JPS646479B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to an envelope control device suitable for multiplying a sound source waveform and an envelope waveform, such as in an electronic musical instrument, or for performing multiplication in image processing.

Conventional structure and its problems Conventionally, in the field of electronic musical instruments, when creating a musical sound waveform with an envelope by multiplying a sound source waveform and an envelope waveform, a digital multiplier of 12 bits x 12 bits or more has been used as a digital multiplier. If one with a width was not used, there was a problem that when the envelope was attenuated, noise due to quantization errors would be generated and be audible. Furthermore, the DA converter required a 14-bit to 16-bit one, and the envelope waveform also required a resolution of 14 to 16 bits. JP-A-54-1609 and JP-A-57-181593 are known as methods that display a binary floating point number instead of using a 16-bit linear code as an envelope waveform. However, for multiplication with the sound source waveform, analog multipliers or linear parallel multipliers such as 16 bits x 16 bits were used as before.

OBJECTS OF THE INVENTION The present invention provides an envelope control device that can perform multiplication of a waveform and an envelope using a multiplier that is smaller than conventional ones.

Structure of the Invention The present invention generates envelope data E _i divided into two parts, a lower code P _i and an upper code Q _i , and combines the lower code P _i or a modified version thereof and waveform data W _i. Multiply and shift the multiplication result according to the upper code Q _i ,
It is characterized in that the multiplier is made smaller to W _i ×P _i or slightly larger than W _i ×E _i instead of W i ×E i.

DESCRIPTION OF EMBODIMENTS FIG. 1 is a block diagram showing the principle of the present invention. Reference numeral 1 denotes a waveform generator that generates waveform data W _i of a musical sound source, which samples one waveform of a musical sound and repeatedly generates quantized data. 3 is a waveform data register. 2 is an envelope generator, which generates envelope data E _i . The envelope data E _i has a structure that can be separated into lower P _i and upper Q _i . 4 is a register for the lower code P _i and 5 is a register for the upper code Q _i . The waveform data W _i and the lower code _i are applied to a multiplier 6 . At this time, "1" is added above the most significant bit of the lower code P _i . Therefore, when P _i is 8 bits, multiplier 6
is given 9-bit data. Multiplier 6
Waveform data W _i is input to the other input.
This data will consist of 10 bits. For the output of the multiplier 6, the most significant bits out of 10+9=19 bits are used. This product code
WP _i is input to shifter 8. The shifter shifts the bits of the input code and outputs it, and the number of bits to be shifted upwards or downwards is specified by the upper code Q _i of the envelope data.

Here, if the envelope data is represented by {q ₃ q ₂ q ₁ q ₀ p ₇ p ₆ p ₅ p ₄ p ₃ p ₂ p ₁ p ₀ }, {p ₇ ~ p ₀ } represents the mantissa part, and { q ₃ to q ₀ } represents an exponent part.

The size of E _i is It is expressed as a binary floating value, such as .

By adding “1” to the top of P _i , E _i becomes becomes. The value inside { } is expressed as P^ _i .

On the other hand, if W _i is 10 bits, W _i =−w ₉ 2 ⁹ + ₈ 〓 ⁱ⁼⁰ w _i 2 ⁱ . However, this uses two's complement representation of negative numbers.

W _i ×E^ _i is The WP _i term is calculated by a 10 bit x 9 bit multiplier 6.

The term corresponds to performing 16 shifts depending on the upper code Q _i , where Q _i = (0000) there is no shift, Q _i = (0001) there is a 1-bit shift to the upper order, and Q _i
= (0010), shift 2 bits to the upper position, Q _i =
(1111) shifts 15 bits upward. Therefore, if WP _i is 12 bits, the output of shifter 8 will have a width of 12+15=27 bits, but only the upper 16 bits will be taken out and input to DA converter 9. In this way, since WP _i is 12 bits, the S/N against quantization noise is approximately 72 dB, and the dynamic range is 96 dB, corresponding to 16 bits, making it possible to obtain sufficiently high quality sound. I can do it.

FIG. 2 is another example illustrating the principles of the invention.
The differences from Figure 1 will be explained. The digital shifter 8 is omitted, and instead, an attenuator 10 corresponding to an analog shifter is provided at the output of the DA converter 9, and the attenuation AT is switched by the upper code Q _i . The attenuation AT is It is expressed as K is a constant. This arrangement is more advantageous than the example shown in FIG. 1, since the DA converter 9 can be of 12-bit precision.

A in FIG. 3 is an example of P^ _i and Q _i . It shows the attenuation process of the envelope, the so-called release. Since P _i changes from 255 to 0, P^ _i
varies from 511 to 256. Figure 4A shows Q _i ,
This is the format of P _i and P^ _i .

FIG. 5 is another example illustrating the principle of the invention.
The waveform data W _i and the lower code P _i are applied to a multiplier 6, and the product output thereof is shifted by a shifter 8 according to the upper code Q _i . The waveform data W _i is also shifted by the shifter 18 . shifter 8 and 18
The outputs of are added to a subtractor 20 where their difference is taken. The output of the subtracter 20 is DA
input to the converter.

In FIG. 6, the order of the subtracter 20 in the shifters 8 and 9 of the example in FIG. 5 is reversed to reduce the number of shifters to one, and the bit width to be handled by the subtracter 20 is reduced.

In the examples in Figures 5 and 6, the envelope data
Assume that E _i is generated so that it becomes an increasing function in its attenuation process. Examples of values are shown in FIG. 3B. FIG. 4B shows the format of envelope data.

FIG. 7 is an example of an analog shifter. In FIG. 7, the resistors are a well-known R-2R type ladder network. A switch is provided in the 2R resistor example to select the flow of current into the + and - inputs of the operational amplifier 31. R _F is a feedback resistor that determines the gain, and 30 through which current flows into the -input is a decoder, which operates according to the upper code Q _i.
Select and switch one of the 16 switches to the - input side. There is no attenuation for switch _SO , -6 dB for switch S ₁ , and -12 dB for switch S ₂ . For switch S _i , it becomes -6×idB. the most attenuated
In S ₁₅ , Q _i =(1111). In this case, the switch may be fixed to the + input side so that the input is cut off.

In Figure 7, 16 different levels are generated, which is more than enough for normal sounds. Q _i may be 3 bits. Even if Q _i is set to 4 bits, only 10 to 12 bits may be used instead of using all 16 bits.

In the case of the code shown in FIG. 3B, the correspondence between the decoder and the switch shown in FIG. 7 may be reversed.

In the examples shown in Figures 1 and 2, as P _i and Q _i decrease until Q _i = (0000), and further decreases beyond (00000000), _{Q i}
= (1111) and BORROW occurs, but
In this case, Q _i = (1111) or BORROW is detected, Q^ _i = (0000) is supplied to the shifter instead of Q _i , and the upper additional bit of R _i is changed from “1” to “0”. ”, as P^ _i ,
For normal cash register Q _i = (0001) ~ (1110), P^ _i = 511 ~
256, and when Q _i = (0000), P^ _i = 511 to 0.

In each of the above examples, we have explained the case where the product of waveform data and envelope data is directly DA converted, but based on this principle, the present invention calculates the difference value between the two subsequent products, and shifts the difference value. It was designed to do so.

FIG. 8 shows an embodiment of the present invention in which a difference value with respect to i of W _i ×E _i is output.

The waveform data W _i and P^ _i obtained by adding "1" to the lower order data P _i are added to the multiplier 60 to obtain the product WP _i . Since the product WP _i is applied to the latch 40, the previous product WP _i-1 is held at time i. The upper code Q _i of the envelope data E _i is the latch 42,
The signal is supplied to a comparator 43... and an attenuator 10, which is an analog shifter. The latch 42 holds the previous upper code Q _i-1 and supplies it to the comparator 43 . The comparator 43 compares Q _i and Q _i-1 , and then outputs an output based on the conditions.

Q _i >Q _i-1 :a Q _i =Q _i-1 :b Q _i <Q _i-1 :c The output WP _i-1 of the latch 40 is supplied to the shifter 41 . The shifter 41 shifts WP _i-1 up or down by 1 bit, and can be configured by a combination of an AND gate and an OR gate. The signal entering input terminal A is shifted one bit lower. Input terminal B
The input signal is output as is. Since the signal input to the input terminal C is shifted upward by one bit, the input WP _i-1 is converted as follows.

a: Select A: 1/2WP _i-1 b: Select B: WP _i-1 c: Select C: 2WP _i-1 The output of the shifter 41 and the product are supplied to the subtracter 61, and the difference is Calculated. The differential output is as follows.

When Q _i > Q _i-1 ΔWP _i = 1/2WP _i-1 −WP _i Q _i = Q _i-1 When ΔWP _i = WP _i-1 −WP _i Q _i <Q _i-1 ΔWP _i = 2WP _i-1 - WP _i The output ΔWP _i of the subtracter 61 passes through the DA converter 9 and the attenuator 10, and then is transferred to the capacitor C and the operational amplifier 11.
is added to an integrator consisting of . The difference value ΔWP _i is
Since it corresponds to the characteristic value of WP _i , the output of the integrator is
The waveform is the product of W _i and E _i .

The comparator 43 and shifter 41 are connected to the upper code Q _i and
Q _i-1 is different. That is, this is to correct the difference between P^ _i and P^ _i-1 when the ranges of the index display part of the envelope data are different, so that a correct difference value ΔWP _i can be obtained.

Since ΔWP _i is the difference between WP _i and WP _i-1 , WP _i
Even if ΔWP i is 12 bits, the word width of ΔWP _i is usually small. Therefore, in this embodiment in which the difference is calculated, the word width of the DA converter 9 can be made 12 bits or less.

In the embodiment shown in FIG. 8, the comparator 4
3, a shifter 41, and a subtracter 61 constitute a differentiator.

FIG. 9 shows another example of the DA converter 9 used in FIG. 8. The difference value ΔWP _i is the shifter 70,
The signal is outputted via a DA converter 71 and an analog shifter 72, and applied to the attenuator 10 in FIG.
Shifter 70 and shifter 72 are supplied with the upper 3 bits of 7-bit data of ΔWP _i to control the shift operation. When the upper 3 bits of ΔWP _i are "0" as shown in FIG. 10A, the lower 4 bits are directly applied to the DA converter 71, and the shifter 72 is
The gain becomes 1/4, that is, -12dB. As shown in Figure 10B, when the upper 3 bits are “001”,
The middle 4 bits are output by the shift operation of the shifter 70, and are applied to the DA converter 71, so that the shifter 72 has a gain of 1/2, that is, -6 dB. 1st
When the upper 2 bits of ΔWP _i are “01” as shown in Figure 0C, the upper 4 bits excluding the MSB (most significant bit) are added to the DA converter 71, and the shifter 72 adjusts the attenuation degree.
Let it be OdB.

The above explanation applies when ΔWP _i is positive. ΔWP _i
FIG. 11 shows a case in which the two's complement representation is performed when is negative. In FIG. 11, when the upper three bits are "111", the lower bits shown in FIG. 11D are directly added to the DA converter 71. shifter 72
is -12dB. When the upper 3 bits are "110", the middle 4 bits in Figure 11E are transferred to shifter 7.
0 is output. The shifter 72 becomes -6 dB. When the upper 2 bits are “10”, 4 as shown in Figure 11F
Bits are output from shifter 70. shifter 72
becomes OdB. ΔWP _i Among the 7 bits, the MSB is
It is inverted and applied to the MSB input of the DA converter 71. In this way, this complement representation is converted to an offset binary representation.

The waveform generator 1 outputs a digital sound source waveform, and may be one that sequentially reads out data written in a storage device. envelope generator 2
, an up-down counter or the like is used.

It goes without saying that even when using increasing function type envelope data E _i as shown in FIGS. 5 and 6, it is possible to adopt a configuration that calculates the difference value as shown in FIG. 8.

Effects of the Invention As is clear from the above description, the present invention generates waveform data W _i from a waveform generator, and generates envelope data E _i that can be divided into a lower code P _i and an upper code Q _i from an envelope generator. , add the above waveform data W _i and the above lower order code P _i to the above multiplier to obtain their product WP _i , add the above product to a differentiator, calculate the difference value of the following two products, and calculate the above difference. The value output is added to the shifter and shifted according to the above upper code Q _i , so
Multipliers and
The scale of the DA converter can be reduced.

Further, in calculating the difference value of the product WP _i , if the upper code Q _i changes, the change is corrected and the difference calculation is performed, so that a correct difference can be obtained.

[Brief explanation of the drawing]

FIG. 1 is a block diagram for explaining the principle of the present invention, FIG. 2 is a block diagram showing another example for explaining the principle of the present invention, and FIG. 3 is a block diagram for explaining the principle of the present invention.
FIG. 4 is a diagram showing an example of the format of envelope data, and FIGS. 5 and 6 are blocks showing further examples for explaining the principle of the present invention. FIG. 7 is a diagram showing an example of a shifter used in each of the above examples,
Figure 8 is a block diagram of an embodiment of the present invention that outputs a difference value, and Figure 9 is a block diagram of the embodiment of Figure 8 in which the DA converter is configured as a floating type DA converter. , FIG. 10, and FIG. 11 are diagrams showing the code table of FIG. 9. 1... Waveform generator, 2... Envelope generator, 6... Multiplier, 8... Shifter, 9... DA converter, 10... Analog shifter.

Claims (1)

[Scope of Claims] 1. Waveform data is generated from a waveform generator, envelope data that can be divided into a lower code and an upper code is generated from an envelope generator, and the waveform data and the lower code are added to a multiplier to process them. Obtain the product of An envelope control device that shifts according to the code. 2. The envelope control device according to claim 1, wherein in calculating the difference value, when the code changes, the change is corrected and the difference calculation is performed.