JP3256316B2 - Digital electronic musical instrument system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital electronic musical instrument system, and more particularly, to digitizing and storing a part of performance sounds such as musical instrument sounds recorded by a microphone and the like, and storing the stored actual musical instrument sounds. And the like are sequentially converted in accordance with pitch information (melody and the like) separately input from the outside, and a digital electronic musical instrument system as an acoustic signal reproducing system for reproducing and outputting the same.

[0002]

2. Description of the Related Art Conventionally, in a sampling type synthesizer, a recorded PCM waveform is reproduced not only at the same pitch as during recording but also at a different pitch. For output at this different pitch, PC
A technique of reproducing the M waveform at different sampling frequencies and changing the pitch is employed.

In a digital electronic musical instrument of this kind, when the above-mentioned reproduction pitch is changed, that is, when the sound stored in the apparatus is converted into a specified pitch, a method based on a direct modulation of a clock frequency and a sinc function are generally convoluted. method,
There are three known methods, an interpolation method using an algebraic curve passing through a sample point.

FIG. 7 shows an example of the "method based on direct modulation of clock frequency". In the conventional example shown in FIG. 7, first, when pitch instruction information such as vibrato is given, a clock signal having a frequency corresponding to the level is output from the variable frequency clock generator 51. The frequency of this clock signal is, for example, doubled per “pitch + 1 octave”.

The clock signal is supplied to an address counter 5
2 to increase the output address of the address counter. In this case, the higher the frequency of the clock signal, the sooner the counting is performed, and the lower the frequency of the clock signal, the slower the counting. PCM (Pulse Code Mod) corresponding to the output address of the address counter 52
)) is read from the PCM waveform memory 53, converted into an analog sound signal by the D / A converter 54, and output to the outside via the low-pass filter 55.

FIG. 8 shows a conventional example of the aforementioned "
An example of “a method of convolving a sinc function” is shown. This FIG.
Is a conceptual example, and for the sake of explanation, the sampling frequency is assumed to be F _S [Hz]. In FIG. 8, D / A converter 61, the input PCM waveform p · _{F S}
Digital / analog conversion (D / A conversion) is performed at [Hz] to create an analog waveform. Then, _"F C = min _(F output of the D / A converter 61 the cut-off frequency as _{F C}
Enter the _{S / 2, p · F S} / 2) low pass filter 62 of the "cuts the high frequency component. Subsequently, the A / D converter 6
Analog / digital conversion (A / D) at F _S [Hz] at 3
Conversion) and outputs a PCM waveform.

Here, to obtain an analog waveform by applying a low-pass filter by performing D / A conversion on the PCM waveform described above is theoretically equivalent to convolving a sinc function with the PCM waveform. A / D conversion of the resulting analog waveform again is equivalent to extracting the value of the analog waveform at a predetermined sampling cycle. Therefore, from this viewpoint, if only the value for each sampling period (constant time interval) becomes clear, the value of the time during that period becomes unnecessary. That is, the PCM sampling value y (n) after the A / D conversion can be calculated from the original PCM sampling value x (m) before the D / A conversion by the following equation.

[0008]

(Equation 4)

For this reason, in the conventional example shown in FIG. 8, an analog filter and an A / D converter, in addition to a D / A converter, are actually unnecessary, and can be processed in a time-division manner. There is an advantage that the pitch can be converted only by changing the value of.

FIG. 9 shows a conventional example of the aforementioned "
An example of "an interpolation method using an algebraic curve passing through sample points" is shown. In the conventional example shown in FIG. 9, a quadratic curve passing through three adjacent PCM sample values is determined, and the curve is sampled again. The method shown in FIG. 9 is effective for cost reduction because the processing is simple.

[0011]

SUMMARY OF THE INVENTION
The embodiments have the following inherent disadvantages.
Exist. First, the conventional example shown in FIG.
In the case of "direct modulation method", only a single sound is generated
Simplifies and is convenient, but multiple
If it is necessary to make two sounds at the same time,
A corresponding number of D / A converters must be provided. So
The reason is that since the clock frequency is different for each sound, one D /
This is because time division cannot be performed by the A converter. For this reason,the above
Synthesizer that must output the sound of
The electronic piano has a disadvantage that the cost is high.

Further, in the case of the conventional example of "convolution of sinc function" shown in FIG. 8, the calculation of the sinc function is complicated, and the range of m of Σ is (negative infinity, positive infinity). There are two problems. The sinc function can be solved by making a table in advance,
Since the convergence of the sinc function is slow in the range of m of Σ, the calculation error of the PCM sampling value y (n) increases when the range of m is cut off finitely, and this results in aliasing noise in the output signal. Because of its appearance, it has been unsuitable for improving sound quality even though it can contribute to reducing product costs.

Further, in the case of the conventional example of "interpolation method using an algebraic curve passing through sample points" in FIG. 9, "sharpness" occurs at a PCM sample in a virtual analog signal formed by this method. Since the sharp signal has a discontinuous first derivative, its Fourier transform attenuates only to the order of “1 / ω ² ” as the ω changes to infinity with respect to the frequency ω. For this reason, aliasing noise increases, and as in the case of FIG. 8 described above, there is an inconvenience that high-quality sound output cannot be expected even though the simplicity of processing can contribute to a reduction in product cost.

In these series of various methods, if the aliasing noise at the time of resampling is large, an unpleasant ghee or a humming sound such as a wah-waan is generated separately from the original signal component. As described above, there is an inconvenience that the sound is remarkably generated in the case of a musical instrument sound having a high harmonic component.

[0015]

The object of the present invention is to improve the disadvantages of the prior art, and in particular to enable time-division processing in signal processing and reduce the occurrence of aliasing noise, thereby enabling one of the previously recorded ones. It is an object of the present invention to provide a digital electronic musical instrument capable of outputting reproduced sound outputs of different pitches with high quality using musical sounds and the like.

[0016]

According to the present invention, a part of a performance sound such as a musical instrument sound recorded by a microphone or the like is resampled after a predetermined pitch difference is given to digitized acoustic information. and pitch change means for, and a pitch control unit for controlling based on the operation of the pitch conversion means to an external command. The pitch conversion means discretely converts discrete filter coefficients of FIR and IIR.
To get a discrete time sampled value output by convolving with the waveform
N-th ward with a finite base and equivalent to equation (2) above
Discrete wave with impulse response as continuous polynomial in continuous time
It comprises an anti-aliasing filter that obtains a continuous-time output in the form of a piecewise polynomial by convolving with the form . or
It is combined frequency characteristic correction unit for setting a high-frequency component of the digitized audio information at a high level, and employs a configuration equal. This aims to achieve the above-mentioned object.

[0017]

The operation will be described with a specific example. An acoustic signal such as a musical instrument sound recorded by a microphone or the like is first digitized by an A / D converter. Specifically, the sampling frequency F of the input signal is twice the highest frequency in the spectrum distribution of the signal.
_S sampling is performed. The output of this A / D converter is sent to the signal processing means as an original signal. Signal processing means, immediately oversampling unit functions, higher sampling frequency "L · F _S from the sampling frequency of the input
(L> 1) ", re-sampling of the digitized sound signal, that is, oversampling is performed.

[0018] In this signal processing means, subsequently, the frequency characteristic correction unit functions, filtering with an inverse characteristic of the anti-aliasing filter that the piecewise polynomial and the impulse response is performed. Then, the output signal of the frequency characteristic correction unit is stored in the information storage unit.

On the other hand, when an oversampling unit is provided in addition to the frequency characteristic correction unit, the oversampling unit immediately functions at the signal input stage or the signal output stage of the frequency characteristic correction unit, and is higher than the input sampling frequency. At the sampling frequency “L · F _S (L> 1)”, re-sampling of the digitized audio signal, that is, oversampling, is performed.

The PCM waveform stored in the information storage means is
Under the control of the pitch control means, the signal is appropriately read out by the above-described piecewise polynomial impulse response anti-aliasing filter. In this anti-aliasing filter, a predetermined filtering polynomial coefficient is calculated under the control of the pitch control means, and the filtering polynomial determined based on the coefficient is resampled. Then, the resampled PCM waveform is reproduced and output as a resampling signal having little noise in accordance with the pitch specified by the above-mentioned pitch control means.

[0021]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to FIGS. First, in the embodiment shown in FIG. 1, an A / A for digitizing an acoustic signal such as a musical instrument sound recorded by a microphone or the like is used.
It comprises a D converter 1 and information storage means 3 for storing the digitized sound signal. The embodiment shown in FIG. 1 converts the acoustic signal stored in the information storage means 3 into a continuous-time signal (section polynomial) having a predetermined pitch, and re-samples the signal. Anti-aliasing of impulse response
A filter 4 is provided.

Further, the embodiment shown in FIG. 1 comprises a pitch control means 5 for controlling the operation of an anti-aliasing filter 4 for a piecewise polynomial impulse response based on an external command, and an anti-aliasing for a piecewise polynomial impulse response. A D / A converter 7 for converting the output of the filter 4 into an analog signal and outputting the analog signal; The pitch control means 5
It has a function of controlling the address information in the information storage means 3 as well. Symbol n indicates a control signal for address setting.
Reference numeral 6 denotes input means such as a keyboard for sending predetermined pitch information to the pitch control means 5 described above. This input means 6 may be replaced by an external computer or the like operated by a predetermined program, for example, as necessary.

Between the A / D converter 1 and the information storage means 3, a signal processing means 2 having an oversampling section 2A is provided. Further, the signal processing means 2 is provided with a frequency characteristic correction unit 2B for setting a high frequency component of the signal after oversampling to a high level.

[0024] In the oversampling section 2A of the signal processing unit 2, resampling of the digitized acoustic signal with a high sampling frequency than the sampling frequency of the input _{"L · F S (L> 1} ) ", i.e. over Sampling is performed.

In the present embodiment, the anti-aliasing filter 4 of the piecewise polynomial impulse response as the pitch conversion means has an impulse response h _N (t) of the Nth order polynomial and is expressed by the following equation (2). . Equation (2) shows a case where the sampling frequency is 1. It may be replaced with "t · F _S" and "t" in the following equation if the sampling frequency is F _S (2).

[0026]

(Equation 5)

Here, the function of the oversampling section 2A of the signal processing means 2 will be described in more detail. First,
It is assumed that a PCM waveform that is the output of the A / D converter 1 is input to the anti-aliasing filter 4 having the above-described impulse response. Here, as the anti-aliasing filter 4, it is assumed that an impulse response h _N (t) composed of a cubic piecewise polynomial (an anti-aliasing filter with an impulse response h ₃ (t)) is used. .

FIG. 2 shows the spectrum of the filtered polynomial signal. In FIG. 2, the symbol Ω indicates a normalized angular frequency. The symbol π corresponds to the Nyquist frequency. Here, a component in the range of 0 ≦ Ω <π (cross-hatched portion) indicates a desirable signal. Also, π <
The component (hatched portion) in the range of Ω is a portion that is folded at the time of resampling and becomes noise (aliasing noise).

On the other hand, FIG. 3 shows the spectrum of the filtering polynomial signal output from the anti-aliasing filter 4 when the PCM waveform obtained when the signal processing means 2 oversamples twice is input. As can be clearly seen from FIG. 2, the range in which the aliasing noise component exists is narrowed, and the signal-to-noise ratio (S / N ratio) is significantly improved. In this case, how many times the oversampling is adopted is determined by the specifications of the spectrum shape of the input signal and the required aliasing S / N ratio.

Next, the function of the frequency characteristic correction section 2B of the signal processing means 2 will be described. The frequency characteristic correction unit 2B is provided corresponding to the anti-aliasing filter 4 described above. In other words, the anti-
Aliasing filter 4, the piecewise polynomial of the impulse response "h _{N (t)"} is represented by N times convolution to其its following formula (3) are shown as h _{0 (t).}

[0031]

(Equation 6)

Accordingly, the frequency characteristic “H _N (Ω)” of the anti-aliasing filter 4 is given by the following equation (4).

[0033]

(Equation 7)

In this embodiment, a frequency characteristic correction unit 2B is provided to flatten the frequency characteristic H _N (Ω). That is, the frequency characteristic correction unit 2B performs the anti-
For the input signal to the aliasing filter 4,
It functions so that a digital filter of “H _N ⁻¹ (Ω); 0 ≦ Ω <π” can be applied in advance. As a result, the corrected frequency characteristic of “H _N (Ω)” is expressed by the following equation (5), and the frequency characteristic of the pass band is clearly flat.

[0035]

(Equation 8)

Such a correction is effective because the anti-aliasing filter 4 has a non-flat frequency characteristic between DC and Nyquist frequency. If such correction is not made, the timbre deteriorates (specifically, the timbre becomes darker). Therefore, the frequency characteristic correcting portion 2B of the present embodiment has a that plays an important mission.

The correction of the input signal to the anti-aliasing filter 4 by the frequency characteristic correction unit 2B may be performed in real time, or may be temporarily stored after the correction. However, in the present embodiment, the PCM waveform that is the input signal is oversampled in advance by the oversampling unit 2A and a digital filter of “H _N ⁻¹ (Ω)” is applied in the frequency characteristic correction unit 2B, and the result is obtained. The information storage means 3 can record and hold the information. For this reason, oversampling and correction of frequency characteristics are not required at the time of reproduction, so that the amount of calculation required at the time of pitch conversion is significantly reduced, and more effective and rapid processing of frequency conversion is possible.

Next, the relationship between the anti-aliasing filter 4 of the piecewise polynomial impulse response as the above-mentioned pitch conversion means and the pitch control means 5 will be specifically described.

It is assumed that a sampling waveform x (t) of the following equation (6) is input from the information storage means 3 to the anti-aliasing filter 4 having the impulse response h _N (t) as a pitch control signal. I do. Here, this sampling waveform x (t) is intended for a waveform that has been oversampled and whose frequency characteristics have been corrected (recorded and held in the information storage means 3).

[0040]

(Equation 9) Here, continuous time is represented by the δ function, but x
(M) is a discrete value of {x0, x1, x2,...}.

The filter output y (t) in this case is given by the following equation (7).

[0042]

(Equation 10) Hereinafter, the derivation process of Expression (7) will be described in detail.

[Equation 11]

Then, y (n + Δ) representing the value of the filter output y (t) at time t (= n + Δ) is given by the following equation (8) (where n is an integer part of t and Δ is fractional part of t: 0 ≦ Δ <1). That is, n is an integer of t
N∈Z, Δ indicates the fractional part of t, and 0 ≦ Δ <1
Δ∈R.

[0044]

(Equation 12)

[0045] _{Here, k = m-n h N} (Δ-k) is "- (N + 1) / 2 ≦ (Δ-k) <
+ (N + 1) / 2 ". _kmin = ceil (Δ- (N + 1) / 2); _kmax
= Floor (Δ + (N + 1) / 2)

For k _min, a round-up numerical value is used.
_{For max,} a depreciation value is used. In equation (8), y
(N + Δ) is called a filtering polynomial in this embodiment.

In this equation (8), the time "t" is set by the output signal from the pitch control means 5 described above. That is, the n component at the time "t = n + .DELTA." Is sent to the information storage means 3 as a control signal for setting the address, and the .DELTA. Component is an anti-aliasing filter as the t coordinate signal of the filtering polynomial impulse response. It is sent to 4. The control signal is output from the pitch control means 5 and functions as described above, whereby the resampling signal of the filtering polynomial specified by the equation (5) is sequentially output from the anti-aliasing filter 4. .

The D / A converter 7 functions to convert the re-sampled signal output from the anti-aliasing filter 4 into an analog signal and to output it as a frequency-converted reproduced sound to the outside.

Next, the overall operation of the above embodiment will be described with reference to FIG.

An acoustic signal such as a musical instrument sound recorded by a microphone or the like is first digitized by the A / D converter 1. Here, the input signal is sampled at a sampling frequency F _S that is twice the highest frequency in the spectrum distribution of the signal (refer to the symbol A in FIG. 4). This A / D converter 1
Is sent to the signal processing means 2 as an original signal. In the signal processing means 2, the oversampling unit 2A
Code of but resampling function, and the digitized audio signal at a high sampling frequency than the sampling frequency of the input _{"L · F S (L> 1} ) ", i.e. over-sampling is performed (in FIG. 4 B).

In the signal processing means 2, the frequency characteristic correction section 2B subsequently functions to correct the high-frequency component to a high level, that is, the inverse characteristic of the anti-aliasing filter 4 using a piecewise polynomial as an impulse response. Is applied. That is, digital filtering of “H _N ⁻¹ (Ω); 0 ≦ Ω <π” is performed corresponding to the output of the anti-aliasing filter 4 having the frequency characteristic H _N (Ω) (see FIG. 4). C). The oversampled and digitally filtered PCM waveform is stored in the information storage unit 3.

The PCM stored in the information storage means 3
The waveform is controlled by the pitch control means 5 and read out to the anti-aliasing filter 4 having a piecewise polynomial impulse response as appropriate. In the anti-aliasing filter 4 is controlled by the pitch control unit 5 predetermined filtering polynomial are calculated (see FIG. 4 numeral D), calculated filtering polynomials of its is resampled. Then, the resampled PCM waveform is reproduced and output as a resampling signal having little noise according to the pitch specified by the pitch control means 5 described above (reference numeral E in FIG. 4).
reference). That is, the filtering polynomial is calculated
(See D in FIG. 4).
The term is resampled. FIG.
The white circle where the Taling polynomial curve is the discrete waveform sample value
The present invention uses the conventional algebraic curve interpolation method
Has also been shown to produce smooth output that cannot be obtained with
I have.

Here, the calculation of the aliasing noise (check of the S / N ratio) at the time of the impulse input in the ideal state in the above embodiment is performed based on FIG. 3 described above.

In FIG. 3, the S / N ratio is defined by the ratio of the power of the signal component indicated by double hatching to the power of the portion (noise component) indicated by hatching.
Now, when the S / N ratio at the time of L-times oversampling is calculated using the N-order filtering polynomial, the following equation (9) is obtained.
Becomes

[0055]

(Equation 13)

Here, H _N (Ω) is calculated by the aforementioned equation (5).
Is represented by the following equation (10).

[0057]

[Equation 14]

Therefore, the following equation (11) is obtained as the S / N ratio.

[0059]

(Equation 15)

FIG. 5 shows a numerical calculation of the change in the S / N ratio when the L value of the L-times oversampling and the order of the filtering polynomial are variously changed and tabulated.

As is apparent from FIG. 5, the S / N ratio increases as the order N of the filtering polynomial and the oversampling ratio L increase. This fact has been confirmed experimentally. 100 [dB] theoretically
It is also possible to obtain a high S / N ratio.

In the above-described embodiment, when processing a signal of an acoustic signal such as a musical instrument sound recorded by a microphone or the like, a digital filter is applied by the frequency characteristic correction unit 2B after oversampling, and the information storage unit 3 Has been described, but a digital filter may be applied by the frequency characteristic correction unit 2B before oversampling. By doing so, there is an advantage that the signal processing time in the frequency characteristic correction unit 2B can be reduced to 1 / L, and that an acoustic signal such as a musical instrument sound to be recorded can be stored in the information storage unit 3 more quickly. There is.

Next, another embodiment will be described with reference to FIG. The embodiment shown in FIG. 6 has been researched and developed in the case of performing the calculation in the anti-aliasing filter 4 of the piecewise polynomial impulse response more quickly in the embodiment of FIG. 1 described above.

In the embodiment shown in FIG. 6, a filtering polynomial coefficient calculator 10 is additionally provided between the signal processing means 2 and the information storage means 3 in the embodiment of FIG. Other configurations are the same as those of the embodiment of FIG. 1 described above.

The filtering polynomial coefficient calculator 10
When the PCM waveform oversampled by the oversampling unit 2A and subjected to digital filtering by the frequency characteristic correction unit 2B is subjected to predetermined arithmetic processing by the anti-aliasing filter 4, resampled and output, This is for calculating the coefficients of the filtering polynomial calculated by the anti-aliasing filter 4 in advance, and presupposes that the storage capacity of the information storage means 3 is relatively large.

The third-order filtering polynomial will be described in further detail below by way of example.

The third-order filtering polynomial y ₃ (n +
Δ) is given by the following equation (12) from the PCM sample x and the small deviation amount Δ.

[0068]

(Equation 16)

Referring to equation (12), there are many product-sum operations on PCM samples x. Here, when the coefficient of Δ in each row is represented by X _i (n),

Y ₃ (n + Δ) = X ₀ (n) + X ₁ (n)
Δ + X ₂ (n) Δ ² + X ₃ (n) Δ ³ (13)

Here, X ₀ (n), X ₁ (n),...
X ₃ (n) is represented by the following equation (14).

X ₀ (n) = [x (n + 1) + 4x (n) + x (n−1)] / 6 X ₁ (n) = [x (n + 1) −x (n−1)] / 2 X ₂ (n) = [x (n + 1) -2x ( n) + x (n-1) ] / ₂ X 3 (n) = [x (n + 2) -3x ( n + 1) + 3x (n) -x (n-1) / 6 ……… (14)

Therefore, instead of storing the entire PCM sample x in the information storage means 3, the filtering polynomial coefficients X ₀ (n) to X ₃ (n) are calculated in advance and stored in the information storage means 3. The product-sum operation for the PCM sample x can be significantly reduced. This degree of reduction in the amount of computation functions more effectively as the degree of the filtering polynomial is higher.

The other constructions, functions and effects are the same as those of the embodiment shown in FIG.

Here, in each of the above embodiments, the information storage means 3 is detachably provided to the signal processing means 2 (or the filtering polynomial coefficient calculating unit 10) and further to the anti-aliasing filter 4. Is also good. By doing so, the type of reproduced sound and the performance area can be significantly expanded. Also, the one incorporating the filtering polynomial coefficient calculating unit 10 has the advantage that the amount of calculation required for pitch conversion in the system can be significantly reduced.

Further, in each of the above embodiments, the case where the oversampling unit 2A is used is particularly exemplified. However, the oversampling unit 2A may be deleted as necessary.

In each of the embodiments described above, particularly, a configuration in which the components are collectively mounted on the instrument body is disclosed. However, the present invention is not necessarily limited to this.
The information storage means 3 may be provided detachably with respect to its input side and output side. Further, the information storage means 3, the anti-aliasing filter 4 of the piecewise polynomial impulse response, the D / A converter 7, and the pitch control means 5 are housed in the main body of the musical instrument. It may be provided at a predetermined outside location, and the information storage means 3 may be provided detachably with respect to the instrument main body.

Further, the information storage means 3, the anti-aliasing filter 4 for the piecewise polynomial impulse response and the pitch control means 5 are housed in the instrument body, and the information storage means 3, the anti-aliasing for the piecewise polynomial impulse response are stored. Each component other than the filter 4 and the pitch control means 5 may be provided at a predetermined location outside the main body of the musical instrument, and the information storage means 3 may be provided detachably from the main body of the musical instrument.

In this way, it is possible to sufficiently address various demands of the user, and for example, it is possible to reduce the size of the musical instrument by equipping only the minimum necessary performance functions in the musical instrument main body. It is possible to transmit data at a distant place, and it is possible to respond to various demands by storing digital signals of various original sounds in the information storage means in advance, thereby providing a rich system with a wide range of performances and variations. Can be set, and the cost of the musical instrument body can be effectively reduced.

[0080]

According to the present invention, equation (1) having a finite base is
Response to impulse response of continuous N-th order polynomial
Convolved with a discrete waveform to produce a continuous-time output in the form of a piecewise polynomial.
By obtaining the power, high-order FIR operation is not necessary,
No large-scale coefficient table is required, and convolution and re-
Sampling can be performed. Also, the continuous-time polynomial
Generates a virtual analog waveform in the form
Time position can be specified with high accuracy without using sampling
Resampling is possible, and distortions due to time errors
Does not occur. In other words, since the present invention is configured and functions as described above, according to this, the signal processing is performed based on the PCM waveform, so that the time division processing can be performed, and the pitch converting means is provided with a segmented polynomial impulse response. Since it is composed of an anti-aliasing filter, it is possible to greatly reduce the aliasing noise by appropriately selecting and setting the order of the impulse response, and since the frequency characteristic correction unit is provided, the anti-aliasing filter is provided.
Conventionally, the high-frequency component of the filter output can be automatically set and corrected to an appropriate level, and as a result, a reproduced sound with better sound quality can be obtained at an arbitrary pitch (melody) at an arbitrary place. An excellent digital electronic musical instrument system can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of the present invention.

FIG. 2 is an explanatory diagram showing a spectrum of a resampled signal output from an anti-aliasing filter in FIG. 1;

FIG. 3 is an explanatory diagram showing a spectrum of an anti-aliasing filter output (resampling signal) in FIG. 1 for a pulse signal that has been oversampled twice.

FIG. 4 is an explanatory diagram showing the operation of the embodiment shown in FIG.

FIG. 5 is an explanatory diagram showing an effect (a reduction state of the S / N ratio) in the embodiment shown in FIG. 1;

FIG. 6 is a block diagram showing another embodiment.

FIG. 7 is an explanatory diagram showing a conventional example.

FIG. 8 is an explanatory diagram showing another conventional example.

FIG. 9 is an explanatory diagram showing another conventional example.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 A / D conversion part 2 Signal processing means 2A Oversampling part 2B Frequency characteristic correction part 3 Information storage means 4 Anti-aliasing filter as pitch conversion means 5 Pitch control means 6 Keyboard 7 D / A conversion part 10 Filtering polynomial coefficient Calculation department

Claims (2)

(57) [Claims] A part of a performance sound such as a musical instrument sound recorded by a microphone or the like gives a predetermined pitch difference to digitized acoustic information.
A pitch conversion means for resampling after was example, a pitch control unit for controlling based on the operation of the pitch conversion means to an external command, the pitch conversion means, FIR and IIR discrete filter coefficients
Convolves a number with a discrete waveform to obtain a discrete-time sampled value output
Rather than having the finite base, it is equivalent to the following equation h _N (t)
N-th order polynomial as continuous time impulse response
Convolution with a discrete waveform to produce a continuous-time output in the form of a piecewise polynomial
A digital electronic musical instrument system, wherein the digital electronic musical instrument system is obtained and resampled . (Equation 1) 2. The N-th order polynomial h _N (t) is expressed by the following equation.
As shown, h ₀ (t) is represented by N convolutions of itself with itself.
Is, [number 2] When the Fourier transform H N _(Omega) is expressed by the following equation, [Equation 3] The input signal of the pitch conversion means is expressed as “H _N (Ω)”.
The frequency inverse characteristic “H _N ⁻¹ (Ω); 0 ≦ Ω <π”
2. A frequency characteristic correction unit for performing a filtering process.
Placing digital electronic musical instrument system.