JPS5840760B2 - Musical waveform noise removal method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a musical sound waveform noise removal method for removing noise on a musical sound waveform of an electronic musical instrument that performs digital processing.

Conventionally, in electronic musical instruments that perform analog processing, a method using an analog filter or the like has been adopted as a means for removing or reducing waveform distortion and noise appearing on a musical sound waveform.

However, recently developed electronic musical instruments that perform digital processing produce noise in musical waveforms that is different from that of electronic musical instruments that perform analog processing.

There is no suitable method for eliminating or reducing this that is suitable for electronic musical instruments.

FIG. 1 is a schematic diagram of a conventional digital processing electronic organ, in which 1 is a stop or tablet, 2 is a digital processing section, 3 is a D/A converter, 4 is a sound device, and 5 is a keyboard.

Information on each tablet, stop, piston, knob, etc. is provided from the stop or tablet 1, and information on the pressed state of each weight is provided from the keyboard 5 to the digital processing section 2.

The digital processing section 2 processes the musical sound waveform according to the input information and outputs it as a digital signal.

The D/A converter 3 converts the digital signal into an analog signal, and the audio device 4 outputs it as a musical tone.

In this case, the size of the musical sound waveform differs depending on the type of tone selected by the tablet.

Among various tones, if the ratio of the waveform size with the minimum amplitude to the waveform size with the average amplitude is m, the number of tablets in the electronic organ is n, and the maximum number of pressed keys on the keyboard is p, then the electronic The dynamic range of the musical sound waveform of an organ is m
It becomes np times.

However, in this case, volume adjustment using exgression pedals etc. is not included.

Therefore, the digital output signal from the digital processing section shown in FIG. 1 also has this dynamic range e.

m, that is, the relative level ratio between each tone is at least m=3
is necessary.

Originally, p is the number of keys, but due to the simplification of the digital processing unit, in the digital processing organs that have appeared to date, the number pf! : There are many restrictions, and considering the number of fingers on the hand, etc., it can be set as p-12.

The number of tablets n is considered to be at least n=10 in an electronic organ.

Therefore, m-n-p=3.10.12=360, and its representation requires 9 bits in binary code.

Normally, this dynamic range ranges from several hundred to several thousand times.
More than 10 bits are required for the decimal code.

Further, the number of bits required to sufficiently represent a musical sound waveform by, for example, pulse code modulation is normally 7 to 8 bits, and at least 5 bits are required.

From the above, the number of digital output bits of the digital processing section 2 is usually close to 20 bits, and at least 14 bits are required.

The digital processing section 2 is composed of logic circuits such as various registers, counters, adders, gate multipliers, etc., and the delay time increases in proportion to the number of pixels due to carry of the adder, etc.

Delay time can be reduced by constructing high-speed logic elements, but for reasons such as lower power consumption, higher integration, higher reliability, and lower cost, it is necessary to construct logic circuits with the lowest possible operating frequency using low-speed logic elements. It is desirable to do so.

Therefore, since a circuit that outputs at least 14 bits or more in parallel is constructed of low-speed logic elements, the change point of each bit may shift due to accumulation of delay time, etc., or a hazard may occur.

Then, thin pulse-like noise such as a so-called glitch appears in the analog signal of the D/A converter 3 due to the switching element.

The magnitude, or energy, of this noise is determined by the product of the width and height of the thin pulse-like waveform.

The width of this noise is determined by the speed of the logic element and the number of bits, but the height of the noise is determined by the point where many bits change among each bit, and the most significant bit (MSB).
), etc., appears higher at the point where the bit changes, and is considered to be proportional to the sum of the weights of the changing bits.

The most significant change is that all bits change, and when the bit configuration is , the bit string changes from (1) to (2), and from (2) to (
This is the point where it changes to 1).

Generally, when an envelope is added to a musical tone signal,
The waveform has a substantially symmetrical shape in the positive and negative directions with the zero level as the center.

FIG. 2 shows one example.

Therefore, in order to use the D/A converter 3 effectively, it is necessary to
This zero level should be set approximately in the middle of the output range of the A converter.

In other words, the digital power of the D/A converter is the bit string (1)
Or in case (2), the analog output outputs zero.

The points indicated by bit string (1) or (2) are pl, p2, . ......This is the position of p6.

1) t + p3) p5 and p2, p4 +
Since pa appears at the same frequency as one cycle of the waveform, the noise has a component at the same frequency as the frequency component of the musical sound waveform, and the noise always has a nearly constant magnitude for the analog output range of the D/A converter. However, since the dynamic range of the musical sound waveform is wide, the noise appears relatively large, especially for small amplitude waveforms.

This noise is extremely audible and must be removed.

SUMMARY OF THE INVENTION An object of the present invention is to provide a musical sound waveform noise removal method for reducing noise on a musical sound waveform of an electronic musical instrument that performs digital processing.

In order to achieve the above object, the musical waveform noise removal method of the present invention sequentially writes musical waveform input signals sampled at a predetermined clock into a first storage means, and further stores the contents of the first storage means into a second storage means. The contents of the fourth and second storage means are compared by a comparator, and based on the coincidence signal, the contents of the second storage means are written and outputted to the third storage means. By doing so, noise is reduced from the musical sound waveform.

The present invention will be described in detail below with reference to examples.

FIG. 4 is an explanatory diagram showing the configuration of an embodiment of the present invention, and a dotted line frame 10 indicates a digital waveform shaping circuit.

In the figure, 11, 12, and 13 are registers, 14 is a clock generator, and 15 is a comparator.

A digitally displayed tone waveform signal from the digital processing section is given to the register 11, and written into the register 11 by the clock CL1 from the clock generator 14.

The output of register 11 is then written to register 12 by clock CL1.

At this time, since the registers 11 and 12 are written by the clock CL1, the sampling position of the input data on the time axis is shifted by one pulse of the clock CL1.

The outputs of registers 11 and 120 are compared by comparator 15, and a match signal is given to register 13 only when they match.

The output of the register 12 is written into the register 13 using a match signal, and at the same time, the output is sent to the D/A converter.

If there is a shift in the output signal from the digital processing section due to the delay of each bit, the clock C
Since they are written simultaneously by L1, they are performed in synchronization with CL1.

Figure 5 shows tl, t2゜t3 of CLl for bits 1 to 5.
．． This shows the waveform information of each bit when sampled at time t4.

In this case, in the case of the position of 11.13 degrees t4, the deviation of each bit is eliminated.

However, in the case of the position t2, although each bit is synchronized, there is a good possibility that erroneous data will be written.

Therefore, the register 11 outputs incorrect data in the period from t2 to t3.

Register 11 and register 12 are compared by comparator 15, and only when they match are written to register 13, so if they do not match, register 13 is outputting the previous data.

That is, tl, t2. This data is written into the register 13 only when the values sampled at t3, t4, . . . are equal at the two sampling points and the data is constant.

Therefore, a hazard-like pulse that appears within one pulse width of clock CL1, for example, bit 3 in the figure, is in register 13.
Since it is not written to , it is removed by the waveform shaping circuit.

Further, the change points of each bit are synchronized in each register.

The above-described operation removes the noise.

Furthermore, regarding the setting of the cycle of clock CL1, since the delay time differs depending on the elements that make up the digital processing section, the width is wider than the maximum deviation of each bit shown in FIG.
Just set it.

Since musical waveforms are sampled using registers, consider the sampling theorem for audio frequency signals.
It is sufficient to set it narrower than the Nyquist interval.

Generally, a few microseconds is sufficient.

Next, FIG. 6 is an explanatory diagram showing the configuration 2 of another embodiment of the present invention, in which a dotted line frame 20 indicates an analog input waveform shaping circuit.

In the figure, reference numerals 21, 22, 23 are sample and hold circuits, 24 is a clock generator, and 25 is a comparator.

The waveform shaping circuit 20 is an example of the waveform shaping circuit 10 adapted to apply to analog signals.

In other words, the output of the D/A converter is sent to the sample hold circuit 2.
1 is given.

Further, the output of the sample hold circuit 23 is sent to an audio device.

The sample and hold circuits 21, 22, 230, write timing, etc. are the same as in the case of FIG. 4, and the clock generator 24 is the same as the clock generator 14.

The comparator 25 compares the analog signals of the sample and hold circuits 21 and 22, and the comparison timing is based on the clock C.
Given by L2.

A coincidence signal is given to the sample and hold circuit in synchronization with clock CL2 only when the difference between the input analog signals is smaller than a certain set value and almost equal.

The comparator 25 is composed of an analog subtraction circuit, an analog comparator, a gate circuit, and the like.

The waveform shaping circuit having the configuration shown in FIG. 6 can remove the noise even in the analog stage.

As explained above, according to the present invention, it is possible to very easily reduce noise on a musical tone waveform using the above-described method using three registers and a comparator.

The above noise can be removed to some extent by applying a deglitch type D/A converter, but this type of D/A converter
A/A converter is extremely expensive compared to a normal D/A converter and is not suitable for electronic musical instruments.

In addition, this type of D/A converter requires a write pulse to be given to the built-in register, but the write pulse is set when the input waveform is a composite waveform of musical waveforms having two or more different frequencies. It was not used in electronic musical instruments because it had the disadvantage of being extremely difficult to perform.

In contrast, the present invention has a simple configuration and is inexpensive, and can be said to be a waveform shaping method suitable for electronic musical instruments without the above-mentioned problems.

[Brief explanation of drawings]

Figure 1 is a schematic configuration diagram of a digital electronic organ, Figure 2
3 is a diagram explaining the principle of the present invention, FIG. 4 is an explanatory diagram showing the configuration of an embodiment of the present invention, FIG. 5 is a diagram explaining the same operation, and FIG.
The figure is an explanatory diagram showing the configuration of another embodiment of the present invention, in which 10 is a digital waveform shaping circuit, 1L12 and 13 are registers, 14.24 is a clock generator, 15.25 is a comparator, and 20 is a The analog waveform shaping circuit includes sample and hold circuits 21, 22, and 23.

Claims (1)

[Claims] 1. Sequentially write musical waveform input signals sampled at a predetermined clock into the first storage means, and
The contents of the storage means are sequentially written into the second storage means at the predetermined clock, the contents of the first and second storage means are compared by a comparator, and the contents of the second storage means are written based on the coincidence signal. A musical sound waveform noise removal method characterized in that noise is reduced from a musical sound waveform by writing and outputting it to a third storage means.