JPH06295194A - Signal comparing device - Google Patents

Abstract

(57) [Summary] [Purpose] To calculate an error that excludes instantaneous effects such as timing deviation between two types of signals. [Structure] A sampling unit 405 samples the feature amount of the first signal in accordance with a control signal from the control unit 403. The averaging means 406 averages the feature quantities of the first signal sampled by the sampling means 405.
Further, the sampling means 408 and the averaging means 409 perform similar processing on the second signal. The comparison means 410 compares and outputs the feature amounts of the first and second signals averaged by the averaging means 406 and 409.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal comparison device for comparing two kinds of signals, for example, a model singing recorded in a karaoke medium and a singing voice of a user singing karaoke.

[0002]

2. Description of the Related Art In recent years, karaoke apparatuses have been equipped with various functions such as key control and addition of reverberation so that a user can easily sing a song. In this
In order to automatically set a key controller, a karaoke device has been devised which includes a model singing recorded on a karaoke medium and a signal comparison device for comparing the singing voice of a user singing karaoke.

As such a conventional karaoke apparatus,
For example, there is Japanese Patent Laid-Open No. 59-204095, "Musical tone pitch changing device". In this conventional karaoke device, before the user sings a song, the pitch of the necessary portion recorded in the karaoke media is checked, while the user is uttered to check the vocal range of the user, the two are compared, and the result is compared. The key controller is controlled according to.

A conventional musical tone pitch varying apparatus is a frequency analyzing means for frequency-analyzing an input signal to determine a pitch, and two different kinds of voices which are separately input from the result of the pitch determination by the frequency analyzing means. A tone range checking means for checking the tone range of the signal in advance, and a frequency shift amount determining means for determining a frequency shift amount from each tone range of the two types of input audio signals checked by the tone range checking means, It is composed of frequency shift means for shifting the frequency of one input voice by the shift amount determined by the frequency shift amount determining means.

The operation of the conventional musical tone pitch varying device constructed as above will be described below.

First, a necessary part of the karaoke media, for example, the first part of the song to be sung is reproduced, and this is input to the frequency analysis means. Then, the sound range checking means checks the highest and lowest tones of the reproduced karaoke media based on the analysis result of the frequency analysis means. Next, the voice of the user is input to the frequency analysis means.
The user sequentially pronounces in the vocal range where he can correctly pronounce "do, le, mi, ...". The pronunciation of the user is analyzed by the frequency analysis means, and the range of the user is obtained by using the range check means.

As described above, the musical range recorded in the karaoke medium and the vocal range of the user can be obtained. These musical ranges are compared by the frequency shift amount determining means to determine the frequency shift amount. For example, when the range of the karaoke media is higher than that of the user, the frequency shift amount is determined so as to lower the range of the karaoke media. By reducing the reproduced sound of the karaoke media by the frequency shift means according to the frequency shift amount, the reproduced sound in the range suitable for the user's vocal range can be obtained.

[0008]

However, the conventional configuration as described above has a problem that the operation is troublesome. That is, before the user starts singing karaoke, he / she must first play a necessary part of the karaoke media and also check the vocal range of the user himself, and it takes a considerable time to actually start singing karaoke. It is virtually impossible to perform such an operation when a large number of people enjoy karaoke.

The present invention solves the above-mentioned conventional problems, and is necessary for controlling the pitch converting means or for displaying the pitch according to the song of the user while the user is singing karaoke. An object of the present invention is to provide a signal comparison device that compares signals.

[0010]

In order to achieve this object, a signal comparison apparatus according to the present invention comprises sampling means for sampling the feature quantity of the first signal according to the feature quantity detection flag of the first signal, Averaging means for averaging the sampled feature amounts of the first signal, sampling means for sampling the feature amount of the second signal in accordance with the feature amount detection flag of the second signal, and the sampled second It is composed of averaging means for averaging the feature amounts of the signals and comparison means for comparing the feature amounts of the averaged first and second signals.

Further, the signal comparison device of the present invention includes control means for outputting a control signal corresponding to the feature amount detection flag of the first signal and the feature amount detection flag of the second signal, and the control means in response to the control signal. Sampling means for sampling the characteristic amount of the first signal, averaging means for averaging the sampled characteristic amounts of the first signal, and sampling means for sampling the characteristic amount of the second signal according to the control signal. , And averaging means for averaging the sampled feature amounts of the second signal, and comparison means for comparing the averaged feature amounts of the first and second signals.

[0012]

According to the present invention, with the above configuration, the feature quantities of the first and second signals are sampled and averaged according to the feature quantity detection flags of the first and second signals, and then these feature quantities are compared. .

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described below with reference to the drawings. The features to be compared in the present invention include the strength of the signal, the temporal change in strength (rhythm),
Although there are pitches and the like, the case where the feature amount is the pitch of the signal will be described in the following embodiments.

FIG. 1 is a block diagram showing the configuration of a signal comparison apparatus according to the first embodiment of the present invention. In FIG. 1, 101 is a basic cycle input terminal for inputting the basic cycle of the first signal, 102 is a basic cycle detection flag input terminal of the first signal, and 103 is a basic cycle of the first signal according to the basic cycle detection flag. The period is sampled and the pitch data (cent '
Sampling means for accumulating after conversion into a value), 104 is an averaging means for averaging the accumulated first fundamental period data, 1
Reference numeral 05 is a basic cycle input terminal for inputting the basic cycle of the second signal, 106 is a basic cycle detection flag input terminal of the second signal, and 107 is a sampling of the basic cycle of the second signal.
Sampling means for converting into pitch data (cent 'value) and accumulating, 108 for averaging means for averaging accumulated pitch data, 109 for comparing means for comparing the averaged first and second pitch data. , 110 are output terminals for outputting the comparison result. In the first embodiment, the sampling means 103 and 107, the averaging means 104 and 108, and the comparing means 109 are realized by the microcomputer 111.

The operation of the signal comparison apparatus of the first embodiment constructed as above will be described.

First, the basic cycle data input from the basic cycle input terminals 101 and 105 will be described. There are methods of extracting the pitch of the voice, such as a method of using autocorrelation and a method of performing frequency analysis by FFT or the like. In the first embodiment, the case of using the method of extracting the pitch from the zero cross interval will be described. For details of the method for extracting the pitch from the zero-cross interval, see, for example, Japanese Patent Laid-Open No. 2-1
No. 08100, “Pitch Extractor”. This pitch extraction device inputs a voice signal and outputs data indicating how many times the fundamental period of the input voice is the sampling period.

The pitch extracting device detects the basic period of the input speech signal. However, when the continuously detected basic period is within a certain ratio, the basic period detection flag (PRL for the first signal, PRL, Set the second signal for PRR). These flags are reset in the timer interrupt, and are set when the pitch detecting device detects the basic period.
Therefore, if there is no zero crossing during sampling at fixed time intervals (when the input signal is low frequency),
Even if there is a zero cross, if the basic period cannot be detected (consonant, etc.), the basic period detection flag may remain reset.

Next, the contents of data sampling / accumulation processing will be described. FIG. 2 is a PAD for explaining sampling / accumulation of pitch data performed in interrupt processing. Of the variables used in the explanation, is
L0, isR0, scL, scR, and icnt are assumed to be initialized to zero.

The sampling means 103 and 107 sample the input basic cycle using the timer interrupt function of the microcomputer 109. In the case of human singing voice, a sampling interval of about 10 [ms] is appropriate. First, when a timer interrupt occurs, the basic cycle detection flag (PRL) is checked to check whether the basic cycle of the first signal has been detected (step 1).

When the fundamental period is detected, the frequency is obtained from the fundamental period and converted into the cent 'value (cnt0L) (step 2). Cen when the basic cycle is t [s]
The t'value c is given by equation (1).

C = 1024 * log2 (t0 / t), t0 = 1/55 (1) Equation (1) is an equation when the reference frequency is 55 [Hz]. If lower frequencies are to be handled, the reference frequency also needs to be lowered accordingly.

The conversion method to cent 'uses a table of the format for reading out the cent' value with the number of samples of the basic period as an index. If the basic cycle is short,
Since the interval between cent 'values becomes large, it is efficient to reduce the step size of the basic cycle and increase the step size stepwise as the basic cycle becomes longer.

The detected cent 'value of the first signal is accumulated in the accumulation variable (scL), and the accumulated number counter (isL0) is incremented (step 3). The basic period detection flag (PRL) is reset (step 4).

If the basic period is not detected, no processing is performed. Similarly, for the second signal basic cycle, basic cycle detection check (step 5), cent 'conversion (step 6), accumulation of scR, increment of counter isR0 (step 7), basic cycle detection flag (PRR)
Is reset (step 8).

The interrupt count counter (icnt) is incremented each time an interrupt occurs regardless of whether or not the basic cycle is detected (step 9).

It is checked whether or not the number of interrupts has reached the reference value (icnt0) (step 10). If it has, the accumulated result (scL, scR) is stored in the buffer (sscL, sscR).
Then, the accumulated number (icL0, isR0) is stored in the buffer (isL1, isR1) (step 11), and each variable is cleared (step 12). And the pitch data accumulation completion flag (CCNT)
Set.

The above is the contents of the data sampling / accumulation processing which is processed by the interrupt.

Next, the averaging process of the accumulation results of the averaging means 104 and 108 will be described. FIG. 3 is a PAD of a process for averaging accumulated results.

The averaging process does not need to be interrupted as long as the process is completed by the next sampling, as compared with the sampling in which the time must be accurately managed at fixed time intervals. Normal processing is used to reduce the interrupt load. When the pitch data accumulation completion flag (CCNT) is set during the loop processing such as the key scan after initialization by the microcomputer, the averaging processing is executed (step 1).

In the averaging process, first, it is judged whether or not the cumulative number (isL) of the first signals is equal to or larger than the reference number (MIN) (step 2). If there is more than the reference number of pitch data,
The accumulated data (sscL) / accumulated number (isL1) is used as the average pitch data (avcL), and the average completion flag (SRL) of the first signal
Is set (step 3). When the accumulated number (isL) is less than the reference number (MIN), the average pitch data (avcL) is set to 0 and the average completion flag is reset.

Similar processing (steps 5 to 7) is carried out for the second signal. The above is the content of the averaging process.

Next, the processing of the comparison means 109 will be described. Many karaoke devices have a pitch conversion function. For example, if the first signal is the user's microphone input voice and the second signal is the model singing voice recorded in the karaoke medium, the pitch conversion is performed in karaoke. If so, the current pitch change value is added to the average pitch data (avcR).
It is necessary to add key [cent '] to make a correction. After this,
An error keyerr of the pitch detection result is obtained by subtracting the average pitch data (avcR) of the second signal from the average pitch data (avcL) of the first signal.

The error keyerr is a cent 'value. Therefore, when there is an error of 1 octave or more, it becomes ± 1024 or more according to the definition of the equation (1). However, the effective pitch error is within one octave, that is, the 1024 remainder of the error. Furthermore, +512 [cent '] and -512 [cent'] are
Actually, the pitch is the same. From the above, lower 1 of keyerr
If 0 bit is treated as a two's complement display, the error in pitch detection will be in the range of ± 512 [cent ']. If the resolution is not so high, the lower 2 bits may be discarded and the lower 3 to 8 bits may be treated as error data. Processing can be simplified when using an 8-bit microcomputer. This error data is output from the comparing means 107.

As described above, in the first embodiment, (1) the two signals are averaged, so that the error can be calculated by eliminating the instantaneous influence such as the timing deviation of the two kinds of songs. (2) Since two types of signals are sampled independently, an accurate average value can be obtained even when the number of detected basic cycles is small when the signal changes slowly.
(3) It is possible to provide a signal comparison device having characteristics such that the time interval for performing the averaging is constant and is suitable for applications such as displaying a pitch difference.

The second embodiment of the present invention will be described below with reference to the drawings. FIG. 4 is a block diagram showing the configuration of the signal comparison apparatus according to the second embodiment of the present invention. In FIG. 4, 401 is a basic cycle input terminal for inputting a basic cycle detection flag of a first signal, 402 is a basic cycle input terminal for inputting a basic cycle detection flag of a second signal,
Reference numeral 403 is a control means for inputting the first and second basic cycle detection flags, 404 is a basic cycle input terminal for inputting the basic cycle of the first signal, and 405 is a first signal according to the control signal of the control means 403. Sampling means for sampling the fundamental cycle of the above, 406 is an averaging means for averaging after converting the sampled first fundamental cycle data into pitch data (cent 'value), and 407 is inputted with the fundamental cycle of the second signal. An input terminal, 408 is sampling means for sampling the fundamental cycle of the second signal in accordance with the control signal of the control means 403, and 409 is the pitch data (cent 'value) after conversion of the fundamental cycle of the sampled second signal. An averaging means for averaging, 410 is a comparing means for comparing the averaged first and second pitch data, and 411 is an output terminal for outputting a comparison result. In the second embodiment, the control means 403,
Sampling means 405, 408, averaging means 406,
409, comparison means 410 to the microcomputer 412
Described below is the case of implementation.

FIG. 5 is a PAD showing the operations of the control means 403 and the sampling means 405, 408. These processes are performed by interruption. Among the variables used in the description, isLR0, scL, scR, and icnt are assumed to be initialized to zero.

The control means 403 inputs the fundamental cycle detection flags of the first and second signals and determines whether pitch detection is possible for both signals (step 1). If the fundamental period of both signals is detected, each fundamental period is converted into a cent 'value (step 2). The cent 'value of the detected first signal is the accumulation variable (scL)
Then, the cent 'value of the second signal is accumulated in the accumulation variable (scR), and the accumulated number counter (isLR0) is incremented (step 3). The basic cycle detection flags (PRL, PRR) are reset (step 4).

If the basic period is not detected, no processing is performed. Interrupt counter (icnt)
Is incremented each time an interrupt occurs, regardless of whether the basic period is detected (step 5).

It is checked whether or not the number of interrupts has reached the reference value (step 6), and if it has, the cumulative result (scL, scR) is stored in the buffer (sscL, sscR) and the cumulative number (is
LR0) is stored in the buffer (isLR1) (step 7) and each variable is cleared (step 8). Then, the pitch data accumulation completion flag (CCNT) is set (step 9).

The above is the contents of the data accumulation process which is processed by the interrupt. Next, the averaging means 406, 4
The averaging process of the accumulation result of 09 will be described. FIG. 6 is a PAD of a process of averaging accumulated results.

The averaging process does not have to be an interrupt process. Normal processing is used to reduce the interrupt load.
When the pitch data accumulation completion flag (CCNT) is set during the loop processing such as the key scan after initialization by the microcomputer, the averaging processing is executed (step 1).

In the averaging process, first, the cumulative number of cent '(isL
It is determined whether R1) is greater than or equal to the reference number (MIN) (step 2). If there is more than the standard number of pitch data, accumulated data (sscL) / accumulated number (isLR1) is average pitch data (avcL), accumulated data (sscR) / accumulated number (isLR1) is average pitch data (avcR) Then, the average completion flags (SRL, SRR) of the first and second signals are set (step 3).

The above is the contents of the averaging process. The processing of the comparison means 410 is similar to that of the first embodiment.

As described above, the signal comparison apparatus according to the second embodiment (1) calculates the error by eliminating the instantaneous effect such as the timing deviation between the two kinds of songs because the two signals are averaged. (2) Since only the data when the fundamental period is detected from two signals at the same time is adopted, an accurate average value can be obtained when the signal changes drastically. (3) Perform averaging It is possible to provide a signal comparison device having a characteristic that the time interval during which the operation is performed becomes constant and is suitable for applications such as displaying the pitch difference.

The third embodiment of the present invention will be described below with reference to the drawings. The configuration of the signal comparison apparatus of the third embodiment is the same as the block diagram of FIG.

FIG. 7 is a PAD showing the operations of the control means 403 and the sampling means 405, 408. These processes are performed by interruption. It should be noted that isL0, isR0, isLR0, scL, scR, and icnt among variables used in the description are initialized to zero.

First, when a timer interrupt occurs, the first
The basic cycle detection flag (PRL) is checked to see if the basic cycle of the signal is detected (step 1). When the basic cycle is detected, the basic cycle number counter (isL0) is incremented (step 2).

If the basic period is not detected, no processing is performed. Similar processing is performed for the second signal (steps 3 and 4).

Next, it is determined whether pitch detection is possible for both signals (step 5). If the fundamental period of both signals is detected, each fundamental period is converted into a cent 'value (step 6). The detected cent 'value of the first signal is accumulated in the accumulation variable (scL), and the cent' value of the second signal is accumulated in the accumulation variable (scR), and the accumulated number counter (isLR0) It is incremented (step 7). The basic cycle detection flags (PRL, PRR) are reset (step 8).

If both basic periods have not been detected, no processing is performed. The interrupt count counter (icnt) is incremented each time an interrupt occurs, regardless of whether or not the basic cycle is detected (step 9).

It is checked whether or not the number of interrupts has reached the reference value (step 10). If the number of interrupts has been reached, the accumulation result (scL, scR) is stored in the buffer (sscL, sscR) and the number of data (icLR0) is stored in the buffer. Store in (isLR1) (Step 1
1) Clear each variable (step 12). And
A pitch data accumulation completion flag (CCNT) is set (step 13).

The above is the contents of the data accumulation processing which is processed by the interrupt. The averaging processing of the accumulation results of the averaging means 406 and 409 and the processing of the comparing means 410 are the same as those in the second embodiment, but the basic cycle detection number of the first and second signals is counted. , Can be used as auxiliary information for comparison results. That is, for example, when the data number isLR1 is less than the required number MIN in step 2 of FIG. 6, it can be known whether the first signal is detected less or the second signal is detected less. For example, the first signal is the user's microphone input voice,
If the second signal is a model singing voice recorded in a karaoke medium, the number of detected first signals is small,
If the signal of is sufficient, it can be considered that the user determines that the part is not singing a song.

As described above, the signal comparison apparatus according to the third embodiment (1) calculates the error by eliminating the instantaneous influence such as the timing deviation between the two types of songs because each of the two signals is averaged. (2) Since only the data when the basic period is detected at the same time is adopted, an accurate average value can be obtained when the signal changes drastically. (3) The time interval for performing the averaging is It is possible to provide a signal comparison device having features such as being constant and suitable for applications such as displaying a pitch difference, and (4) utilizing the number of data of each signal as auxiliary information.

The fourth embodiment of the present invention will be described below with reference to the drawings. The configuration of the signal comparison device of the fourth embodiment is the same as that of the block diagram of FIG.

FIG. 8 is a PAD showing the operations of the control means 403 and the sampling means 405, 408. These processes are performed by interruption. Note that isLR0, scL, and scR among the variables used in the description are initialized to zero.

The control means 403 inputs the fundamental period detection flags of the first and second signals and determines whether pitch detection is possible for both signals (step 1). If the fundamental period of both signals is detected, each fundamental period is converted into a cent 'value (step 2). The cent 'value of the detected first signal is the accumulation variable (scL)
Then, the cent 'value of the second signal is accumulated in the accumulation variable (scR), and the accumulated number counter (isLR0) is incremented (step 3). The basic cycle detection flags (PRL, PRR) are reset (step 4).

If the basic period is not detected, no processing is performed. It is checked whether the accumulated number (isLR0) has reached the reference value (step 5), and if it has reached, the accumulated result (scL, scR) is stored in the buffer (sscL, sscR) (step 6). Clear the variables (step 7). Then, the pitch data accumulation completion flag (CCNT) is set (step 8).

The above is the contents of the data accumulation processing which is processed by the interrupt. Next, the averaging means 406, 4
The averaging process of the accumulation result of 09 will be described. FIG. 9 is a PAD of processing for averaging accumulated results.

The averaging process need not be an interrupt process. Normal processing is used to reduce the interrupt load.
When the pitch data accumulation completion flag (CCNT) is set during the loop processing such as the key scan after initialization by the microcomputer, the averaging processing is executed (step 1).

In the averaging process, first, accumulated data (sscL)
/ Accumulation number (icnt0) is the average pitch data (avcL), accumulation data (sscR) / Accumulation number (icnt0) is the average pitch data (avcR) (step 2), and the average completion flag for the first and second signals Set (SRL, SRR).

The above is the contents of the averaging process. The processing of the comparison means 410 is similar to that of the first embodiment.

As described above, the signal comparison apparatus of the fourth embodiment (1) calculates the error by eliminating the instantaneous influence such as the timing deviation of the two kinds of songs, because the two signals are averaged. (2) Since only the data when the fundamental period is detected at the same time is adopted, an accurate average value can be obtained when the signal changes drastically. (3) The number of data to be averaged becomes constant. It is possible to provide a signal comparison device having features such that the pitch difference is highly reliable and is suitable for applications such as control data of a key controller.

[0063]

As described above, according to the present invention, the characteristic amounts of the respective signals are sampled and averaged according to the characteristic amount detection flags of the plurality of signals, and the comparison is performed to detect the timing deviation of the respective signals. It is possible to provide a signal comparison device having a feature that comparisons can be performed while eliminating instantaneous influences such as.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a signal comparison device according to a first embodiment of the present invention.

FIG. 2 is a PAD diagram of pitch data sampling / accumulation processing in the first embodiment.

FIG. 3 is a PAD diagram of processing for averaging accumulated results in the first embodiment.

FIG. 4 is a block diagram showing a configuration of a signal comparison device according to a second embodiment of the present invention.

FIG. 5 is a PAD diagram of pitch data sampling / accumulation processing in the second embodiment.

FIG. 6 is a PAD diagram of a process of averaging accumulated results in the second embodiment.

FIG. 7 is a PAD diagram of pitch data sampling / accumulation processing in the third embodiment of the present invention.

FIG. 8 is a PAD diagram of pitch data sampling / accumulation processing according to a fourth embodiment of the present invention.

FIG. 9 is a PAD diagram of processing for averaging accumulated results in the fourth embodiment.

[Explanation of symbols]

101, 105, 404, 407 Basic period input terminal 102, 106, 401, 402 Basic period detection flag input terminal 103, 107, 405, 408 Sampling means 104, 108, 406, 409 Averaging means 109, 410 Comparison means 110, 411 Output terminal 111, 412 Microcomputer 403 Control means

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Tatsuhiko Numamoto 1006 Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Industrial Co., Ltd.

Claims (9)

[Claims] 1. A first sampling means for sampling the feature amount of the first signal according to a feature amount detection flag of the first signal; and a first averaging feature amount of the sampled first signal. 1 averaging means, second sampling means for sampling the feature quantity of the second signal according to the feature quantity detection flag of the second signal, and averaging the feature quantity of the sampled second signal A signal comparison device comprising: a second averaging means for performing comparison and a comparison means for comparing the feature amounts of the averaged first and second signals. 2. A control means for outputting a control signal according to the feature amount detection flag of the first signal and the feature amount detection flag of the second signal, and a feature amount of the first signal according to the control signal. Sampling means for sampling the first signal, and a first sampling means for averaging the feature quantities of the sampled first signals.
Averaging means, second sampling means for sampling the characteristic amount of the second signal according to the control signal, and second averaging the characteristic amounts of the sampled second signals.
Signal averaging means and a comparing means for comparing the feature amounts of the averaged first and second signals. 3. The sampling means samples / accumulates the feature quantity when the feature quantity of the signal is detected within a fixed time ts, and the averaging means has the averaging means that the sampling means has a reference number or more within a fixed time ta. The signal comparison device according to claim 1, wherein the averaging process is performed when the characteristic amounts are accumulated. 4. The signal comparison device according to claim 1, wherein the characteristic amount of the signal is the pitch of the signal. 5. The averaging means sets the basic period t corresponding to the pitch to c = A * log ₂ (t0 / t), t0 = 1 / f0, A = 2n (where f0: reference frequency, n: integer). 3. The signal comparison device according to claim 1, wherein the signal is converted into a value c and averaged. 6. The comparing means has a basic period t corresponding to a pitch.
Is c = A * log ₂ (t0 / t), t0 = 1 / f0, A = 2n (where f0: reference frequency, n: integer) and the lower c [bit]
2. The comparison is performed by using as a two's complement.
Alternatively, the signal comparison device described in 2. 7. The first and second sampling means sample and accumulate the respective characteristic amounts when both the characteristic amounts of the first and second signals are detected within a fixed time ts, and the averages are obtained. 3. The signal comparison device according to claim 2, wherein the conversion means performs an averaging process when the first and second sampling means accumulate feature quantities equal to or larger than a reference number within a fixed time ta. 8. The first and second sampling means sample and accumulate the respective characteristic amounts when both the characteristic amounts of the first and second signals are detected within a fixed time ts. The number of times the characteristic quantities of the first and second signals are detected is independently counted, and the first and second averaging means have the first and second sampling means which are equal to or larger than a reference number within a fixed time ta. 3. The averaging process when the feature quantities are accumulated.
The signal comparison device described. 9. The first and second sampling means accumulate the respective characteristic amounts when both of the characteristic amounts of the first and second signals are detected within a fixed time ts, and the first and second sampling means are accumulated. 3. The signal comparison device according to claim 2, wherein the second averaging means outputs the average value when the characteristic amount equal to or larger than the reference number is accumulated.