JPH07101349B2 - Envelope extraction device - Google Patents

Description

Description: TECHNICAL FIELD OF THE INVENTION The present invention relates to an envelope extracting device for extracting an envelope signal of an input waveform signal.

[Background of the Invention]

BACKGROUND ART Conventionally, research has been conducted to generate various sounds by imitating natural sounds. In this kind of research, it has long been accepted that natural sounds have envelope changes such as their volume over time, and these are extracted and used as parameters for obtaining more realistic sounds. It is understood that it is possible. However, the current situation is that no specific solution has yet been found.

In addition, it has a sampling function of inputting an acoustic signal from the outside called a sampling machine or a sampler in recent years, digitally recording the acoustic signal, reading the recorded waveform signal as a sound source signal, and outputting each scale tone signal. Various electronic musical instruments have been developed.

In this type of electronic musical instrument, an acoustic signal is generally sampled as it is, converted into a digital signal and recorded, and since the input waveform signal has a unique envelope, Even when a digitally recorded waveform signal is read out at a predetermined speed and converted to an analog signal and then an envelope is added by a voltage control type amplifier (VCA), etc., the original envelope and the envelope added later are superimposed. It will be a combination,
It was difficult to perform sufficient envelope control.

In addition, even when a specific section of the waveform signal as described above is repeated to generate a continuous read sound or the like, if the waveform signal originally has an attenuated envelope, for example, the level at the joint of the waveform is There is a disadvantage that the difference appears as a noticeable amount and noise such as a click sound is output, and when the loop function is to be implemented, there is a problem that it can be performed only in a short section.

Therefore, even in an electronic musical instrument having this type of sampling function, it is necessary to extract the envelope signal of the original input waveform signal and remove it from the input waveform signal.

As described above, the necessity of extracting the envelope signal of the input waveform signal is felt, and various concrete configurations have been proposed. For example, a method has been proposed in which the peak value of an input analog waveform signal is held and the held peak value signal is output as an envelope signal via an integrator circuit. However, there are various problems such as the fact that a faithful envelope signal cannot be extracted due to the influence of the time constant of the integrating circuit.

[Object of the Invention]

The present invention has been made in view of the above circumstances,
It is an object of the present invention to provide an envelope extraction device that can extract an envelope signal that is more faithful to the waveform signal from a digital format.

[Main points of the invention]

In order to achieve the above-mentioned object, the present invention has a waveform storage means for digitally storing an input waveform signal, a division means for dividing the waveform signal stored in the waveform storage means into a plurality of blocks, and a plurality of division means by the division means. From the waveform signal divided into blocks, the address where the absolute value of the waveform signal takes the maximum value and the maximum value are detected in each block for each block,
Envelope signal calculating means for calculating the envelope signal by interpolating the maximum value of each block by digital operation processing from the address and the maximum value of the absolute value of the waveform signal detected for each block. The key point is that

[Structure of Invention]

A configuration when the present invention is applied to the electronic musical instrument having the above-described sampling function will be described with reference to FIG.

That is, reference numeral 1 is a waveform storage means for storing the waveform signal, and the waveform signal can be externally input and set in advance.

Then, the waveform signal stored in the waveform storage means 1 is read by the block reading means 2 for each block with the predetermined width of the address as a block.

The waveform signal read by the block reading means 2 is supplied to the maximum absolute value detecting means 3, and the address at which the absolute value takes a large value and the maximum value are detected in the block,
It is given to the maximum value / address storage means 4. Therefore, the maximum value / address storage means 4 sequentially stores the maximum value of all blocks and the address having the maximum value.

The output of the maximum value / address storage means 4 is supplied to the interpolation envelope value calculation means together with the sample value of the waveform read from the waveform storage means 1 by the waveform reading means 5, and the interpolation envelope value is calculated, It is given to the envelope normalization means 7.

The envelope normalizing means 7 performs a predetermined calculation from the sample value supplied from the waveform reading means 5 and the interpolation envelope value supplied from the interpolation envelope value calculating means 6 so as to remove the original envelope of the waveform signal. To obtain a normalized sample value.

Then, this sample value is sent to the waveform writing means 8 and is again supplied to the waveform storing means 1, so that the stored contents can be rewritten.

In this way, the waveform signal having the envelope is converted into the waveform signal of the envelope having a substantially constant value by removing the envelope according to the interpolation envelope value calculated by the interpolation envelope value calculation means 6. . That is, a normalized waveform signal is obtained by the series of processes.

The configuration and features of the present invention will be understood concretely by the embodiments described in detail below.

〔Example〕

Hereinafter, an embodiment in which the present invention is configured by using a CPU such as a microprocessor will be described in detail.

FIG. 2 shows its circuit configuration. An acoustic signal input through the microphone 11 is sampled by the waveform input circuit 12 at an appropriate sampling frequency, converted into a digital signal, and supplied to the CPU 13.

The CPU 13 sends this input waveform signal to the waveform memory read / write circuit 14 and finally stores it in the waveform memory 15.
Therefore, the waveform signal having the envelope of the input sound is digitally stored in the waveform memory 15.

Further, the CPU 13 normalizes the waveform signal stored in the waveform memory 15 to remove the envelope. The working register 16 used at that time includes an address register (ad register), a dt register, a block register (bl register), an ev register, and a madj / mdtj register. Data and addresses, which will be described later, are stored in these registers.

Also, this CPU 13 is connected to the keyboard 17,
In addition to the CPU 13 performing various processes based on the key operation of the keyboard 17, the waveform signal normalized from the waveform memory 15 at a predetermined speed corresponding to each scale according to the operation of the performance keys provided on the keyboard 17. The waveform memory read / write circuit 14 is controlled so as to read.

Then, the digital signal read from the waveform memory 15 is supplied to the D / A converter 18 via the waveform memory read / write circuit 14 and converted into an analog signal, and the voltage controlled amplifier (VCA) 19 Applied to.

The VCA 19 is supplied with the envelope signal calculated and output by the CPU 13 by the soft processing after being converted into an analog signal by the D / A converter 20, and the amplification factor is determined, as a result, from the VCA 19 , A waveform signal having an envelope designated by the CPU 13 is output, and the output of the VCA 19 is converted into an acoustic output by the sound system 21.

Next, the operation of this embodiment will be described with reference to the flowchart of FIG. First, this series of processing is performed by an address where the absolute value of the sample value in the block has the maximum value, a routine A for obtaining the maximum value, and a routine for normalizing the stored contents of the waveform memory 15 from the result of this routine A. It consists of B and.

First, in step A1 in routine A, the block register (bl register) in the working register 16 is initialized. That is, 1 is set to specify the first block. In addition, the block means each of the areas when the waveform memory 15 is divided by a predetermined address width BS.

Next, in step A2, the address register (ad register) is initialized. That is, the content of the ad register is used as data for designating the start address of the waveform memory 15. Then, in step A3, the address of the maximum value among the absolute values of the sample values and the contents of the register (madj / mdtj register) storing the maximum value are initialized.

That is, when the first block of the waveform memory 15 is j = 1 and the last block is j = N, the contents of the madj / mdtj register are set to zero in the range of j = 1 to N, and
Of the virtual block j = 0, which is one before the first block j = 1
The contents of the mado / mdto register, the address of the waveform memory
Width of each block from the start address of (block size B
S) to the value obtained by subtracting the maximum value and setting the maximum value to zero. Similarly, the final block j = N, one virtual block after the final block j = N.
= N + 1 mad _{N + 1} / mdt _{N + 1} register contents, block size from the final address of waveform memory 15
Set to the value with BS added, and set the maximum value to zero.

Then, by providing the above series of initial setting processing,
The maximum value of each block is detected in the processing after A4. First, at step A4, the sample value of the waveform memory 15 designated by the content of the ad register is read and the absolute value is obtained.

Then, at the next step A5, it is detected whether or not the absolute value is larger than the maximum value mad _{bl of the} current block which has already been recorded. If YES, the process proceeds to step A6 and madj / md
Rewrite the value of the tj register.

Then proceed to step A7. Also, when NO is determined in step A5, the process proceeds to step A7 following step A5. At step A7, the value of the ad register is incremented.

Then, in step A8, it is judged whether the value of the ad register is the final address of the block (that is, it is determined whether or not the value of the ad register matches the value of b1 × BS), and if NO. The process returns to step A4 again, and the processes of steps A4 to A8 are continuously executed. In this way, the maximum value of the block and its address are obtained.

FIG. 4 shows the relationship between the sample address of each block and its maximum value, and the values shown in the figure are stored in the madj / matj register (j = bl). The waveform shown by the dotted line in the figure is obtained by rewriting the sample points having a negative value to have a positive value by taking the absolute value of each sample value as described above.

When the processing of the block is completed, step A8
Then, make a YES decision and proceed to the next step A9. Step A
At 9, the value of the bl register is incremented. At the next step A10, that block is the last block (j =
N), if YES, return to Mayflow N
If it is O, the process returns to step A4 to start the process for the next block.

In this way, the address at which the absolute value of the sample value of each block (j = 1 to N) takes the maximum value and the maximum value are
It is stored and set in the madj / matj register.

Then, the normalization processing of the waveform memory 15 of the routine B is executed next.

First, in step B1, in the working register 16
The content of the ad register is set to the address that specifies the start address of the waveform memory 15. It also sets the contents of the bl register to a value of 1 that specifies the start address.

Next, in step B2, the waveform sample value at the address designated by the ad register is read from the waveform memory 15 and set in the dt register in the working register 16.

Next, in step B3, the interpolation envelope value is calculated and ev
Input to the register. That is, as shown in FIG. 4, the interpolation envelope value ev is obtained from the maximum value (mdt _j-1 ) of the previous block, the maximum value (mdt _j ) of the block, and the address (ad) at that time.

That is, the calculation formula is as follows.

Note that if j = bl, it corresponds to the values shown in FIG.

Then, in step B4, the waveform is normalized using the interpolation envelope value ev at the address point obtained as described above. That is, the following calculation is performed in a concrete manner.

Dt = MV × dt / ev Note that MV is a level value to be normalized, and is 700 in hexadecimal display in the embodiment. That is, the interpolation envelope value ev is an approximation of the envelope of the original waveform and causes an error, and depending on the input waveform, a normal value that exceeds the value of MV can occur, so the maximum value of the waveform is , For example
In the case of the 12-bit first complement expression, a value smaller than 7FF (hexadecimal expression) must be taken. Then, the normalized sample value Dt is input and set to the address of the waveform memory 15. Then, in step B5, the content of the ad register is incremented.

At the next step B6, the contents of the ad register and mad _bl
The ad register is mad when the size of the register is judged.
_When the value of the bl register is exceeded, the process proceeds to step B7 and the bl register is incremented.

Then proceed to step B8. Also, when the determination of NO is made in step B6, the process directly goes to step B8. Then, at step B8, it is judged whether the normalization processing has been completed up to the last sample value of the waveform.
Return to B2.

In this way, steps B2 to B8 are repeatedly executed to complete the normalization processing of all sample points. FIG. 5 (B) shows the waveform thus normalized, and the envelope as shown in FIG. 5 (A) (shown by a chain line).
The input waveform signal with is removed from the envelope by the execution of routines A and B and has a substantially constant value (corresponding to MV).
Is converted into a waveform signal having an envelope (indicated by a chain line).

Therefore, according to the present embodiment, even if the waveform sample value re-stored in the waveform memory 15 is read and the envelope is added to it by the VCA 19, the envelope originally possessed by the input waveform signal is removed. Effective envelope control is possible.

Further, even if a part of the waveform stored in the waveform memory 15 is read out by looping, the noise is not generated, such as unnecessary clicking noise, because the envelope is kept substantially constant.

Although the contents of the waveform memory 15 are rewritten to the normalized waveform signal in the above embodiment, they may be written to another waveform memory.

Further, in the above embodiment, the interpolation envelope value is obtained according to the above-described calculation formula, and the sample value of the waveform is divided by this interpolation envelope value to obtain the normalized sample value. You may ask for it. The point is to remove the envelope originally possessed by the input waveform signal.

Further, although the present invention is applied to the electronic musical instrument having the sampling function in the above-described embodiments, the present invention is also applicable to the electronic musical instrument in which the envelope of the input waveform signal is extracted and used as the acoustic parameter. Of course, can be applied.

〔The invention's effect〕

As described in detail above, the present invention enables extraction of an envelope signal from a digital waveform signal, and is very effective in obtaining a sound source waveform signal of an electronic musical instrument or the like and an envelope parameter signal. Moreover, the peak value of the waveform is discretely detected in order to extract the envelope signal as in the conventional case, and the envelope signal is not generated between the peak value and the peak value via the integrating circuit.
The maximum absolute value of each waveform is detected from the address where the absolute value of the waveform signal takes the maximum value and the maximum value, and the address where the absolute value of the waveform signal detected at each block has the maximum value and the maximum value. Since the value is interpolated by digital calculation processing to calculate the envelope signal, it is possible to extract a more faithful envelope signal from the waveform signal without being affected by the time constant of the integrating circuit as in the conventional case. Become.

[Brief description of drawings]

The drawings show the present invention, FIG. 1 is a diagram showing a configuration when the present invention is applied to an electronic musical instrument having a sampling function, FIG. 2 is a circuit configuration diagram of an embodiment of the present invention, and FIG. A flow chart showing the operation of the embodiment, FIG. 4 is an explanatory view of a concrete operation for calculating an interpolation envelope value, and FIG. 5 is a waveform diagram showing an input waveform signal and a normalized waveform signal. . 1 ... Waveform storage means, 3 ... Maximum absolute value detection means, 4 ...
... maximum value / address storage means, 6 ... interpolation envelope value calculation means, 7 ... envelope normalization means, 13 ... CP
U, 15 ... Waveform memory, 16 ... Working register.

Claims (2)

[Claims] 1. A waveform storage means for digitally storing an input waveform signal, a dividing means for dividing the waveform signal stored in the waveform storage means into a plurality of blocks, and a waveform divided into a plurality of blocks by the dividing means. The address at which the absolute value of the waveform signal takes the maximum value in each block from the signal and the maximum value is detected, and the address at which the absolute value of the waveform signal detected in each block takes the maximum value and the maximum And an envelope signal calculation means for calculating the envelope signal by interpolating the maximum value of each block by digital calculation processing from the value. 2. The envelope signal calculating means calculates the envelope value ev by the interpolation as follows: ev = mdt _bl-1 − (mad _bl-1 −ad) × {(mdt _bl −mdt _bl-1 ) / (Mad _bl −mad _bl-1 )} where bl is the block number, mdt is the maximum value of the block, mad is the address that takes the maximum value, and ad is the sample value address of the waveform signal. The envelope extraction apparatus according to claim 1, wherein the envelope extraction apparatus calculates the envelope.