JPS5830548B2 - Zatsuon Pulse Noken Yutsuhoushiki - Google Patents

Description

[Detailed Description of the Invention] When pulse noise is mixed into a music signal or an audio signal, the mixed pulse noise causes discomfort to the listener. In the production of disc records, in order to prevent noise pulses from occurring in the playback signal from disc records in excess of an allowable number, the sukumba used in disc record production must be managed, and the materials used for disc records must be controlled. Various measures have been taken to prevent dust from adhering to the discs and to ensure that there is no dust in the factory.In addition, for disc records that have been produced, sampling inspections are conducted to determine the playback signal by, for example, listening to them. Quality control of disc records is carried out by counting and evaluating the number of noise pulses in the reproduced signal, or electrically counting the number of noise pulses in the reproduced signal.

The conventional method for electrically counting the number of noise pulses in the reproduced signal described above is, for example, as shown in the block diagram shown in Figure 1. 2 is an amplitude detection circuit, 3 is an attenuator, 4 is a low-pass filter, 5 is a comparator, and 6 is a counter. The reproduced signal A is subjected to half-wave rectification in the amplitude detection circuit 2 as shown at the signal port in FIG. 2, and then applied to the attenuator 3 and the low-pass filter 4.

The output from the attenuator 3 is the half-wave rectified signal port which has been moderately attenuated, as shown in signal C shown in FIG. Supplied with input.

Further, the signal port described above is passed through a low-pass filter 4 to form a signal 2 according to the envelope of the signal port, and is provided as the other input to a comparator 5.

The comparator 5 compares the amplitudes of the signal C and the signal 2, and when the amplitude of the signal C is larger than the amplitude of the signal 2 by a predetermined value, the signal H as shown in FIG. Counter 6
Send to.

The conventional method shown in Figure 1 is based on the principle of detecting noise pulses that the noise pulses mixed into the original information signal have sharp rising characteristics. When the original information signal itself contains many mechanical signal components, such as when the original information signal is a music signal, it is difficult to distinguish between the original information signal and the noise pulse. The problem is that false detections occur.

To avoid false detection due to the above causes, for example, use a low-pass filter 4 with a small time constant so that the low-pass filter 4 does not respond only to input signals with extremely sharp rises. Therefore, the comparator 6 must be designed to operate at that time.

However, the above-mentioned noise pulse with an extremely sharp rise generally has a narrow time span, so the energy of the noise pulse is not so great that the peak value of the pulse is high. In many cases, even when reproduced as a noise, the noise does not show a large loudness.In fact, even if the noise pulse has a not-so-sharp rise, a noise pulse with a large pulse width is considered to be a noise that shows a large loudness in the reproduced sound. Therefore, as in the conventional method with the configuration shown in Figure 1 above, we focus only on the sharpness of the rise of the signal and compare the loudness of the noise in the reproduced sound. If a detection method that does not use this as a detection condition is used, the effective number of noise pulses that may generate noise with a loudness that is evaluated as harmful noise in the reproduced sound is calculated. Therefore, it is not possible to detect and count them.

The present invention provides electrical signals (hereinafter referred to as
Sometimes described as a program signal: is generally
Even if the rise of the envelope is steep, the fall is gradual due to the presence of the lingering sound of the instrument or the reverberation of the recording field. We focused on the fact that the rise of the noise pulse is sharp, and the fall is also extremely sharp based on the transient characteristics of pickups and amplifiers. The present invention provides a noise pulse detection method that uses the abrupt falling part of the signal to discriminate between a program signal and a noise pulse, thereby making it possible to effectively detect noise pulses mixed in a program signal. The problem of the conventional method described above is solved, and the noise pulse is After obtaining a level value corresponding to approximately the total energy value of The present invention provides a method for detecting noise pulses, and its contents will be specifically explained below with reference to the accompanying drawings.

FIG. 3 is a block diagram of one embodiment of the noise pulse detection method of the present invention, in which:
Components using the same drawing numerals as those used in the block diagram shown in the first diagram described above are shown in the first diagram.
This is a component corresponding to the component explained in the figure.

The playback signal from a disc record is a program signal mixed with noise pulses, but the program signal has large energy in low frequency components, and
Since the noise pulse mainly consists of high frequency components, input terminal 1 is
It is preferable that the reproduction signal supplied from the disc record is passed through a suitable high-pass filter so that the noise pulses stand out.

The reproduced signal a as shown in FIG. 4a, which is supplied to the input terminal 1 of the reproduced signal, is half-wave rectified by the amplitude detection circuit 2 to produce a signal as shown in FIG. is passed through the low-pass filter 7 set to a high value, resulting in a signal C having valleys between each peak as shown in FIG. 4,
It is applied to the logarithm circuit 14, the differentiation circuit 24, etc.

Now, the noise pulse mixed into the program signal is as follows.
There are various types of noise pulses, such as those with a narrow pulse width but a high peak value, and those with a relatively wide pulse width but a low peak value. Even if the energy of the noise pulses is the same, it will be felt as noise of the same size in the reproduced sound. For noise pulses that have, it is desirable to replace them with signals exhibiting similar voltage levels.

The low-pass filter 7, whose cut-off frequency value is set to a relatively high frequency value, outputs an output corresponding to the average value of each of the noise pulses mixed into the program signal. By doing so, it is possible to obtain signals showing approximately the same voltage level for noise pulses having similar energy, regardless of the difference in the form of the noise pulses.

The outputs of the attenuator 3 and the low-pass filter 4, to which the output signal C from the low-pass filter 7 described above is applied, are applied to the differential amplifier 5.

The output from the low-pass filter 4 represents the average signal level of the signal C applied to it, and the output from the attenuator 3 is the signal C applied to it attenuated to a predetermined value. , the differential amplifier 5 supplies an output signal corresponding to the difference between the two outputs to the square circuit 8 only when the output from the attenuator 3 is larger than the output from the low-pass filter 4;
The output of the multiplier circuit 8 is integrated by an integration circuit 9 and then provided to two Schmitt circuits 10 and 11.

The output signal from the above-mentioned integrating circuit 9 represents the magnitude of the energy of the peak portion that jumps out from the average signal level in the reproduced signal.

The integrator circuit 9 has a fourth integrator which is reset after the noise pulse disappears.
It is reset by a signal l as shown in FIG. 1 (how the signal l is generated will be described later), and at that time the output of the integrating circuit 9 is returned to zero.

By the way, there is a difference between a certain size ratio of the amount of sound energy and the ratio of the size of the sound as perceived by humans, for example, when the ratio of the amount of sound energy is 1:20, Since there is a certain relationship such that the ratio of the sound magnitude to the human sense is 1:3, the output has a magnitude proportional to the amount of sound energy obtained from the above-mentioned integrating circuit 9. of,
Setting a value that corresponds to the human perception of sound loudness is significant because it allows the entire noise pulse to be evaluated numerically.

Therefore, in the illustrated embodiment, the integrating circuit 9
A signal having a magnitude proportional to the amount of sound energy outputted from the is AD converted by two Schmitt circuits io and 1i set to have mutually different threshold values. The outputs are added by an adder 12.

As an example, if the threshold of the Schmitt circuit 11 is set to be approximately 20 times the threshold of the Schmitt circuit 10, the Schmitt circuit 11 will respond to an input approximately 20 times or more higher than the response level of the Schmitt circuit 10. Because you respond to
Adder 12 into which the output from Schmitt circuit 10 is input
By setting the output from the Schmitt circuit 11 to be input to the input terminal 12b of the adder 12 which is one digit higher in binary numbers with respect to the input terminal 12a of the adder 12, only the output from the Schmitt circuit 10 is input to the adder When the output from the Schmitt circuit 10 and the output from the Schmitt circuit 11 are both input to the adder 12, the addition value is 1+2=3, whereas the addition value is 1 when input to the adder 12. This value corresponds to the relationship that when the amount of energy of a continuous sound increases by 20 times, the human perception of the loudness of the sound increases by 3 times. The resulting added value is a numerical value suitable for evaluating the noise of a plurality of noise pulses, each having a different level value.

The addition operation of the adder 12 described above is performed when the pulse m shown in FIG.

Note that the count value of the adder 12 is reset by applying to the terminal 12d a signal generated by the opening/closing operation of a switch in response to the operation of separating the pickup from the armrest in the record reproducing apparatus at the start of the next performance.

Next, the signal C output from the low-pass filter 7 described above is
The differential circuit 24 differentiates the signal d to form a signal d as shown in FIG. 4d, and the waveform shaping circuit 25 shapes the signal d to form a signal e as shown in FIG. 4e.

The falling point t1 of the signal e corresponds to the position of the zero point in the signal d, and the fact that the position of the zero point in the signal d corresponds to the peak position of the signal C means that the signal d differentiates the signal C. This is clear from the fact that it was obtained by

The above-mentioned signal e is given to the first hold circuit 15, inverter 16, monostable multi-by-break 26, etc., but first, the signal C shown in FIG. The supplied first hold circuit 15 holds the signal e at the falling time tl, 1
2. Since the input signal at each time point is sampled and held every t4, the first hold circuit 15 outputs successive peak values PI in the signal C shown in FIG.

A signal f as shown in FIG.
and the hold circuit 17.

The logarithm circuit 14 described above is used to compress the dynamic range of the signal so that the hold circuit can operate well even with input signals having a wide level range.

The above-mentioned signal e is supplied to the second hold circuit 17 after its polarity is inverted by the inverter 16. Therefore, the second hold circuit 17 performs the first hold circuit at each rising point of the signal e. The output signal f from the circuit 15 is sampled and held, and a signal g as shown in FIG. 4g is applied to the comparator 18.

Furthermore, the monostable multi-by-break 26 to which the signal e is applied is activated every time the signal e falls.
A pulse h whose pulse width is sufficiently narrower than the duration of the signal e.
(4th diagram), this pulse h activates another monostable multi-by-break 27 by its fall,
A pulse l as shown in FIG. 4i is generated.

The first point mentioned above. The comparator 18 to which the output signal f2g from the second hold circuits 15 and 17 is applied as a dual force is as follows:
It is designed to operate when the difference between the two human forces f2g is above a certain value, and this is because, for example, the input side has hysteresis characteristics due to a combination configuration of a differential amplifier circuit and a Schmitt circuit. Being done and being the first
When the signal f from the hold circuit 15 becomes larger than the signal g from the second hold circuit 17 by more than a certain value, the output state becomes high level, and then the magnitude relationship of the two signals f2g is reversed. The high level state is maintained until the signal g becomes larger than the signal f by more than a certain value, and the output state becomes low level when the signal g becomes larger than the signal f by more than a certain value. It acts like a signal.

Therefore, the first. When the signal f2g from the second hold circuit 15.17 is as shown in FIG. 4 f2g, the comparator 18 outputs a signal J as shown in FIG. Added as one input to the navel.

Since the output pulse 1 of the monostable multi-bi break 27 described above is added to the AND gate 19 as the other input, if the signal J is at a high level during the period when the pulse i is at a high level, , an output pulse k is applied from the ant gate 19 to the monostable multivib break 20 and the inverter 21.

In the embodiment shown in FIG.
The time position of the rise of the pulse i given to the first hold circuit 15 is delayed by the pulse width of the output pulse h of the monostable multi-bi break 26 from the time point at which the signal level of the output signal f from the first hold circuit 15 changes, for example, time t2. The reason for delaying the time point, for example, time t3, is that the comparison output due to a transient signal change portion when the signal level changes in the output signal from the hold circuit is input from the comparator 18 to the AND gate 19. This is to prevent the anime game 119 from operating during this period.

The monostable multi-by-break 20 described above outputs a pulse l (FIG. 4, FIG. 1) of a predetermined pulse width from the time t3 of the rising edge of the output pulse from the ant gate 19.
It is applied to the AND gate 22, the integrating circuit 9, and the comparator 18.

The occurrence of this pulse l indicates that the newly appearing peak P2 in the reproduced signal has a much higher peak value than the peak P1 that appeared before, and this pulse l The time constant of the monostable multivib break 20 is determined so that the monostable multivib break 20 remains at a high level for a predetermined period of time after it occurs.

If the peak P3 at time t4 that follows the peak P2 at time t2 that occurs in the reproduced signal has a much lower peak value than the peak P2, the signal j from the comparator 18 will change from time t4 to pulse h. Time t5 after the elapse of the pulse width time
When this low level state is made high level by the inverter 21 and applied to the AND gate 22, the monostable multivibrator 20
When the output pulse l of is in a high level state, that is, the peak P3 following the peak P2 has occurred within a predetermined time set by the monostable multi-vibration break 20, and the new peak at time t4 has occurred. Tana Peak P
When the peak value of P3 is determined to be much smaller than the peak value of the preceding peak P2, a high level output from the ant gate 22 is input to the ant gate 23. given as.

Since the AND gate 23 is given the output pulse i of the monostable multi-by-break 27, the AND gate 23 outputs a pulse m as shown in FIG. terminal 1
2c, and the adder 12 performs an addition operation using the pulse m.

The state in which the pulse m is applied from the AND gate 23 to the adder 12 is such that when the output pulse i of the monostable multi-bi break 27 is applied to the AND gate 23, the pulse m is applied from the AND gate 22 to the input of the AND gate 23. This is a case where the output of the AND gate 22 that is applied is at a high level, and this state is the middle pulse of the high level output pulse l of the monostable multi-bi break 20,
This is obtained only when the output of the AND gate 119 is at a low level at the time when the pulse i is generated.In the end, the pulse m is sent from the AND gate 23 to the adder 12 when the signal C is present. Each peak P that appears one after another
I, P2. The successive peaks are low,
This is only the case where the output signal from the comparator 18 changes from high to low and becomes high level only within a predetermined time span.

(As described above, the comparator 18 changes the level of its output signal J only when the level difference between the two human input signals to be compared therein is greater than a certain value.

By the way, the two-power signal f2g applied to the comparator 18 corresponds to the peak value of the adjacent peak in the output signal of the logarithm circuit 14;
Since it is the logarithm value of the peak value of adjacent peaks in the signal C supplied as an input signal to the comparator 18, it is possible to change the level state of the output signal of the comparator 18. Regarding the signal C, the level difference means that the ratio of the peak values of adjacent peaks in the signal C is greater than a certain value.

) Therefore, depending on the setting of the operating conditions of the comparator 18,
The ratio between the peak values at successive peaks in the signal C corresponding to the state in which the pulse m is to be sent to the adder 12 is determined, for example the ratio between the peak values at successive peaks in the signal C mentioned above. comparator 18 so that
If the operating conditions are set, the peak value of peak P2 following peak P1 in signal C is more than three times the peak value of Pl, and the peak value of peak P3 following peak P2 is the peak value of peak P2. AND gate 2 only when the peak P3 appears in the middle of the output pulse l from the monostable multi-bi break 20.
3 will send a pulse m to the adder 12.

Note that the comparator 18 and the integrator 9 are each reset when the signal l becomes low level.

As mentioned above, there are peaks in which the ratio of the peak values of adjacent peaks in signal C is equal to or higher than a certain ratio and the peak values are low, high, and low, and there are also peaks with high peak values. The pulse m is sent from the AND gate 23 to the adder 12 only when the next low peak appears within a predetermined time from the time position at which the high peak appears. It becomes possible to successfully detect only noise pulses that appear with high loudness in the noise pulse.

In other words, as mentioned above, among the noise pulses mixed in the program signal, the noise pulses that cause problems during playback are the noise pulses that appear with a large loudness in the reproduced sound, but these noise pulses are not the same as musical sounds. The present invention is different from the case where an impact noise that often appears in a car has a steep rise but whose envelope falls gradually due to the subsequent vibration. By determining whether or not an isolated peak reaches a peak value of a predetermined ratio within a predetermined time period as in the method of the present invention, it is possible to effectively discriminate between an impact sound in a musical tone and a noise pulse. Accordingly, the problems in the conventional method described above can be satisfactorily solved.

In the embodiment shown in the third figure described above, the predetermined time value set for determining the falling state of the pulse within a predetermined time is obtained using the monostable multi-vibration break 20. In another embodiment, gating means using a pulse counter can be used instead of the monostable multi-bibreak 20 in FIG.

That is, from the AND gate 19 in FIG. 3, a pulse is obtained for each peak in the signal due to the high level signal j from the comparator 18, unless successive peaks in the signal suddenly become low. This pulse is applied to a pulse counter that outputs a high level output when counting pulses below a certain number, and when a certain number of pulses or more are counted, the counted peak group is considered not to be a noise pulse and is counted. The output is reset to a low level, and the signal j from the comparator 18 described above is output when counting within a certain number.
When the signal becomes low level, a high level signal is applied from the AND gate 22 to the terminal 12c of the adder 12 to cause the adder 12 to perform an addition operation.

Note that the counter described above may be reset by the output signal from the AND gate 22.

As is clear from the above detailed explanation, in the noise pulse detection method of the present invention, isolated peaks such as peak P2 in FIG. This means that, for example, the signal Q in Figure 2A is not detected even if its rise is steep, and only the isolated pulse P can be detected as a noise pulse. In addition, in the method of the present invention, even if the rise is steep, such as an impact sound in a musical tone, the envelope falls slowly due to subsequent vibrations. If the peak values of the peaks that appear do not have a predetermined ratio, they are not detected as noise pulses, and as a result, noise pulses can be favorably distinguished from program signals.

Furthermore, in the method of the present invention, the degree of rise and fall of the envelope signal C obtained from the reproduced signal is determined from the magnitude of the ratio between the peak values of the signal C. In the conventional method, the above-mentioned problems in which noise pulses are detected only based on the steepness of the rise of the signal are satisfactorily solved by the method of the present invention. On the other hand, noise pulses having a relatively large pulse width and a wide pulse width can also be detected satisfactorily.

Furthermore, in the method of the present invention, in order to numerically express the amount of noise having a problematic loudness in reproduced sound, a signal with a level corresponding to the energy value of the noise pulse is obtained, and then Since the addition is weighted according to the level of loudness, the amount of noise on the entire surface of the disc record can be expressed as a single number, and the evaluation can be performed numerically. play.

[Brief explanation of the drawing]

Figure 1 is a block diagram of the conventional method, Figure 2 A to E are the first
A waveform diagram for explaining the operation of the illustrated method, FIG. 3 is a block diagram of an embodiment of the noise pulse detection method of the present invention, and FIG. 4 a''-m are for explaining the operation of the method illustrated in the third figure. It is a waveform diagram. 1... Input terminal, 2... Amplitude detection wave circuit, 3... Attenuator, 4, 7... Low pass filter. , 5, 18...Comparator, 6...Counter, 8...Square circuit, 9...Integrator circuit, io, ii... ... Schmitt circuit, 12...
... Adder, 13 ... Display, 14 ...
... Logarithmic circuit, 15... First hold circuit, 16, 21... Impark, 17...
・Second hold circuit, 18... Comparator, 19
,22,23...and gate, 20.26,
27... Monostable multibibreak, 24...
... Differential circuit, 25 ... Waveform shaping circuit.

Claims (1)

[Claims] 1. There are peak values such as low, high, and low in the envelope of the signal generated by the noise pulse mixed in the program signal, and the peak value in the signal envelope is
A noise pulse detected by distinguishing that it has a magnitude greater than a predetermined ratio compared to the peak value before the peak value appears and the peak value after the peak value appears within a predetermined time from the time the peak value appears. , the noise pulse is squared and integrated so that the evaluation judgment corresponds to the loudness of the noise when the noise pulse appears as noise in the reproduced sound, and approximately the total energy value of the noise pulse is calculated. After obtaining the level value corresponding to , it is AD converted and weighted addition is performed, and the entire noise amount of multiple noise pulses mixed in the program signal and each having a different level value is converted into one numerical value. Detection method of noise pulses displayed.