JP3223280B2 - Waveform data interpolation device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a waveform data interpolating apparatus for a musical tone in an electronic musical instrument or the like in which a performance is performed based on a sampled tone signal.

[0002]

2. Description of the Related Art In an electronic musical instrument, when a tone signal having a different pitch is created from one tone waveform data having a clear pitch (pitch), effects such as vibrato and detune are added to the tone. If so,
Usually, a process called pitch interpolation is performed to obtain a new pitch.

FIG. 6 is a diagram showing pitch interpolation according to a conventional linear interpolation method. An interpolation value is obtained on a straight line connecting peak values at sampling points.

[0004]

However, in the linear interpolation method, since the interpolation accuracy is not good as is apparent from FIG. 6, the tone waveform is distorted as shown in FIG. When there are many frequency components close to / 2 or when the oversampling ratio at the time of output is high, there is a problem that this tendency becomes remarkable.

An object of the present invention is to realize a waveform data interpolating device which has less waveform distortion and excellent sound quality.

[0006]

SUMMARY OF THE INVENTION The present invention is based on a waveform data interpolation apparatus which calculates a peak value at an arbitrary waveform position by interpolating a plurality of waveform sample values with respect to a waveform sample value of waveform data. And

Then, a first quadratic curve passing through the waveform sample values of the waveform data at two sampling points before and one sampling point after the designated waveform position, The second 2 that passes through the waveform sample values of the waveform data at the preceding one sampling point and the two following sampling points
Interpolating means for calculating a third curve passing through an arithmetic mean value with the next curve and calculating the value of the third curve at the waveform position as a peak value at the waveform position.

[0008]

A third curve passing through an arithmetic mean value of two quadratic curves shifted forward and backward with respect to a designated waveform position is calculated based on a quadratic curve interpolation which is easy to calculate. The peak value at the waveform position is calculated as the value of the third curve at the position.

Accordingly, the result is the same as approximately increasing the degree of interpolation while minimizing the increase in the amount of calculation, and the interpolation accuracy is improved.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment in which the present invention is applied to an electronic keyboard instrument will be described below in detail with reference to the drawings. Configuration FIG. 1 is a block diagram showing the overall configuration of an embodiment according to the present invention.

In FIG. 1, a CPU (central processing unit) 101 operates by executing a program stored in a program memory 102, and a switch group 1
04 and the keyboard 103 are scanned to capture the operation states of the switches and keys, and the waveform ROM 10 is read from the data memory 105.
In addition to reading the address 7 and the envelope data, it also sends the tone generation control data to the tone generator 106 to control the tone generation operation of the tone signal based on the waveform data read from the waveform memory 107.

A tone signal generated by the tone generator 106
Is converted into an analog signal by the D / A converter 108.
Music is emitted by the audio system 109.
You. Principle of pitch interpolation of this embodiment Next, the operation principle of pitch interpolation according to the present invention will be described with reference to FIG.
Will be described.

In FIG. 2, sampling points x = −1,
Peak values L ₋₁ , L ₀ , L at ₀ , ₁ , 2
_1, with respect to L _2, the sampling point x = 0 (peak value at this time L ₀₎ Consider the coordinate with the origin of. Then, a quadratic curve Y ₁ (x) passing through the peak values L ₀ , L ₁ , L ₂
And a quadratic curve Y ₂ (x) passing through the peak values L ₋₁ , L ₀ , and L ₁
And a curve Y passing through the average of these two curves
An interpolated value between the peak values is obtained on _S (x).

The pitch interpolation operation will be described below with reference to FIGS. FIG. 3 is a configuration diagram showing the internal configuration of the sound source unit 106 of FIG. 3, an integer address calculation block 301 receives a waveform start address from the CPU 101 and a waveform loop address for repeatedly reading waveform data from the waveform ROM 107. Further, the minority address calculation block 302 accumulates the increment of the read address of the ROM 107 given from the CPU 101, and when a carry occurs, sends the carry to the integer address calculation block 301 and increments the integer address. Let it. Then, the minority address calculated in the minority address calculation block 302 is input to the product-sum block 305 as data for performing pitch interpolation.

Next, the waveform ROM data latch block 3
Reference numeral 04 denotes a circuit for fetching waveform data from the waveform ROM 107 based on the address designated by the integer address calculation block 301 to calculate a difference value, accumulate the difference value, and the like, and use the calculation results for pitch interpolation. The data is sent to the product-sum block 305 as data.

The sum-of-products block 305 performs pitch interpolation on the tone signal and multiplication of the envelope data generated by the envelope generation block 303, as described later, and outputs a time-division output DCO to the accumulation block 306. The accumulation block 306 converts the time-division output DCO into RI based on the panning data output from the CPU 101.
The GHT (channel) accumulation register 307 and the LEFT (channel) accumulation register 308 are allocated to perform accumulation.

The accumulated time-division output DCO is converted into an analog signal by the D / A converter 108 and reproduced by the audio system 109 as a musical tone. Next, in FIG. 2, a quadratic curve Y ₁ (x) passing through the peak values L ₀ , L ₁ , and L _2.
And a quadratic curve Y ₂ (x) passing through the peak values L ₋₁ , L ₀ , and L ₁
Curve Y _s (x) through each average value of is represented by the difference value and the peak value L ₀ of the waveform data in the following manner.

First, a quadratic curve Y ₁ (x) is expressed as

[0019]

## EQU1 ## If Y ₁ (x) = ax ² + bx is obtained, FIG.

[0020]

## EQU2 ## When x = 1, Y ₁ (x) = d ₁ = a + b

[0021]

## EQU3 ## When x = 2, Y ₁ (x) = d ₂ = 4a + 2b. From Equation 3−Equation 2 × 2, d ₂ −2d ₁ = 2a∴a = １ ／ (d ₂ −2d
₁ ) In addition, from Equation 2 × 4 Equation 3, 4d ₁ −d ₂ = 2b∴b =＝ (4d ₁ −d
₂ ) Therefore, from equation ( ₁ ), Y ₁ (x) = １ ／ (d ₂ −2d ₁ ) × ² + ／ (4
d ₁ −d ₂ ) x.

Next, a quadratic curve Y ₂ (x) is expressed as

[0023]

## EQU4 ## If Y ₂ (x) = ax ² + bx,

[0024]

## EQU5 ## When x = 1, Y ₂ (x) = d ₁ = a + b

[0025]

[6] When _{x = -1, Y 2 (x} ) = - a d ₀ = a-b. From Expression 5 and Expression 6, d ₁ −d ₀ = 2aｄa = ２ (d ₁ −d ₀ ). In addition, from Expression 5 to Expression 6, d ₁ + d ₀ = 2bｂb = １ ／ (d ₁ + d ₂ ). Therefore, from equation (4), Y ₂ (x) = １ ／ (d ₁ −d ₀ ) × ² + ／ (d ₁
+ D ₀ ) x.

The above is a coordinate system having the peak value L ₀ as the origin, and if the coordinate system is restored by adding the peak value L ₀ , it is expressed as follows.

[0027]

## EQU7 ## Y ₁ (x) = １ ／ (d ₂ −2d ₁ ) x ² +1
_{/ 2 (4d 1 -d 2)} x + L 0

[0028]

Y ₂ (x) = １ ／ (d ₁ −d ₀ ) x ² + 1 /
2 (d ₁ + d ₀ ) x + L ₀ Next, the average value Y _S (x) of Y ₁ (x) and Y ₂ (x) is
It is obtained as follows using the equations (7) and (8).

[0029]

[Expression 9] Y _S (x) = 1 / 2１ ／ Y ₁ (x) + Y
₂ (x)｝ = (1/4) (d ₂ −d ₁ −d ₀ x ² ) +
_{(1/4) (- d 2 +} 5d 1 + d 0) x + L 0 = (1 /
4) x ｛d ₂ (x−1) + d ₁ (−x + 5) + d ₀ (−
x + 1)｝ + L ₀ = (１ ／) x ｛−d ₂ (1-x) +
d ₁ (1-x) + d ₀ (1-x) + 4d ₁ ｝ + L ₀ = x
_{{1/4 (-d 2 + d 1} + d 0) (1-x) + d 1} +
L ₀ Here, when expressed by the peak value L _-1 ~L ₂ sampling points consecutively d ₀ to d _2, from FIG. 2

[0030]

## EQU10 ## d ₂ = L ₂ −L ₀

[0031]

## EQU11 ## d ₁ = L ₁ -L ₀

[0032]

The [number 12] d ₀ = L ₀ -L _-1. Therefore, from Equation 9,

[0033]

(13) (−d ₂ + d ₁ + d ₀ ) = − (L ₂ −L ₀ )
+ (L ₁ −L ₀ ) + (L ₀ −L ₋₁ ) = − (L ₂ −L ₁ )
+ (L ₀ −L ₋₁ ).

Next, the relationship between the peak values L _{-1 to} L _{2 at} successive sampling points and the difference values δ _{-1 to} δ ₊₁ is shown in FIG.

[0035]

Δ ₊₁ = L ₂ −L ₁

[0036]

Δ ₀ = L ₁ −L ₀ (= d ₁ )

[0037]

Δ ₋₁ = L ₀ −L ₋₁

Therefore, using equations (14) and (16),
Equation 13 is

[0039]

(17) (−d ₂ + d ₁ + d ₀ ) = − δ ₊₁ + δ ₋₁

As a result, Y _s (x) of the equation (9) is as follows from the equations (17), (11) and (15).

[0041]

Y _S (x) = x ｛1/4 (−δ ₊₁ + δ ₋₁ )
(1−x) + δ ₀ ｝ + L _{0 A} quadratic curve Y ₁ (x) passing through the peak values L ₀ , L ₁ , and L ₂ represented by Expression 18, and the peak values L ₋₁ , L ₀ , and L ₁ The peak values on the curve Y _S (x) passing through the average value of the quadratic curve Y ₂ (x) passing through are calculated by the waveform ROM data latch block 304 and the product-sum block 305 in the tone generator 106 in FIG. Is done.

Next, the waveform ROM data latch block 3
04 and the configuration and operation of the product-sum block 305 will be described with reference to FIG. Waveform ROM data latch block 304 First, multiplexer 40 for data bus switching
5, 406 are both switched to the "1" side, and a difference value accumulation register L ₀ for restoring the peak value from the difference value.
401 and the data of the register δ ₀ 403 storing the current difference value calculated based on the waveform data read from the waveform ROM 107 as described above are added to the adder / subtractor 40.
7, and the result of the addition is written to the difference value accumulation register L ₀ 401 as a peak value at a new sampling point.

Next, a register δ _-1 404 storing the difference value at the immediately preceding (previous) sampling point and a register δ ₀ storing the difference value at the current sampling point.
403 and the register δ ₊₁ 402 storing the difference value at the next sampling point, δ ₋₁ ← δ ₀ , δ ₀ ← δ ₊₁
Data is updated as the waveform RO in the register [delta] ₊₁
A new difference value is written from M107.

Then, the inputs of the multiplexers 405 and 406 are both switched to the “2” side, and the adder / subtractor 407 subtracts the data of the next difference value register δ ₊₁ from the previous difference value register δ _-1 404. , Register 408
Is written to.

The above operation is repeated each time the integer address calculation block 301 in FIG. 3 updates the address of the waveform ROM 107 in FIG. 1, and the product sum block 305 is subjected to the peak value at the current sampling point. L ₀ , and the difference −δ between the respective difference values at the previous and next sampling points
₊₁ + δ _-1 will be sent. Note that the register L ₀₄
The contents of 01, δ ₀ 403, and δ _-1 404 are cleared by a reset signal when new waveform data is assigned. In FIG. 4, first, the multiplexers 415, 416,
The inputs 417 are both switched to "1". FIG.
The minority address x sent from the minority address calculation block 302 and written in the register x409 is
After all bits have been inverted at 11, the incrementer 412
, The value is increased by 1 LSB, and the data value is further multiplied by ４ by the right 2-bit shifter 413.

In this manner, a data value 1/4 (1-x) is obtained in the range of 0.ltoreq.x <1 for the small number address x, and stored in the register 414. Next, since the inputs of the multiplexers 415 and 416 are both switched to the “1” side, the output of the multiplier 418 outputs the multiplication result of the register 408 and the register 414, that is, １ ／ (1-x) (− δ ₊₁ + δ _-1 ) is obtained. Further, since the input of the multiplexer 417 is switched to the “1” side, the output of the adder 419 has the result of addition of the result of the multiplication and the value of the register δ ₀ 403, that is, （(1-x) _{(-δ +1 + δ -1) +} δ 0 is obtained, the addition result is stored in the register 420.

Next, the multiplexers 415, 416, 4
17 are both switched to the "2" side. As a result, the multiplication result of the register 420 and the register x409, that is, x ｛(1−x) (− δ) is output to the output of the multiplier 418.
₊₁ + δ ₋₁ ) + δ ₀ }, and the addition result of the multiplication result and the register L ₀ 401, that is, x ｛1/4 (1-x) (− δ) is added to the output of the adder 419. ₊₁ + δ ₋₁ ) + δ ₀ ｝ + L ₀ is obtained and stored in the register 421.

Thereafter, the multiplexers 415, 416,
As a result of the inputs 417 being both switched to the “3” side,
A register 421 and a register 41 are output to the output of the multiplication 418.
The result of the multiplication with the envelope value ENV from the envelope generation block 303 stored in 0, that is, ENV × [x ｛(1−x) (− δ _{+ 1} + δ− ₁ ) + δ
₀ ｝ + L ₀ ] is obtained. The obtained value is added to an adder 419.
Then, the value "0" is added, stored in the register 422, and output to the accumulation block 306 in FIG.

As described above, in the waveform ROM data latch block 304 and the product-sum block 305, the quadratic curve Y ₁ (x) passing through the peak values L ₀ , L ₁ , and L ₂ shown in the above equation (18) ) And a peak value on a curve Y _S (x) passing through each average value of the quadratic curve Y ₂ (x) passing through the peak values L ₋₁ , L ₀ , and L ₁ , and further, Is multiplied by the envelope value.

[0050]

According to the present invention, based on the quadratic curve interpolation which is easy to calculate, the third waveform passing through the arithmetic mean value of the two quadratic curves shifted forward and backward with respect to the designated waveform position. Is calculated, and the peak value at the waveform position is calculated as the value of the third curve at the waveform position,
While suppressing an increase in the amount of calculation as much as possible, the same effect as approximately increasing the order of interpolation can be obtained, and it is possible to improve interpolation accuracy and reduce waveform distortion.

[Brief description of the drawings]

FIG. 1 is an overall configuration diagram of an embodiment of the present invention.

FIG. 2 is an explanatory diagram of interpolation in the present embodiment.

FIG. 3 is a configuration diagram of a sound source unit 106.

FIG. 4 is a configuration diagram of a waveform ROM data latch block.

FIG. 5 is a waveform diagram of a signal obtained by interpolation according to the embodiment.

FIG. 6 is an explanatory diagram of conventional linear interpolation.

FIG. 7 is a waveform diagram of a signal obtained by conventional linear interpolation.

[Explanation of symbols]

101 CPU 102 Program Memory 103 Keyboard 104 Switch Group 105 Data Memory 106 Sound Source 107 Waveform ROM 108 D / A Converter 109 Audio System 301 Integer Address Calculation Block 302 Decimal Address Calculation Block 303 Envelope Generation Block 304 Waveform ROM Data Latch Block 305 Product Sum block 306 Accumulation block 307 RIGHT accumulation register 308 LEFT accumulation register 401 Difference value accumulation register 402, 403, 404, 408, 409, 410, 4
14, 420, 421, 422 Register 405, 406, 415, 416, 417 Multiplexer 407 Adder / subtractor 411 Inverter 412, 419 Adder 413 Right 2-bit shifter 418 Multiplier

Claims (1)

(57) [Claims] 1. A waveform sample value of waveform data
In a waveform data interpolating apparatus for calculating a peak value at an arbitrary waveform position by interpolating a plurality of the waveform sample values, sampling of two points before and one point behind the specified waveform position with respect to a designated waveform position A first quadratic curve passing through the waveform sample value of the waveform data at a point, and a second quadratic curve passing through the waveform sample value of the waveform data at one sampling point before and two sampling points after the waveform position. Waveform data interpolating apparatus, further comprising: an interpolating means for calculating a third curve passing through the arithmetic mean value of the above, and calculating a value of the third curve at the waveform position as a peak value at the waveform position.