JP5036677B2 - Variable characteristic type signal conversion apparatus and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a variable characteristic signal conversion apparatus and signal conversion method capable of playing back signals with sound quality and image quality according to user preferences for discrete data. <P>SOLUTION: In a signal conversion processing part 2, by calculating an interpolation value reflecting the value of a variable parameter &alpha; to be multiplied by the value of a control sampling function c<SB>0</SB>(t), the characteristics of analog signals obtained by interpolation processing via a sampling function s<SB>N</SB>(t) are adjusted corresponding to the variable parameter &alpha; by changing the value of the variable parameter &alpha;. Thus, the user can properly change the variable parameter &alpha; in response to various conditions such as a music playback environment, a sound source and a music tone to reproduce high-quality music with the sound quality desired by the user in which the frequency characteristics of the analog signals are changed. <P>COPYRIGHT: (C)2010,JPO&amp;INPIT

Description

  The present invention relates to a conversion apparatus and method for converting a discrete signal into a continuous signal, for example, by interpolating between discrete data arranged in a time direction sampled at a predetermined sampling frequency, and discrete data at a frequency higher than the sampling frequency at the time of input. Alternatively, the present invention relates to a variable characteristic signal conversion apparatus and method that can be applied when generating an analog signal and that is suitable for obtaining the frequency characteristic of an output signal obtained by changing the frequency characteristic of an input signal.

  In particular, the present invention is suitable for changing the sound quality according to the genre of music and the listener's preference for acoustic signals, and is a technique suitable for image quality adjustment for photographs and videos.

  In this specification, the generation of discrete data at high frequency discrete intervals and the generation of analog signals are referred to as “analog signal generation” as the same processing. Further, the case where the value of the function has a finite value other than 0 in the local region and becomes 0 in the other region will be referred to as a “finite platform”.

Conventionally, when generating an analog signal from discrete data such as digital data, a sampling function represented by Shannon's sampling theorem has been widely used. Here, as shown in FIG. 14, Shannon's sampling function becomes 1 only at the sample position of t = 0, and becomes 0 at all other sample positions, and theoretically from −∞ to + ∞. A waveform in which vibration continues indefinitely. For this reason, when an interpolation process between discrete data is actually executed by various processors using a Shannon sampling function, the process is forcibly terminated in a finite interval. As a result, there is a problem that an error due to truncation occurs. In order to solve such a problem, a sampling function that has no censoring error and can reproduce even higher-order band components and converges in a finite range has been devised (see, for example, Patent Document 1). ). In this sampling function, since it converges to 0 at two sample positions before and after the origin, it is confirmed that signal reproduction can be performed with a small amount of calculation and that there is a band up to a high frequency.
International Publication No. 99/38090 Pamphlet

  Incidentally, the sampling function used in the prior art of Patent Document 1 has a problem that the waveform shape is fixed, and the characteristics of sound quality and image quality cannot be changed. There is a demand for a tailor-made method that can freely adjust the sound quality and the like according to various conditions such as a music reproduction environment, a sound source, and a tune for various users such as a hearing impaired person and an elderly person. However, in the past, such a request could not be satisfied, and an equalizer that performs frequency analysis of an input signal and adjusts sound pressure for each frequency band is used. In this case, frequency band separation is required, and further, since sound pressure is adjusted for each frequency band, continuous adjustment cannot be performed when the frequency changes, resulting in sound quality without smoothness. In addition, in an image such as a video or a photograph, image quality adjustment according to human visual characteristics is necessary. However, in the current situation where color adjustment is performed by an expert, such image quality adjustment is difficult.

  The present invention has been created in view of the above points, and an object of the present invention is to provide variable characteristic signal conversion capable of reproducing a signal with discrete sound quality and image quality according to user preferences. An apparatus and a signal conversion method are provided.

In order to solve the above-described problem, the variable characteristic signal conversion apparatus of the present invention is a sampling system configured by a linear combination of a basic sampling function and a control sampling function each represented by a piecewise polynomial of a finite stage. A function processing unit that calculates an interpolated value between the discrete data by performing a convolution operation on a plurality of discrete data using a function, and when the sample position of the discrete data is t, the basic sampling function is t -1 to +1 is the range of the finite stage, becomes 1 at t = 0, converges to 0 toward t = ± 1, and the control sampling function is set for each sample included in the range of the finite stage. It is 0 at position t .

Further, the variable characteristic signal conversion method of the present invention uses a sampling function composed of a linear combination of a basic sampling function and a control sampling function, each of which is represented by a piecewise polynomial of a finite platform. A function processing step for calculating an interpolated value between discrete data by a function processing unit by performing a convolution operation on the data, and when the sample position of the discrete data is t, the basic sampling function is in a finite range. In this case, the signal becomes 1 at t = 0, converges to 0 toward t = ± 1, and the control sampling function becomes 0 at each sample position t included in the range of the finite platform .

  Further, the data processing apparatus further includes a discrete data extraction unit that extracts a predetermined number of discrete data having a continuous input order from a plurality of discrete data that are sequentially input, and the function processing unit includes a predetermined number of discrete data extracted by the discrete data extraction unit. It is desirable to calculate an interpolation value using discrete data. Alternatively, the method further includes a discrete data extraction step by which a discrete data extraction unit extracts a predetermined number of discrete data having a continuous input order from a plurality of discrete data that are sequentially input. It is desirable to calculate an interpolation value using a predetermined number of extracted discrete data.

  Further, when the basic sampling function described above is f (t), the control sampling function is C (t), and a parameter that can be arbitrarily set by the user is α, the function processing unit converts the linear combination to f (t It is desirable to calculate by t) + αC (t). Alternatively, in the function processing step described above, it is desirable to calculate the linear combination by f (t) + αC (t).

  The function processing unit described above includes a basic term operation unit that performs a convolution operation on a plurality of discrete data using a basic sampling function, and a control term operation that performs a convolution operation on a plurality of discrete data using a control sampling function. And a coefficient multiplication unit that multiplies the calculation result by the control term calculation unit by a parameter. Alternatively, the function processing step described above includes a basic term calculation step in which a basic term calculation unit performs a convolution operation on a plurality of discrete data using a basic sampling function, and a convolution operation on a plurality of discrete data using a control sampling function. It is desirable to include a control term calculation step performed by the control term calculation unit, and a coefficient multiplication step performed by the coefficient multiplication unit for multiplying the calculation result in the control term calculation step by the parameter.

  In addition, the function processing unit described above specifies a sampling function obtained by linearly combining a basic sampling function and a control sampling function when a parameter value is specified, and uses this sampling function. It is desirable to calculate the interpolation value by performing a convolution operation on the discrete data. Alternatively, in the function processing step described above, when a parameter value is specified, a sampling function obtained by linearly combining a basic sampling function and a control sampling function is specified, and this sampling function is used. It is desirable to calculate the interpolation value by performing a convolution operation on the discrete data.

  Further, it is desirable to further include a parameter setting unit that arbitrarily sets a parameter value in accordance with a user instruction. Alternatively, it is desirable to further include a parameter setting step for arbitrarily setting a parameter value by the parameter setting unit in accordance with a user instruction.

  Moreover, it is desirable to further include a selector that selects one of a plurality of preset parameter values when operated by the user. Alternatively, it is desirable to further include a parameter selection step of selecting one of a plurality of preset parameter values in accordance with a user operation instruction.

  Further, the basic sampling function described above is a piecewise polynomial that can be differentiated only once in the interval [−1, 1] of the sample position of the discrete data, and the other intervals are functions that are represented by 0 uniformly. The control sampling function is preferably a piecewise polynomial that can be differentiated only once in the interval [−2, 2] of the sample position of the discrete data, and is a function that is uniformly 0 in the other intervals. .

  When the sample position of the discrete data described above is t, the basic sampling function f (t) is

Represented by
When the control sampling function is C ₀ (t) = C _r (t) + C _r (−t), C _r (t) is

It is desirable that

  Further, it is desirable to program a programmable signal processor that forms an arithmetic configuration having a user-desired control form based on program data designated by the user.

  In addition, each of the basic sampling function and the control sampling function described above tabulates function values calculated in advance according to a predetermined number of sections between discrete data and holds them as table values. It is desirable that a convolution operation is performed on discrete data using the function value thus determined. Alternatively, in the functionalization processing step described above, it is desirable that a convolution operation is performed on discrete data using the table-like function values.

  In addition, for a plurality of different division numbers between the discrete data described above, it is preferable that a table value corresponding to the division number of the least common multiple of the division numbers is calculated and held in advance.

  In the present invention, by constructing the sampling function by a linear combination of the basic sampling function and the control sampling function, it becomes easy to modify the entire sampling function by modifying the term related to the control sampling function, and the discrete function It is possible to reproduce a signal with sound quality and image quality according to the user's preference.

  In particular, when the basic sampling function is f (t), the control sampling function is C (t), and the parameter arbitrarily settable by the user is α, the above linear combination is expressed as f (t) + αC (t). Thus, the shape of the sampling function can be changed by changing the value of the parameter α, and it becomes easy to change the characteristics of the obtained interpolation value.

  Hereinafter, a variable characteristic signal converter according to an embodiment to which the present invention is applied will be described in detail with reference to the drawings.

(1) Basic Concept of the Present Invention FIG. 1 is a diagram showing the waveform shapes of the basic sampling function f (t) and the control sampling function c ₀ (t) that constitute the sampling function used in the interpolation processing of the present invention. It is. In FIG. 1, the horizontal axis indicates the sample position t of discrete data, and the vertical axis indicates each sample function value. A sampling function s ₂ (t) composed of a basic sampling function f (t) and a control sampling function c ₀ (t) between sampling positions [−2, 2] of the discrete data is expressed by the following equation. The

Generally, if the control sampling function is c _k (t) and c _k (t) = c _r (t−k) + c _r (−t−k), the sampling position [−N , N], the sampling function s _N (t) is expressed by the following equation.

Α _k represents a variable parameter and has an arbitrary value that can be set by the user. The value may not be variable depending on k, such as α ₁ = α ₂ = α ₃ .

  The basic sampling function f (t) is a finite number of piecewise polynomial functions that focus on differentiability. For example, the basic sampling function f (t) can be differentiated only once in the entire region, and the sample position t along the horizontal axis varies from −1 to +1. (That is, a function having a finite value other than 0 when in the section [-1, 1]) and being represented by 0 uniformly in other sections. Specifically, the basic sampling function f (t) is an nth-order polynomial function in each subsection obtained by dividing the section [-1, 1] into two or more, and is continuous (values) at the boundaries of each subsection. And slope are continuous functions). This basic sampling function f (t) indicates a convex waveform shape that can be differentiated n-1 times (n is an integer of 2 or more) in the entire range, and becomes 1 only at a sample position of t = 0. It has a feature that it converges to 0 toward t = ± 1, and maintains the value of 0 as it is until the sample position of t = ± 2.

  The basic sampling function f (t) may be a function of a finite impulse response waveform, or may be a continuous n-order piecewise polynomial function that can be differentiated at least once at an arbitrary position in the sample position section. .

  As a specific example, a basic sampling function f (t) expressed by a quadratic piecewise polynomial function is expressed by the following equation.

Then, by performing superposition based on each discrete data using this basic sampling function f (t), a value at an arbitrary point between the discrete data can be temporarily interpolated.

On the other hand, the control sampling function c ₀ (t) is a finite number of piecewise polynomial functions focusing on differentiability, and is expressed by an nth-order polynomial function as in the case of the basic sampling function. For example, it can be differentiated only once in the entire region, and has a finite value other than 0 when the sample position t along the horizontal axis is in the range of −2 to +2 (that is, the interval [−2, 2]). In other sections, it is a function that is uniformly represented by 0. The control sampling function c ₀ (t) shows a waveform that can be differentiated only once in the entire range, and has a feature that it becomes 0 at each sample position of t = 0, ± 1, ± 2.

The control sampling function c ₀ (t) may be a function of a finite number of impulse response waveforms, and may be a continuous n-order piecewise polynomial function that can be differentiated at least once at an arbitrary position in the sample position section.

Here, the control sampling function c ₀ (t) is expressed by the control sampling function c ₀ (t) = c _r (t) + c _r (−t) as described above. This c _r (t) is expressed by the following expression when expressed by a second-order piecewise polynomial, for example.

By performing superposition based on each discrete data using the control sampling function c ₀ (t), a value between the discrete data can be temporarily interpolated.

In this way, the temporary interpolation value calculated based on the basic sampling function f (t) (hereinafter referred to as “basic interpolation value”) and the control sampling function c ₀ (t) are calculated. By linearly adding the provisional interpolation values (hereinafter referred to as “control interpolation values”), values at arbitrary points between discrete data can be interpolated.

Incidentally, in the case of a sampling function represented by a quadratic piecewise polynomial function, the following six conditions are satisfied in the linear combination of the basic sampling function f (t) and the control sampling function c ₀ (t).
First, S ₂ (0) = 1 and S ₂ (± 1) = S ₂ (± 2) = 0.
Secondly, it is symmetric with respect to the even function, that is, the y-axis.
Thirdly, it should be zero in the sample position interval [−∞, −2], [2, + ∞].
The fourth is that each section [n / 2, (n + 1) / 2] (−4 ≦ n ≦ 3) is at most a quadratic polynomial.
・ Fifth, it should be continuous and differentiable once in all sections.
-Sixth, in the sample position section [-1/2, 1/2], it is expressed by the following equation.

The sampling function s ₂ (t) when N = 2 is simply described as the sampling function s _N (t) for convenience of explanation.

In addition to this, at this time, the control sampling function c ₀ (t) is multiplied by a variable parameter α set with an arbitrary numerical value. As a result, the control sampling function c ₀ (t) corresponds to the numerical value of the variable parameter α between the sample positions −2 and +2 while keeping 0 at the sample positions of t = 0, ± 1, ± 2. The waveform amplitude can be changed. As a result, the calculation result by the convolution calculation using the control sampling function c ₀ (t) can be changed. As described above, the variable parameter α is changed in numerical value, thereby changing the frequency characteristic of the analog signal calculated by the sampling function s _N (t) and adjusting the change state of the signal. It has the feature that it can.

Therefore, in the present invention, when the interpolation result is obtained by linearly adding the calculation result of the basic sampling function f (t) and the calculation result of the control sampling function c ₀ (t), the control sampling function c _The interpolation value can be adjusted by the variable parameter α multiplied by ₀ (t). As a result, it is possible to freely adjust the frequency characteristic of the analog signal obtained by interpolating between these discrete data with the interpolation value by the variable parameter α.

(2) Overall Configuration of Variable Characteristic Signal Conversion Device Next, a variable characteristic signal conversion device that performs interpolation processing using the sampling function s _N (t) described above will be described.

  FIG. 2 is a diagram illustrating a configuration of a variable characteristic signal conversion device according to an embodiment. In FIG. 2, reference numeral 1 denotes a variable characteristic signal conversion device, and a signal conversion processing unit 2 is provided by being programmed in a programmable signal processor 3. In the programmable signal processor 3, a plurality of circuit blocks and wiring blocks are regularly arranged on a semiconductor chip, and electrical connection or non-connection of circuits is programmed in the circuit blocks and wiring blocks. A large number of devices that can be used are arranged, and the user can program (define) these devices so that connections inside the blocks and connections between the blocks can be designed in the field (site of use).

  This variable characteristic type signal converter 1 is obtained by a personal computer (PC) 5 via an external interface (external I / F) 4 and the basic sampling function and the control sampling function shown in the above-described equations 9 and 10. By programming various sampling functions such as sampling functions completely different from these, the connection state between the circuit block and the wiring block of the programmable signal processor 3 is changed, and interpolation processing by various sampling functions is performed. The circuit configuration can be changed to hardware that can be executed. Thereby, in this variable characteristic type signal converter 1, since it is possible to change to a circuit configuration desired by the user simply by programming the programmable signal processor 3, various sampling functions can be used when searching for an optimal sampling function. Accordingly, it is not necessary to actually manufacture the circuit board each time, and the cost can be reduced accordingly.

  In the above-described embodiment, FIG. 2 shows a method for realizing the programmable signal processor 3. However, the present invention is not limited to this, and an FPGA (Field Programmable Gate Array) or DSP (Digital Signal Processor) is used. Such a programmable signal processing device can also be realized. For example, an arithmetic device composed of a CPU (Central Processing Unit) and a memory may be used.

In the variable characteristic signal conversion apparatus 1 of the present embodiment, the basic sample shown in Equation 9 that satisfies the condition of the sampling function s _N (t) shown in Equation 8 from the personal computer 5 via the external interface 4. The programmable signal processor 3 is programmed with a quantization function f (t) and a control sampling function c ₀ (t) in which c _r (t) is expressed by Equation 10.

  Thereby, in the variable characteristic signal conversion device 1, the programmable signal processor 3 performs overall control according to a predetermined program, thereby reproducing various recording media such as CD and DVD by the input unit 6, and as a result. The obtained discrete data arranged in the time direction are sequentially sent to the signal conversion processing unit 2. Incidentally, the discrete data is discrete data obtained by, for example, sampling a continuous signal that changes smoothly at regular time intervals and quantizing the sampling data obtained as a result.

The programmable signal processor 3 is connected to a parameter setting unit 7 that can arbitrarily set the numerical value of the variable parameter α in accordance with a user instruction. When the user sets the variable parameter α to an arbitrary numerical value by the parameter setting unit 7, information indicating the set numerical value is sent from the parameter setting unit 7 to the signal conversion processing unit 2. The signal conversion processing unit 2 as a signal processing device performs a so-called oversampling process that artificially increases the sampling frequency by interpolating between discrete data using the sampling function s _N (t), and sets the numerical value of the variable parameter α. Is calculated and sent to the output unit 8.

  The output unit 8 includes, for example, a digital-analog converter, an amplifier, and a speaker. When an interpolation value is input from the signal conversion processing unit 2 at a predetermined period, the interpolation value is converted to an analog corresponding to the interpolation value. It converts it into a signal and outputs voice, music, etc. based on this analog signal.

  As described above, the variable characteristic signal conversion device 1 according to the present embodiment generates an analog signal having desired characteristics reflecting the numerical value of the variable parameter α by changing the numerical value of the variable parameter α. Can do.

The programmable signal processor 3 is connected to a selector 10 having a plurality of selector buttons 9a, 9b, 9c. In this selector 10, a variable parameter α having a different numerical value is associated with each of the selector buttons 9a, 9b, 9c, and when any one of the selector buttons 9a, 9b, 9c is selected by a user operation. Then, the value of the variable parameter α corresponding to the selected selector button is multiplied by the control sampling function c ₀ (t), and interpolation processing by the sampling function s _N (t) is executed.

Specifically, in the case of the present embodiment, for example, when the selector button 9a is selected, the interpolation process is executed by the sampling function s _N (t) with the variable parameter α set to −1.5. When the other selector button 9b is selected, interpolation processing is executed by a sampling function s _N (t) with the variable parameter α set to −0.25. When another selector button 9c is selected, an interpolation process is executed using a sampling function s _N (t) with a variable parameter α set to 1.5.

  Thus, in the variable characteristic signal conversion device 1, the user can set the variable parameter α to an arbitrary numerical value by the parameter setting unit 7, and select any one of the selector buttons 9a, 9b, and 9c. As a result, the interpolation processing using the desired variable parameter α can be easily executed without finely setting the variable parameter α each time by the parameter setting unit 7.

(3) Circuit Configuration of Signal Conversion Processing Unit (3-1) Outline Explanation of Interpolation Processing in Signal Conversion Processing Unit FIG. 3 is a diagram showing a configuration of the signal conversion processing unit 2 in the programmable signal processing processor 3. The signal conversion processing unit 2 sequentially extracts a predetermined number (in this case, four) of discrete data and holds the discrete data extraction unit 15 and the predetermined number of discrete data extracted and held by the discrete data extraction unit 15 once. And a function processing unit 14 that performs interpolation processing using these discrete data, and interpolates data between discrete data sequentially input from the input unit 6 at a predetermined time interval.

The function processing unit 14 includes a basic term calculation unit 16 that calculates a term of the basic sampling function f (t) in the sampling function s _N (t) based on the discrete data, and a sampling function based on the discrete data. a control term computing unit 17 that computes a term of the control sampling function c ₀ (t) in s _N (t), a coefficient multiplication unit 18 that multiplies the calculation result of the control term computing unit 17 by a variable parameter α, The linear addition unit 19 linearly adds the calculation result of the basic term calculation unit 16 and the calculation result of the coefficient multiplication unit 18.

  In the case of this embodiment, the discrete data extraction unit 15 extracts the last four discrete data from the sequentially input discrete data, and holds the four discrete data until new discrete data is input next. Then, these four discrete data are sent to the basic term calculation unit 16 and the control term calculation unit 17, respectively.

  The basic term calculation unit 16 stores a basic sampling function f (t) in a predetermined storage means (not shown), and when an interpolation position between discrete data is designated, the interpolation position and the discrete data The value of the basic sampling function f (t) is calculated based on the distance between. The basic term calculation unit 16 calculates the value of the basic sampling function f (t) for each of the four discrete data transmitted from the discrete data extraction unit 15. The basic term calculation unit 16 multiplies the values of the four basic sampling functions f (t) obtained for each discrete data by the values of the corresponding discrete data, and then handles the four discrete data. The convolution calculation is performed, and the calculation result of the convolution calculation is sent to the linear adder 19.

Further, the control term calculation unit 17 stores a control sampling function c ₀ (t) in a predetermined storage means (not shown), and when an interpolation position is designated, the interpolation between the interpolation position and the discrete data is performed. The value of the control sampling function c ₀ (t) is calculated based on the distance between them. The control term calculation unit 17 calculates the value of the control sampling function c ₀ (t) for each of the four discrete data transmitted from the discrete data extraction unit 15. In addition, the control term calculation unit 17 multiplies the values of the four control sampling functions c ₀ (t) obtained for each discrete data by the corresponding discrete data values, and then adds them. A convolution operation corresponding to the four discrete data is performed, and a calculation result of the convolution operation is sent to the coefficient multiplication unit 18.

The coefficient multiplying unit 18 multiplies the calculation result of the convolution calculation of the control sample function c ₀ (t) received from the control term calculating unit 17 by the variable parameter α, and the variable parameter multiplication result obtained as a result is linearly added unit 19. To send. The linear addition unit 19 linearly adds the calculation result of the convolution calculation of the basic sampling function f (t) received from the basic term calculation unit 16 and the variable parameter multiplication result received from the coefficient multiplication unit 18 to obtain 4 A linear addition result corresponding to two discrete data is obtained. A value obtained by this linear addition is an interpolation value at an interpolation position between two predetermined discrete data. The value of this interpolation position is updated at a predetermined time interval, specifically, every 1 / N period (= T / N) of the period T corresponding to the input interval of discrete data.

  The processing by the function processing unit 14 described above is the function processing step operation, the processing by the discrete data extraction unit 15 is the operation of the discrete data extraction step, the processing by the basic term calculation unit 16 is the operation of the basic term calculation step, and the control term. The selector 10 is used for the operation of the control term calculation step for the processing by the calculation unit 17, the operation for the coefficient multiplication step for the processing by the coefficient multiplication unit 18, and the parameter setting processing for the parameter setting step by the parameter setting unit 7. Each parameter setting process corresponds to the operation of the parameter selection step.

(3-2) Specific Example of Obtaining Interpolation Value Based on Four Discrete Data Next, an interpolation process for calculating an interpolation value between two predetermined discrete data based on four discrete data arranged successively in time Will be described.

  FIG. 4 is a diagram illustrating a positional relationship between four continuous discrete data and a point of interest that is an interpolation position. Hereinafter, each value of discrete data d1, d2, d3, d4 sequentially input corresponding to each of the sample positions t1, t2, t3, t4 is represented by Y (t1), Y (t2), Y (t3), Assume that Y (t4) is used, and a predetermined position t0 (distance b from t2) between the sample positions t2 and t3 is used as an interpolation position to obtain an interpolation value y corresponding to this interpolation position.

Since the sampling function s _N (t) used in the present embodiment converges to 0 at the sample position of t = ± 2, the discrete data d1, d2, d3, d4 up to t = ± 2 may be taken into consideration. . Therefore, when obtaining the interpolation value y shown in FIG. 4, only four discrete data d1, d2, d3, and d4 corresponding to t = t1, t2, t3, and t4 need to be considered, and the amount of calculation Can be greatly reduced. In addition, discrete data (not shown) of t = ± 3 or more should be considered originally, but is not ignored in consideration of the amount of calculation and accuracy, and should be considered theoretically. There is no truncation error.

  FIG. 5 is a diagram illustrating a specific configuration of the signal conversion processing unit 2. As shown in FIG. 5, the discrete data extraction unit 15 includes three shift circuits 20a, 20b, and 20c connected in cascade. When continuous discrete data is input, the discrete data is shifted, for example, at a CD sampling period (44.1 kHz) for each shift circuit 20a, 20b, 20c, and the previous discrete circuit is shifted by each shift circuit 20a, 20b, 20c. One piece of data d1, d2, d3, d4 is extracted and held.

  That is, when the discrete data extraction unit 15 receives four continuous discrete data d1, d2, d3, and d4, the latest discrete data d4 is directly used as the basic term calculation circuit 21a of the basic term calculation unit 16 and the control term calculation. To the control term calculation circuit 22a of the unit 17.

  Also, the discrete data extraction unit 15 sends a discrete data string composed of four continuous discrete data d1, d2, d3, and d4 to the shift circuit 20a, and the discrete data string is shifted by the shift circuit 20a to obtain the latest discrete data string. The previous discrete data d3 is extracted from the data d4 and is sent to the basic term calculation circuit 21b of the basic term calculation unit 16 and the control term calculation circuit 22b of the control term calculation unit 17, respectively.

  Further, the discrete data extraction unit 15 sequentially transmits the discrete data string to the remaining shift circuits 20b and 20c, and further shifts the discrete data string by the shift circuit 20b, so that the two previous data from the latest discrete data d4 are shifted. The discrete data d2 is sent to the basic term calculation circuit 21c and the control term calculation circuit 22c, respectively, and the discrete data string is further shifted by the shift circuit 20c to calculate the discrete data d1 three times before the latest discrete data d4. The data is sent to the circuit 21d and the control term calculation circuit 22d.

  FIG. 6 is a diagram showing an outline of the interpolation processing for the predetermined interpolation position t0 in the basic term calculation unit 16. FIG. 7 is a diagram showing an outline of interpolation processing for a predetermined interpolation position t0 in the control term calculation unit 17.

  As the contents of the interpolation processing, as described above, first of all, the calculation processing for calculating the basic interpolation value in the basic term calculation unit 16 (hereinafter, this calculation processing is referred to as “basic interpolation value calculation processing”), the control term A calculation process for calculating the control interpolation value in the calculation unit 17 and the coefficient multiplication unit 18 (hereinafter, this calculation process is referred to as “control interpolation value calculation process”) is executed. Hereinafter, basic interpolation value calculation processing and control interpolation value calculation processing will be described with reference to FIGS. 6 and 7.

(3-2-1) Basic Interpolation Value Calculation Processing As shown in FIGS. 6A to 6D, the basic interpolation value calculation processing is performed for each sample position t1, t2, t3, t4. The peak heights at t = 0 (center position) of the sampling function f (t) are matched, and the value of each basic sampling function f (t) at the interpolation position t0 at this time is obtained.

  Focusing on the discrete data d1 at the sample position t1 shown in FIG. 6A, the distance between the interpolation position t0 and the sample position t1 is 1 + b. Therefore, the value of the basic sampling function at the interpolation position t0 when the center position of the basic sampling function f (t) is aligned with the sample position t1 is f (1 + b). Actually, in order to match the peak height of the center position of the basic sampling function f (t) so as to match the value Y (t1) of the discrete data d1, the above-described f (1 + b) is multiplied by Y (t1). The value f (1 + b) · Y (t1) is the desired value. The calculation of f (1 + b) is performed by the basic term calculation circuit 21a of the basic term calculation unit 16, and the calculation of multiplying f (1 + b) by Y (t1) is performed by the basic term multiplication circuit 23a of the basic term calculation unit 16. (FIG. 5).

  Similarly, focusing on the value Y (t2) of the discrete data d2 at the sample position t2 shown in FIG. 6B, the distance between the interpolation position t0 and the sample position t2 is b. Therefore, the value of the basic sampling function at the interpolation position t0 when the center position of the basic sampling function f (t) is aligned with the sampling position t2 is f (b). Actually, in order to match the peak height of the center position of the basic sampling function f (t) so as to coincide with the value Y (t2) of the discrete data d2, the above-described f (b) is multiplied by Y (t2). The value f (b) · Y (t2) is a desired value. Calculation of f (b) is performed by the basic term calculation circuit 21b of the basic term calculation unit 16, and calculation for multiplying f (b) by Y (t2) is performed by the basic term multiplication circuit 23b of the basic term calculation unit 16. (FIG. 5).

  Focusing on the value Y (t3) of the discrete data d3 at the sample position t3 shown in FIG. 6C, the distance between the interpolation position t0 and the sample position t3 is 1-b. Therefore, the value of the basic sampling function at the interpolation position t0 when the center position of the basic sampling function f (t) is aligned with the sampling position t3 is f (1-b). Actually, in order to match the peak height of the center position of the basic sampling function f (t) so as to coincide with the value Y (t3) of the discrete data, the above-described f (1-b) is multiplied by Y (t3). The obtained value f (1-b) · Y (t3) is a desired value. The calculation of f (1-b) is performed by the basic term calculation circuit 21c of the basic term calculation unit 16, and the calculation of multiplying f (1-b) by Y (t3) is the basic term multiplication circuit of the basic term calculation unit 16. 23c (FIG. 5).

  Focusing on the value Y (t4) of the discrete data d4 at the sample position t4 shown in FIG. 6D, the distance between the interpolation position t0 and the sample position t4 is 2-b. Therefore, the value of the basic sampling function at the interpolation position t0 when the center position of the basic sampling function f (t) is aligned with the sampling position t4 is f (2-b). Actually, in order to match the peak height of the center position of the basic sampling function f (2-b) so as to coincide with the value Y (t4) of the discrete data d4, the above-described f (2-b) is changed to Y ( t4) The multiplied value f (2-b) · Y (t4) is a desired value. The calculation of f (2-b) is performed by the basic term calculation circuit 21d of the basic term calculation unit 16, and the calculation of multiplying f (2-b) by Y (t4) is the basic term multiplication circuit of the basic term calculation unit 16. 23d (FIG. 5).

  Next, in the basic term calculation unit 16, four values f (1 + b) · Y (t1), f (b) · Y (t2), and f (1−1) obtained corresponding to the point of interest at the interpolation position t0. b) · Y (t3) and f (2-b) · Y (t4) are convolutionally calculated by the basic term convolution circuit 24, and a basic interpolation value ya corresponding to the point of interest is calculated. Note that the operation performed by the entire basic term operation unit 16 is a convolution operation, and the basic term convolution circuit 24 simply adds the multiplication results of the respective basic term multiplication circuits 23a to 23d. In this embodiment, the values f (1 + b) · Y (t1) and f (2-b) · Y (t4) obtained corresponding to the point of interest at the interpolation position t0 are shown in FIG. And since it becomes 0 as shown in (D), the basic interpolation value ya becomes (f (b) · Y (t2)) + (f (1−b) · Y (t3)).

(3-2-2) Control Interpolation Value Calculation Processing On the other hand, the contents of the control interpolation value calculation processing are as shown in FIGS. 7A to 7D for each sample position t1, t2, t3, t4. , to match the t = 0 (center position) of the control sampling function c ₀ (t), the discrete data d1 corresponding to each control sampling function _{c 0 (t), d2,} d3, d4 of the value Y (t1 ), Y (t2), Y (t3), and Y (t4), and the value of each control sampling function c ₀ (t) at the interpolation position t0 at this time is obtained.

Focusing on the value Y (t1) of the discrete data d1 at the sample position t1 shown in FIG. 7A, the distance between the interpolation position t0 and the sample position t1 is 1 + b. Therefore, the value of the control sampling function at the interpolation position t0 when the center position of the control sampling function c ₀ (t) is aligned with the sampling position t1 is c ₀ (1 + b). Actually, in order to match the waveform height of the control sampling function c ₀ (t) in correspondence with the value Y (t 1) of the discrete data d 1, a value c obtained by multiplying the above c ₀ (1 + b) by Y (t 1) ₀ (1 + b) · Y (t1) is a desired value. The calculation of c ₀ (1 + b) is performed by the control term calculation circuit 22a of the control term calculation unit 17, and the calculation of multiplying c ₀ (1 + b) by Y (t1) is performed by the control term multiplication circuit 25a of the control term calculation unit 17. Performed (FIG. 5).

Similarly, focusing on the value Y (t2) of the discrete data d2 at the sample position t2 shown in FIG. 7B, the distance between the interpolation position t0 and the sample position t2 is b. Therefore, the value of the control sampling function at the interpolation position t0 when the center position of the control sampling function c ₀ (t) is aligned with the sample position t2 is c ₀ (b). Actually, in order to match the waveform height of the control sampling function c ₀ (t) in correspondence with the value Y (t 2) of the discrete data d 2, a value c obtained by multiplying the above c ₀ (b) by Y (t 2). ₀ (b) · Y (t2) is a desired value. The calculation of c ₀ (b) is performed by the control term calculation circuit 22b of the control term calculation unit 17, and the calculation of multiplying c ₀ (b) by Y (t2) is performed by the control term multiplication circuit 25b of the control term calculation unit 17. Performed (FIG. 5).

Focusing on the value Y (t3) of the discrete data d3 at the sample position t3 shown in FIG. 7C, the distance between the interpolation position t0 and the sample position t3 is 1-b. Therefore, the value of the control sampling function at the interpolation position t0 when the center position of the control sampling function c ₀ (t) is aligned with the sample position t3 is c ₀ (1-b). Actually, in order to match the waveform height of the control sampling function c ₀ (t) in correspondence with the value Y (t 3) of the discrete data d ₃ , the above c ₀ (1-b) is multiplied by Y (t 3). The value c ₀ (1-b) · Y (t3) is the desired value. The calculation of c ₀ (1-b) is performed by the control term calculation circuit 22c of the control term calculation unit 17, and the calculation of multiplying c ₀ (1-b) by Y (t3) is the control term of the control term calculation unit 17. This is performed by the multiplication circuit 25c (FIG. 5).

Focusing on the value Y (t4) of the discrete data d4 at the sample position t4 shown in FIG. 7D, the distance between the interpolation position t0 and the sample position t4 is 2-b. Therefore, the value of the control sampling function at the interpolation position t0 when the center position of the control sampling function c ₀ (t) is aligned with the sampling position t4 is c ₀ (2-b). Actually, in order to match the waveform height of the control sampling function c ₀ (2-b) in correspondence with the value Y (t4) of the discrete data d4, the above c ₀ (2-b) is changed to Y (t4). The multiplied value c ₀ (2-b) · Y (t4) is the value to be obtained. The calculation of c ₀ (2-b) is performed by the control term calculation circuit 22d of the control term calculation unit 17, and the calculation of multiplying c ₀ (2-b) by Y (t4) is the control term of the control term calculation unit 17. This is performed by the multiplication circuit 25d (FIG. 5).

Next, in the control term calculation unit 17, four values c ₀ (1 + b) · Y (t1), c ₀ (b) · Y (t2), c ₀ obtained corresponding to the point of interest at the interpolation position t0 are obtained. (1-b) · Y (t3), c ₀ (2-b) · Y (t4) are convolved by the control term convolution circuit 26, and the coefficient multiplier 18 multiplies the variable parameter α, thereby paying attention. A control interpolation value yb corresponding to the point is calculated. The operation performed by the entire control term operation unit 17 is a convolution operation, and the control term convolution circuit 26 simply adds the multiplication results of the control term multiplication circuits 25a to 25d.

(3-2-3) Interpolation Value Calculation Processing The linear addition unit 19 is calculated by the basic interpolation value ya corresponding to the point of interest calculated by the basic term calculation unit 16, the control term calculation unit 17, and the coefficient multiplication unit 18. The interpolation value y at the interpolation position t0 is output by linearly adding the control interpolation value yb corresponding to the target point.

(3-3) Interpolation processing result when variable parameter value is changed In addition to the configuration described above, in the signal conversion processing unit 2, the parameter setting unit 7 changes the value of the variable parameter α of the coefficient multiplication unit 18. As a result, the value of the sampling function s _N (t) is changed, and as a result, the interpolation value y varies and the frequency characteristic of the analog signal changes. In the following, regarding how the sampling function s _N (t) changes when the variable parameter α is changed, the waveform indicated by the basic sampling function f (t) shown in FIG. Description will be made by paying attention to a waveform obtained by synthesizing the waveform indicated by the function c ₀ (t).

FIG. 8 is a diagram showing a waveform of the sampling function s _N (t). As shown in FIG. 8, the waveform of the sampling function s _N (t) obtained by synthesizing the waveform indicated by the basic sampling function f (t) and the waveform indicated by the control sampling function c ₀ (t) is a variable parameter. It varies greatly depending on the numerical value of α. In the present embodiment, when the variable parameter α is changed in the order of −1.5, −0.25, and 1.5, the region of −2 ≦ t ≦ −1 and the region of 1 ≦ t ≦ 2 Then, it was confirmed that the amplitude of the wavelength of the sampling function s _N (t) gradually increased and the polarity of the waveform was inverted. On the other hand, in the region of −1 ≦ t ≦ 0 and the region of 0 ≦ t ≦ 1, it was confirmed that the waveform amplitude of the sampling function s _N (t) gradually decreased and the waveform polarity was inverted.

Next, the violin song “Zigeunerweisen” recorded on the CD as a test song was reproduced in the variable characteristic signal converter 1 for about 23 seconds. At this time, the signal conversion processing unit 2 sets the variable parameter α to −0.25, −1.5, and 1.5, respectively, and interpolates the discrete data input during about 23 seconds. Then, when comparing the frequency characteristics of the analog signal interpolated with each sampling function s _N (t) at this time, the result as shown in FIG. 9 was obtained. FIG. 9 is a diagram illustrating a relationship between a plurality of variable parameters α and frequency characteristics of an analog signal obtained by interpolation processing.

As shown in FIG. 9, in the interpolation process using each sampling function s _N (t) in which the numerical value of the variable parameter α is changed, even if the numerical value of the variable parameter α is changed, all signals are generated in a high frequency range of 20 kHz or higher. The level increased, and it was confirmed that the high-frequency component could be reproduced compared with the conventional Shannon sampling function. Further, when the variable parameter α is set to 1.5, the signal level is decreased at less than about 26 kHz, but the signal level is increased except in the vicinity of 44.1 kHz in the high frequency range of about 26 kHz or more. It was confirmed that the high frequency component can be reproduced as compared with the case of setting to 0.25 and -1.5.

  On the other hand, when the variable parameter α is set to −1.5, the signal level suddenly decreases around 26 kHz, but the signal level increases as a whole at less than about 26 kHz, and about 26 kHz except for around 44.1 kHz. It was confirmed that the signal level increased even in the higher region, and the high frequency component could be reproduced with a different signal level compared to the case where the variable parameter α was set to -0.25 and -1.5.

  Furthermore, when the variable parameter α is set to −0.25, the signal level generally increases except for the vicinity of 44.1 kHz, compared to the case where the variable parameter α is set to 1.5 and −1.5. It was confirmed that the high frequency components can be reproduced with different signal levels.

Next, a sound having a reproduction frequency fixed at 10 kHz and 20 kHz as a test song was reproduced by the variable characteristic signal converter 1. At this time, the signal conversion processing unit 2 sequentially switches the numerical value of the variable parameter α from −5 to 5 and interpolates the discrete data sequentially input from the input unit 6. Then, when the signal levels of the analog signals obtained by performing the interpolation processing with the sampling functions s _N (t) having different variable parameters α at this time were compared, the results as shown in FIG. 10 were obtained. FIG. 10 is a schematic diagram showing signal levels when the value of the variable parameter is changed with the reproduction frequency fixed.

  As shown in FIG. 10, at the reproduction frequency of 10 kHz, as the variable parameter α is increased, the signal level gradually decreases, and when the variable parameter α is between 2 and 3, the signal level rapidly increases. After that, it was confirmed that the signal level increased rapidly again. On the other hand, at a reproduction frequency of 20 kHz, as the variable parameter α is increased, the signal level gradually decreases, and when the variable parameter α is near 4, the signal level rapidly decreases, and then the signal level again. It was confirmed that it rose rapidly. As described above, it was confirmed that the signal conversion processing unit 2 can reproduce at the same reproduction frequency with different signal levels by changing the variable parameter α.

(4) Operation and Effect As described above, in the signal conversion processing unit 2 of the present embodiment, the basic sampling function f (t) is stored in the basic term calculation unit 16 and extracted by the discrete data extraction unit 15. The value of the basic sampling function f (t) is calculated for each discrete data d1, d2, d3, d4 with the distance to the interpolation position t0 as t, and is associated with each of the discrete data d1, d2, d3, d4. The basic interpolation value ya at the interpolation position t0 is calculated by performing a convolution operation on the value of the basic sampling function f (t).

In the signal conversion processing unit 2, the control sampling function c ₀ (t) is stored in the control term calculation unit 17, and each discrete data d 1, d 2, d 3, d 4 extracted by the discrete data extraction unit 15 is stored. to calculate the value of the control sampling function c ₀ (t) the distance of the interpolation position t0 as t, the discrete data d1, d2, d3, control sampling function is made to correspond to each of d4 c ₀ of (t) After the value is convolved, the control interpolation value yb at the interpolation position t0 is obtained by multiplying the convolution calculation result of the control sampling function c ₀ (t) by a variable parameter α set to an arbitrary numerical value by the user. I'm calculating.

In the signal conversion processing unit 2, the control sampling function c is calculated by linearly adding the basic interpolation value ya calculated in this way and the control interpolation value yb to calculate the interpolation value y between discrete data. _The interpolation value y reflecting the numerical value of the variable parameter α multiplied by the value of ₀ (t) is calculated.

Therefore, the signal conversion processing unit 2 can adjust the interpolation value y obtained by performing the interpolation process with the sampling function s _N (t) according to the variable parameter α by changing the numerical value of the variable parameter α. That is, the user can change the variable parameter α as appropriate according to various conditions such as the music playback environment, the sound source, the tune, and the like, thereby reproducing the high-quality music having the desired sound quality in which the frequency characteristics of the analog signal are adjusted. It becomes possible.

Further, the signal conversion processing unit 2 uses a basic sampling function f (t) and a control sampling function c ₀ (t) of a finite stage that can be differentiated only once over the entire area as the sampling function s _N (t). Since the control sampling function c ₀ (t) is multiplied by the variable parameter α, the amount of computation required for the interpolation processing between discrete data is greatly reduced as compared with the conventional Shannon sampling function. Can do. In addition, the truncation error that occurs when the Shannon sampling function is used does not occur, and the occurrence of aliasing distortion can be prevented.

In the present embodiment, it is possible to converge the waveform value of the sampling function s _N (t) to 0, particularly in a range that is the same as or narrower than two sample positions before and after the interpolation position t0. Therefore, when data interpolation or the like is performed using this sampling function s _N (t), it is only necessary to use a total of four discrete data items before and after the position of interest, and when the Shannon sampling function is used. Compared with, the processing burden can be remarkably reduced.

Further, in the present embodiment, the sampling function s _N (t) is separated into the basic sampling function f (t) and the control sampling function c ₀ (t) and stored separately, and each of them is discrete data. Is subjected to a convolution operation, the convolution operation result of the control sampling function c ₀ (t) and the discrete data is multiplied by the variable parameter α, and this is multiplied by the basic sampling function s _N (t) and the discrete data. since the convolution calculation result added to the so as to obtain the output signal of the control sampling function c ₀ (t) may be one holding, is as simple as possible the control sampling function c ₀ (t) And variable control of the control sampling function c ₀ (t) can be easily performed.

(5) Other Embodiments In the above-described embodiment, the basic term calculation unit 16 and the control term calculation unit 17 sequentially calculate a plurality of interpolated values between discrete data one by one. However, the present invention is not limited to this, and a plurality of interpolated values between discrete data may be calculated collectively.

  FIG. 11 is a diagram illustrating a modification example in which a plurality of interpolation values are calculated at once. In FIG. 11, the same reference numerals are given to configurations common to FIG. 5. As shown in FIG. 11, the signal conversion processing unit 30 includes a discrete data extraction unit 15 and a conversion function matrix calculation unit 31. In the conversion function matrix calculation unit 31, the values Y (t1), Y (t2), Y (t3), and Y (t4) of the discrete data d1, d2, d3, and d4 are multiplied by the conversion matrix A (described later). By doing this, a plurality of interpolated values yk-2 (1), yk-2 (2),..., Yk-2 (n) between discrete data are calculated sequentially or collectively.

  In this modified example, as shown in FIG. 12, the past second discrete data d2 and the past third discrete data d3 are divided into 1 to n among four consecutive discrete data d1, d2, d3, d4. Then, the values are divided by a predetermined number of sections (in this case, n + 1), and interpolation values yk-2 (1), yk-2 (2),..., Yk-2 (n) at each position are calculated.

  Here, the transformation matrix A is expressed by the following equation.

This transformation matrix A calculates a sampling function s _N (t) using four discrete data d1, d2, d3, d4, and n interpolation values yk−2 (1) between the discrete data d2 and d3. ), Yk-2 (2),..., Yk-2 (n), the sampling function s _N (t) is used as an element, and it has n rows and 4 columns. The transformation matrix A is multiplied by a one-column matrix X whose elements are the values Y (t1), Y (t2), Y (t3), and Y (t4) of the discrete data d1, d2, d3, and d4. Thus, correction values yk-2 (1), yk-2 (2),..., Yk-2 (n) can be obtained. That is, the correction values yk-2 (1), yk-2 (2),..., Yk-2 (n) can be obtained by the following equation.

  Here, the transformation matrix A is the sum of the basic term matrix B of the following equation and the control term matrix C of the following equation multiplied by the variable parameter α, and is represented by A = B + αC.

  The basic term matrix B has a basic sampling function f (t) as an element, and the control term matrix C has a control sampling function c (t) as an element (t indicates a distance between an interpolation point and a sample position). Therefore, the interpolation values yk-2 (1), yk-2 (2),..., Yk-2 (n) are expressed by the following equations.

  In practice, as shown in FIG. 13, the conversion function matrix calculation unit 31 executes a basic term matrix calculation circuit 32 that executes calculation of the basic term matrix B and the matrix X, and calculation of the control term matrix C and the matrix X. .., 18an for multiplying the calculation result of the control term matrix calculation circuit 33, the calculation result of the control term matrix calculation circuit 33 by the variable parameter α, and the calculation result from the basic term matrix calculation circuit 32 .., 18an are linearly added to the calculation results from the coefficient multipliers 18a1, 18a2,..., 18an, and a plurality of linear adders 19a1, 19a2,.

  The basic term matrix calculation circuit 32 calculates a basic term matrix B as a basic sampling function in advance according to the number of sections between discrete data, and generates a basic term matrix B in which the calculated values are tabulated. It is stored in a predetermined storage means. Further, upon receiving the discrete data d1, d2, d3, d4 from the discrete data extraction unit 15, the basic term matrix calculation circuit 32 converts the discrete data into the basic term matrix B as a table value stored in advance in a predetermined storage means. The values Y (t1), Y (t2), Y (t3), and Y (t4) of d1, d2, d3, and d4 are multiplied as a matrix X in one column. Then, the fundamental term matrix calculation circuit 32 sends the value of each row of the matrix obtained as a result to the corresponding linear adders 19a1, 19a2,. That is, the basic term matrix calculation circuit 32 calculates (f1 (n + 1) · Y (t1)) + (f2 (1) · Y (t2)) + (f3 (n) in the first row of the matrix obtained as the calculation result. -1) .Y (t3)) + (f4 (2n-1) .Y (t4)) is sent to the linear adder 19a1, and (f1 (n + 2) .Y (t1)) + in the next second row (F2 (2) · Y (t2)) + (f3 (n−2) · Y (t3)) + (f4 (2n−2) · Y (t4)) is sent to the next linear adder 19a2, Thereafter, the values from the third line to the nth line are sent to different linear adders 19a3,.

  On the other hand, the control term matrix operation circuit 33 calculates in advance a control term matrix C as a control sampling function according to the number of sections between discrete data, and a control term matrix in which the operation values obtained thereby are tabulated. C is stored in a predetermined storage means. Further, when receiving the discrete data d1, d2, d3, d4 from the discrete data extraction unit 15, the control term matrix arithmetic circuit 33 stores the discrete data in the control term matrix C as a table value stored in advance in a predetermined storage unit. The values Y (t1), Y (t2), Y (t3), and Y (t4) of d1, d2, d3, and d4 are multiplied as a matrix X in one column. Then, the control term matrix operation circuit 33 sends the values of the respective rows of the matrix obtained as a result thereof to the corresponding coefficient multipliers 18a1, 18a2,. In other words, the control term matrix operation circuit 33 obtains (c1 (n + 1) · Y (t1)) + (c2 (1) · Y (t2)) + (c3 (n) in the first row of the matrix obtained as the operation result. −1) · Y (t3)) + (c4 (2n−1) · Y (t4)) is sent to the coefficient multiplier 18a1, and (c1 (n + 2) · Y (t1)) + in the next second row (C2 (2) · Y (t2)) + (c3 (n−2) · Y (t3)) + (c4 (2n−2) · Y (t4)) is sent to the next coefficient multiplication unit 18a2. Thereafter, the values from the third line to the nth line are sent to different coefficient multipliers 18a3,.

  The coefficient multipliers 18a1, 18a2,..., 18an multiply the variable parameter α set by the user in the parameter setting unit 7 by the value of each row of the matrix calculated by the control term matrix calculation circuit 33. Are sent to the corresponding linear adders 19a1, 19a2,. Each of the linear addition units 19a1, 19a2,..., 19an performs the same calculation of the calculation results received from the basic term matrix calculation circuit 32 and the calculation results received from the coefficient multiplication units 18a1, 18a2,. , And outputs the interpolation values yk-2 (1), yk-2 (2),..., Yk-2 (n).

  As described above, the signal conversion processing unit 30 stores the basic term matrix B in the basic term matrix calculation circuit 32 and controls the control term matrix calculation circuit 33 in addition to the same effects as the above-described embodiment. The term matrix C is stored, and the values Y (t1), Y (t2), Y (t3), and Y (t4) of the discrete data d1, d2, d3, and d4 are multiplied as a one-row matrix X. Thus, interpolation values yk-2 (1) and yk-2 (2) of interpolation positions 1 to n between predetermined discrete data d2 and d3 among four continuous discrete data d1, d2, d3, and d4. ,..., Yk-2 (n) can be easily calculated collectively.

  In the above-described embodiment, the case where the basic term matrix B and the control term matrix C that are applied to the discrete data sequence in which the number of sections between the discrete data of interest is n + 1 and a constant number is used has been described. However, the present invention is not limited to this, and a basic term matrix and a control term matrix applicable to a plurality of discrete data strings having different numbers of sections between discrete data of interest may be used. That is, in this case, since the conversion function matrix calculation unit is applied to a plurality of discrete data strings having different numbers of divisions, the basic term matrix B and the control term matrix C are calculated in advance with the number of divisions of the least common multiple of the number of divisions. Then, according to the number of divisions set at the start of discrete data input, an operation value corresponding to the number of divisions is selected from the basic term matrix B and the control term matrix C as a table value, A convolution operation is performed on the selected table value and discrete data. As a result, the conversion function matrix calculation unit can correspond to a plurality of discrete data strings having different numbers of sections only by storing only one basic term matrix B and control term matrix C in advance. Therefore, the processing capacity of the entire apparatus can be reduced. For example, considering the number of sections 2 (when one interpolation value is generated between discrete data) and the number of sections 3 (when two interpolation values are generated between discrete data), the minimum number of these sections Since the common multiple is 6, if the table value (function value) corresponding to the number of divisions 6 is held, it is possible to perform convolution operations of both the number of divisions 2 and 3 using this one table. become.

The present invention is not limited to the above-described embodiments, and various modifications can be made within the scope of the gist of the present invention. For example, although the sampling function s _N (t) is a finite function that can be differentiated only once over the entire area, the number of differentiable times may be set to 2 or more.

In the above-described embodiment, the analog signal is generated by performing the interpolation process using the sampling function s _N (t). However, the present invention is not limited to this, and the sampling is performed. It is also possible to simply perform oversampling by performing an interpolation process using the function s _N (t), and then generate an analog credit with an analog-digital converter.

Furthermore, in the above-described embodiment, the sampling function s _N (t) converges to 0 when t = ± 2, but the present invention is not limited to this, and converges to 0 when t = ± 3 or more. You may do it. For example, when t = ± 3 and converges to 0, the discrete data extraction unit 15 extracts the previous six discrete data, and the function processing unit 14 performs sampling functions on these six discrete data. The value of s _N (t) is calculated.

Further, in the above-described embodiment, the basic sampling function f (t) is stored in the basic term calculation unit 16, and separately, the control sampling function c ₀ (t) is stored in the control term calculation unit 17. Then, the basic interpolation value ya and the control interpolation value yb are calculated by performing a convolution operation on the discrete data d1, d2, d3, and d4 for each of the basic sampling function f (t) and the control sampling function c ₀ (t). After that, the case where the interpolation value y is calculated by linear sum addition of the basic interpolation value ya and the control interpolation value yb has been described, but the present invention is not limited to this. For example, the function processing unit 14 adds the result obtained by multiplying the basic sampling function f (t) by the variable parameter α and the control sampling function c ₀ (t) and performs a linear combination thereof, thereby changing the variable. A specific sampling function s _N (t) corresponding to the parameter α is specified, and using this sampling function s _N (t), a convolution operation is performed on the discrete data d1, d2, d3, d4, and an interpolation value is obtained. You may make it calculate y directly. As a result, the function processing unit 14 reduces the number of multiplications compared to the case where the convolution calculation is separately performed using each of the basic sampling function f (t) and the control sampling function c ₀ (t), and the calculation time is reduced. This is suitable for the case of using an arithmetic device that can reduce the number of multipliers and the number of multipliers and has a low processing speed.

It is a figure which shows the waveform shape of the basic sampling function and control sampling function which comprise the sampling function used for the interpolation process of this invention. It is a figure which shows the structure of the variable characteristic type signal converter of one Embodiment. It is a figure which shows the structure of the signal conversion process part in a programmable signal processor. It is a figure which shows the positional relationship of four continuous discrete data and the attention point which is an interpolation position. It is a figure which shows the specific structure of a signal conversion process part. It is a figure which shows the outline of the interpolation process with respect to the predetermined | prescribed interpolation position in a fundamental term calculating part. It is a figure which shows the outline of the interpolation process with respect to the predetermined | prescribed interpolation position in a control term calculating part. It is a figure which shows the waveform of a sampling function. It is a figure which shows the relationship between the several variable parameter (alpha) and the frequency characteristic of the analog signal obtained by the interpolation process. It is the schematic which shows the signal level when fixing the reproduction frequency and changing the value of the variable parameter. It is a figure which shows the modification which calculates a several interpolation value collectively. It is the schematic which shows the positional relationship of four discrete data and a view point, and an interpolation position. It is a block diagram which shows the detailed structure of the signal conversion process part by other embodiment. It is the schematic which shows the waveform of the sampling function of Shannon in the past.

Explanation of symbols

2 Signal Conversion Processing Unit 3 Programmable Signal Processing Processor 10 Selector 14 Function Processing Unit 15 Discrete Data Extraction Unit 16 Basic Term Calculation Unit 17 Control Term Calculation Unit 18 Coefficient Multiplying Unit 19 Linear Addition Unit 32 Basic Term Matrix Operation Circuit 33 Control Term Matrix Operation circuit

Claims (23)

By performing a convolution operation on a plurality of discrete data using a sampling function composed of a linear combination of a basic sampling function and a control sampling function each represented by a finite piecewise piecewise polynomial, includes a function processing unit for calculating the interpolated value,
When the sample position of the discrete data is t,
In the basic sampling function, t ranges from −1 to +1 within a finite range, becomes 1 when t = 0, converges to 0 toward t = ± 1,
It said control sampling function, variable characteristic type signal conversion device according to claim Rukoto a 0 at each sampling position t included in the scope of finite support. In claim 1,
A discrete data extraction unit for extracting a predetermined number of discrete data in which the input order is continuous from the plurality of discrete data sequentially input;
The function processing unit calculates an interpolation value by using a predetermined number of the discrete data extracted by the discrete data extraction unit. In claim 1 or 2,
When the basic sampling function is f (t), the control sampling function is C (t), and a parameter that can be arbitrarily set by the user is α,
The function processing unit performs the linear combination,
f (t) + αC (t)
A variable characteristic type signal converter characterized in that it is operated by In claim 3,
The function processing unit
A basic term operation unit that performs a convolution operation on the plurality of discrete data using the basic sampling function;
A control term operation unit that performs a convolution operation on the plurality of discrete data using the control sampling function;
A coefficient multiplier for multiplying the calculation result by the control term calculator by the parameter;
A variable characteristic signal conversion device comprising: In claim 3,
The function processing unit specifies the sampling function obtained by linearly combining the basic sampling function and the control sampling function when a value of the parameter is specified, and uses the sampling function. The interpolated value is calculated by performing the convolution operation on the discrete data. In claim 4 or 5,
A variable characteristic signal conversion apparatus, further comprising: a parameter setting unit that arbitrarily sets a value of the parameter according to a user instruction. In claim 4 or 5,
A variable characteristic signal conversion apparatus, further comprising a selector for selecting one of a plurality of preset values of the parameter by a user operation. In any one of Claims 1-7,
The basic sampling function is a piecewise polynomial that can be differentiated only once in the interval [−1, 1] of the sample position of the discrete data, and the other interval is a function represented by 0 uniformly.
The control sampling function is a piecewise polynomial that can be differentiated only once in the interval [-2, 2] of the sample position of the discrete data, and is a function that is uniformly 0 in other intervals. A variable characteristic type signal converter. In any one of Claims 1-8,
The basic sampling function f (t) is

Represented by
When the control sampling function is C ₀ (t) = C _r (t) + C _r (−t), the C _r (t) is

A variable characteristic type signal converter characterized by the following. In any one of Claims 1-9,
A variable characteristic signal conversion device programmed to a programmable signal processor that forms an arithmetic configuration having a control form desired by a user based on program data designated by a user. In any one of Claims 1-10,
In each of the basic sampling function and the control sampling function, function values calculated in advance according to a predetermined number of sections between the discrete data are tabulated and held as table values,
A variable characteristic signal conversion apparatus characterized in that a convolution operation is performed on the discrete data using the table-like function values. In claim 11,
The variable characteristic signal conversion apparatus according to claim 1, wherein the table values corresponding to the number of divisions of the least common multiple of the number of divisions between the discrete data are calculated and held in advance. By performing a convolution operation on a plurality of discrete data using a sampling function composed of a linear combination of a basic sampling function and a control sampling function each represented by a finite piecewise piecewise polynomial, have a functional operation step of calculating the function processor interpolated values,
When the sample position of the discrete data is t,
The basic sampling function becomes 1 at t = 0 and converges to 0 toward t = ± 1 within a finite range.
It said control sampling function, variable characteristic type signal conversion method according to claim Rukoto a 0 at each sampling position t included in the scope of finite support. In claim 13,
A discrete data extraction step of extracting a predetermined number of discrete data in which the input order is continuous from among the plurality of discrete data sequentially input by a discrete data extraction unit;
In the function processing step, an interpolation value is calculated using a predetermined number of discrete data extracted in the discrete data extraction step. In claim 13 or 14,
When the basic sampling function is f (t), the control sampling function is C (t), and a parameter that can be arbitrarily set by the user is α,
In the function processing step, the linear combination is
f (t) + αC (t)
A variable characteristic type signal conversion method, characterized in that the calculation is performed by the method. In claim 15,
The function processing step includes:
A basic term operation step in which a basic term operation unit performs a convolution operation on the plurality of discrete data using the basic sampling function;
A control term computation step in which a control term computation unit performs a convolution computation on the plurality of discrete data using the control sampling function;
A coefficient multiplication step in which a coefficient multiplication unit performs an operation of multiplying the calculation result in the control term calculation step by the parameter;
A variable characteristic type signal conversion method comprising: In claim 15,
In the function processing step, when the value of the parameter is specified, the sampling function obtained by linearly combining the basic sampling function and the control sampling function is specified, and the sampling function is used. A variable characteristic signal conversion method characterized in that the interpolated value is calculated by performing the convolution operation on the discrete data. In claim 16 or 17,
A variable characteristic signal conversion method, further comprising: a parameter setting step of arbitrarily setting the value of the parameter by a parameter setting unit in accordance with a user instruction. In claim 16 or 17,
A variable characteristic signal conversion method, further comprising a parameter selection step of selecting one of a plurality of preset values of the parameter in accordance with a user operation instruction. In any one of Claims 13-19,
The basic sampling function is a piecewise polynomial that can be differentiated only once in the interval [−1, 1] of the sample position of the discrete data, and the other interval is a function represented by 0 uniformly.
The control sampling function is a piecewise polynomial that can be differentiated only once in the interval [-2, 2] of the sample position of the discrete data, and is a function that is uniformly 0 in other intervals. A variable characteristic type signal conversion method. In any one of Claims 13-20,
The basic sampling function f (t) is

Represented by
When the control sampling function is C ₀ (t) = C _r (t) + C _r (−t), the C _r (t) is

The variable characteristic type signal conversion method characterized by the above-mentioned. In any one of Claims 13-21,
In each of the basic sampling function and the control sampling function, function values calculated in advance according to a predetermined number of sections between the discrete data are tabulated and held in the function processing unit as table values. ,
In the functionalization processing step, a convolution operation is performed on the discrete data using the tabulated function values. In claim 22,
A variable characteristic type characterized in that, for a plurality of different division numbers between the discrete data, the table value corresponding to the division number of the least common multiple of the division numbers is calculated in advance and held in the function processing unit Signal conversion method.

Images