefore, it is possible to easily obtain a musical tone with a desired timbre regardless of the pitch range, and it is possible to effectively solve the problem of deterioration in sound quality depending on the pitch range, which has been a problem with conventional sampling instruments. Furthermore, as the sound source data, spectral compressed data is used instead of the waveform data of the original sound, which has the advantage that the memory capacity for storing the sound source data can be saved.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram of a sampling instrument according to an embodiment of the present invention, FIG. 2 is a functional configuration block diagram of one sine wave generator of the sine wave generator group 8 in FIG. 1, and FIG. is a waveform diagram showing an example of the original sound, Figure 4 is a diagram used to explain the selection of frequency components after fast Fourier transform, Figure 5 is a diagram used to explain envelope approximation, and Figure 6 is a diagram used to explain the selection of frequency components after fast Fourier transform. 3 is a flowchart showing the operation of this embodiment. 1...A/D converter, 3...CPU,
4... FFT unit, 6... Sine wave generator controller, 7... Work memory for sine wave generation, 8... Sine wave generator group. Patent applicant: Casio Computer Co., Ltd., former Ygurin Figure 2 Figure 3

Claims (3)

[Claims] (1) In a musical tone generating device for a sampling instrument, which includes a sampling means for sampling an acoustic signal and converting it into a digital waveform data format, the sound quality of the acoustic signal is characterized from the waveform data obtained by the sampling means. a time-variable spectrum extracting means for extracting a set of frequencies and a set of envelopes for each frequency; and a sine wave synthesis type musical tone generating means, the musical tone generating means extracting a set of frequencies and an envelope set for each frequency, and the musical tone generating means extracts a set of frequencies and an envelope set for each frequency, A musical tone generating device for a sampling musical instrument, characterized in that a musical tone signal is generated based on the extracted set of frequencies and the set of envelopes. (2) In the musical tone generation device for a sampling instrument according to claim 1, the time-variable spectrum extraction means converts the set of frequencies extracted from the waveform data into a set of frequency ratios relative to a reference frequency. The musical sound generating means includes means for generating a sine wave signal of each frequency calibrated according to the frequency ratio in response to variable pitch designation information provided from the performance input device of the sampling instrument. A musical tone generator for a sampling instrument, characterized by the following. (3) In the musical tone generation device for a sampling musical instrument according to claim 1 or 2, the time-variable spectrum extraction means extracts the waveform data representing the acoustic signal using a fast Fourier method for each predetermined interval. A sampling instrument comprising: means for extracting a set of envelopes in waveform data format by converting; and envelope approximation means for approximating each envelope extracted by the means with a limited number of broken lines. Musical sound generator.