JP2007093504A - Time signal sound detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a time signal sound detector capable of recognizing both a preannouncement sound and a correct time sound using a very easy simple structure, and also capable of stably detecting a time signal sound with the desired frequency. <P>SOLUTION: The time signal sound detector includes a filter 13 and a microcomputer 14. The filter 13 attenuates tripled frequency or higher frequencies of the preannouncement sound frequency (440 Hz) and tripled frequency or higher frequencies of the correct time sound frequency (880 Hz) from voice signals and then supplies to the microcomputer 14. The microcomputer 14 multiplies, from among time-series sampling data of the pre-announcement sounds of the voice signals, a coefficient of opposite sign to the data appearing at near cycles of the doubled frequency of the preannouncement sound alternately to add; determines the presence of a preannouncement sound frequency component by the variations of the addition; when determining the presence of the preannouncement sound, multiplies a coefficient of an opposite sign to data appearing at near the cycles of a doubled frequency of the correct time sound alternately to add; determines the presence of the correct time sound by the variations in the addition; and when determining the both presence of the preannouncement sound and the correct time sound, a time signal is identified. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a time signal sound detection device that detects a time signal sound of a desired frequency from an audio signal of a broadcast signal, for example.

FIG. 1 is a block diagram showing a configuration example of a conventional time signal sound detection device (see, for example, Patent Document 1).
As shown in FIG. 1, the time signal sound detection apparatus 1 includes a reception antenna 2, a reception circuit 3, a tone decoder 4, and a microcomputer (hereinafter referred to as a microcomputer) 5.

In this hourly sound detection device 1, a broadcast signal is received by the antenna 2 and input to the receiving circuit 3. In the receiving circuit 3, a desired channel is selected, and an audio signal of the channel is supplied to a tone decoder 4 which is a dedicated IC for detecting a single audio frequency.
The tone decoder 4 is set so as to detect a signal having the same frequency as that of, for example, a forecast sound signal. In the tone decoder 4, a forecast sound signal of a time signal is detected, and a detection pulse is supplied to the microcomputer 5.
In the microcomputer 5, it is detected whether or not the received signal is a time signal. Then, the time of the clock unit is corrected based on this detection output.

Further, Patent Document 2 discloses a correct report extraction circuit including a predicted sound detection circuit 6 and a correct report sound detection circuit 7 as shown in FIG.
This correct report extraction circuit is configured to extract a correct report sound accurately with no errors in a circuit that detects a time signal composed of three forecast sounds and one correct report sound.
Japanese Patent Laid-Open No. 7-159560 Japanese Patent Laid-Open No. 6-82572

The time signal sound detection apparatus disclosed in the first patent document 1 uses a tone decoder. The tone decoder is basically composed of a PLL circuit.
A tone decoder including a PLL circuit detects a specific single frequency by comparing a reference signal obtained by dividing the reference frequency of the oscillation signal of the CR oscillation circuit with an input audio signal.
However, since it is difficult to obtain a highly stable oscillation frequency with the CR oscillation circuit, the conventional time signal sound detection device cannot stably detect the time signal sound, and the time signal detection sensitivity. Is disadvantageous.

  In addition, the correct signal extraction circuit disclosed in Patent Document 2 needs to be provided with two detection circuits, that is, a predicted sound detection circuit 6 and a correct sound detection circuit 7, leading to an increase in circuit scale and cost.

  The present invention has been made in view of such circumstances, and its purpose is to detect both a forecast sound and a correct report sound with a very simple configuration and stably detect a report sound of a desired frequency. Another object of the present invention is to provide a time signal sound detection device that can improve detection sensitivity.

  In order to achieve the above object, an aspect of the present invention is a time signal sound detection device for detecting a time signal sound of a desired frequency from a sound signal of a broadcast signal, and among time series sampling data of a sound forecast of the sound signal, If data that appears in the neighborhood of twice the frequency of the predicted sound is multiplied by the inverse sign coefficient alternately, and integrated, the presence or absence of the predicted sound frequency component is determined by the change in the integrated value. The data that appears in the neighborhood of twice the frequency of the report sound is accumulated by alternately applying the opposite sign, and the presence or absence of the correct report sound frequency component is determined by the change in the integration, and it is determined that both the forecast sound and the correct report sound exist. In this case, it has a processing circuit that determines that it is a time signal.

  Preferably, a filter for attenuating both the frequency of three times or more of the predicted sound frequency and the frequency of three or more of the correct sound frequency is provided in the previous stage of the processing circuit.

  Preferably, the processing circuit samples the audio signal at a period of a frequency which is a multiple of the desired frequency and can compensate for a phase difference between the input audio signal and the sampling.

  Preferably, the processing circuit samples the audio signal at a period of four times the frequency.

  Preferably, in the sampling data, the odd-numbered signal processing unit for multiplying the odd-numbered data by alternately applying the inverse-signed coefficient and the even-numbered data for the odd-numbered data are alternately multiplied by the opposite-signed coefficient. And an even-numbered signal processing unit for integrating, and the presence / absence of a desired frequency component is determined based on the integration result of the odd-numbered signal processing unit and the integration result of the even-numbered signal processing.

  Preferably, the processing circuit includes a digital filter.

  According to the present invention, it is possible to recognize both a forecast sound and a correct report sound with a very simple configuration, to stably detect a report time sound of a desired frequency, and to improve detection sensitivity. There are advantages.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 3 is a block diagram showing an embodiment of a time signal sound detection apparatus according to the present invention.

  As shown in FIG. 3, the present time signal sound detection apparatus 10 includes a reception antenna 11, a reception circuit 12, for example, a low-pass filter (hereinafter also simply referred to as a filter) 13, and a microcomputer 14.

  The time signal sound detection apparatus 10 according to the present embodiment detects, for example, a single frequency (880 Hz in the present embodiment) included in a radio time signal.

FIG. 4 is a diagram illustrating an example of a waveform of a time signal according to the present embodiment.
This time signal includes three forecast sound signals (57 seconds, 58 seconds, 59 seconds) of 440 Hz and one correct hour sound (correct sound) signal of 880 Hz.
In the hourly sound detection device 10 of the present embodiment, an 880 Hz hourly sound (correction sound) signal is detected from among these four hourly sound signals.
The length of the forecast sound signal is 100 msec, and the interval between the forecast sound signals is set to 900 msec.

  The receiving circuit 12 receives a broadcast signal received by the receiving antenna 11, selects a desired channel, and outputs an audio signal S12 (f (t)) including a time signal sound signal of the channel to the filter 13.

The filter 13 receives the audio signal from the receiving circuit 12, attenuates a higher frequency (for example, a frequency that is three times or more of 880 Hz) from the audio signal at a desired frequency (880 Hz in the present embodiment), and supplies the attenuated frequency to the microcomputer 14.
As described above, the filter 13 of the present embodiment is configured by a low-pass filter, and has a function of attenuating both a frequency that is three times or more the predicted sound frequency (440 Hz) and a frequency that is three times or more the correct sound frequency (880 Hz). Have
The cutoff frequency of the low-pass filter 13 is set to 880 Hz, for example.

  The microcomputer 14 samples the audio signal by the filter 13 by sampling at a predetermined sampling frequency, specifically, a frequency of a multiple of a desired frequency (880 Hz in the present embodiment) (for example, twice or four times). An analog / digital (A / D) converter 141 that converts a signal into a digital signal, and data that appears in a period close to twice a desired frequency (880 Hz in the present embodiment) among time-series sampling data of an audio signal And a signal processing unit 142 that determines the presence or absence of a desired frequency component (880 Hz) based on the change in the integration.

  First, the signal processing unit 142 of the microcomputer 14 multiplies the data that appears in the vicinity of twice the frequency of the forecast sound by multiplying the data of the forecast sound by multiplying the coefficient of the opposite sign alternately from the time series sampling data of the forecast sound. The presence or absence of the forecast sound frequency component is determined from the change, and if there is a forecast sound, the data appearing in the neighborhood of twice the frequency of the correct report sound is alternately multiplied by the reverse sign, and the change in the integration The presence / absence of the correct report sound frequency component is determined, and when it is determined that both the forecast sound and the correct report sound are present, it is determined as a time signal.

  In this embodiment, the hardware filter 13 and the signal processing (digital filter processing) by the microcomputer 14 are used in combination, so that the time signal sound having high selectivity (Q) with simple hardware and signal processing. The time signal sound detection device 10 that detects both the forecast sound and the correct sound is realized.

  Hereinafter, the signal processing (digital filter processing) by the microcomputer and the hardware filter 13 according to the present embodiment will be described in detail, and a time signal sound having high selectivity (Q) with simple hardware and signal processing. It is proved that the detection device 10 can be realized.

  First, signal processing of the microcomputer 14 will be described.

  Detecting a radio time signal In the time signal sound detection device 10, a filter for detecting a time signal sound frequency (NHK of 880 Hz in this embodiment) is an important factor that determines the performance of the system. Here, a digital filter that can be realized by a clock microcomputer will be described.

<Digital filter (pseudo Fourier transform)>
The audio output signal f (t) of the receiving circuit 12 can be expressed by the following Fourier series.

  This Fourier transform includes a Fourier cosine transform represented by the following equation (2) and a Fourier sine transform represented by equation (3).

  For example, the Fourier cosine transform can be expanded as follows.

When mω _m = ω, only the term a _m COS (mω _x −ω) t becomes a constant a _m , and the integral of the trigonometric function and trigonometric function does not increase no matter how much integration is performed, and can be expressed as follows.

  This extracts a frequency component using Fourier transform. The same applies to the Fourier sine transform.

Here, in f (t) COS (ωt), the multiplication of COS (ωt) means that every other part of the cosine waveform shown in FIG. ,-(Negative) sign and multiplying, this is replaced with a process of multiplying 1 and -1 in a cycle of T / 2.
Multiplying by 1 is nothing other than sampling the data at that point. Sampling means multiplying by a δ function (impulse function).

In the present embodiment, signal processing for sampling data with a period T / 2 and alternately accumulating the data with the opposite sign is mathematically expressed.
The process of sampling is mathematically multiplied by a δ function.

  Regarding the δ function sequence C (t) of the period T, first, consider a pulse train E (t) having a size E and a width τ as shown in FIG.

  The pulse train E (t) can be expressed as a Fourier series as follows.

  Therefore, the Fourier series of the pulse train E (t) can be expressed as follows.

  When the pulse becomes a δ function, the following condition is satisfied by the definition of the δ function.

  Therefore, the Fourier series of the δ function sequence C (t) can be expressed as follows.

FIG. 7 is a diagram for explaining signal processing according to the present embodiment and is a diagram illustrating a function D (t).
The function D (t) in FIG. 7 can be obtained by adding the δ function sequence C (t) and C (t) shifted by T / 2 and adding a sign. It can be expressed as

  In this equation (10), when n is an even number, the preceding term and the following term cancel each other out, so that the function D (t) can be expressed as follows.

  Therefore, the process of multiplying the audio signal f (t) by multiplying by the function D (t) means the following.

In Expression (12), F _C (ω) is a Fourier cosine transform of the audio signal f (t) as follows.

As can be seen from equation (12), when the process of multiplying the audio signal f (t) by multiplying by the function D (t) is performed, in addition to the frequency ω, the frequency components of 3ω, 5ω, 7ω. Appears at level.
Therefore, harmonics are picked up unless the filter 13 for cutting frequency components of 3Ω or higher is inserted in the audio output of the receiving circuit 12 (input stage of the microcomputer 14).
Conversely, if the filter 13 is inserted, the same effect as the normal Fourier transform can be obtained. Moreover, it can be said that sampling with higher density than this is meaningless.

  As signal processing of the microcomputer 14, assuming that the audio frequency to be detected is f, as shown in FIG. 8, the A / D converter 141 samples data at a period of twice the frequency 2f, and the data is The signal processing unit 142 may add and add positive and negative signs alternately.

Up to this point, the phase problem has not been mentioned, but in reality it is necessary to consider the phase difference between the period of the input signal and the sampling timing.
The detection level changes depending on the phase. When the phase difference is π / 2, the detection level becomes zero. Therefore, similar signal processing with a phase shift of π / 2 is required.

Accordingly, in the present embodiment, as shown in FIG. 9, data is sampled by the A / D converter 141 at a period of 4 times the frequency 4f, and the odd side signal is compared with the odd side data among the data. The processing unit 142A adds and adds positive and negative signs alternately, and the even-numbered signal processing unit 142B adds and adds positive and negative signs alternately to the even-numbered data, and adds the odd-numbered signal processing unit. It is necessary to make a determination based on a change in the maximum value MAX (A, B) or (A ² + B ² ) ^1/2 of the integration result A of 142A and the integration result B of the even-numbered signal processing unit 142B.
The Q of this digital filter increases as the sampling data increases, so that n is infinite and Q is infinite.

  Here, a case where a simulation is actually performed on a time signal of radio broadcast of NHK (Japan Broadcasting Corporation) will be considered.

FIG. 10 is a diagram showing a simulation result of the filter characteristics of the 880 Hz digital filter.
In FIG. 10, the horizontal axis indicates the frequency, and the vertical axis indicates the output level.

The same waveform as the waveform near 880 Hz shown in FIG. 10 also appears at 2640 (880 × 3) Hz, 4400 (880 × 5) Hz,.
In order to set the sampling period to an integer multiple of 880 Hz, the clock of the microcomputer 14 needs to be an integer multiple of 880 Hz. In order to realize this, a special vibrator must be used. A 2 MHz clock was used.
When the integration time is long, a slight frequency shift becomes a problem, but when the integration time is short, a nearby frequency is sufficient.

As conditions for the simulation of the digital filter, the clock frequency of the microcomputer is 2 MHz, the data sampling rate is 3.521126 kHz (2000000/568 = 4 × 880.28), and the sampling time is 0.1 second (sec).
As a result of the simulation, a Q value of about 100 is obtained with 0.1 second sampling.

  Next, since the frequency 440 Hz of the forecast sound is ½ of the frequency 880 Hz of the correct report sound, the data sampling rate may be halved to perform exactly the same processing.

FIG. 11 is a diagram showing a simulation result of the filter characteristics of the 440 Hz digital filter.
In FIG. 11, the horizontal axis indicates the frequency, and the vertical axis indicates the output level.
Also in this case, similar components appear at 1320 (440 × 3) Hz, 2200 (440 × 5) Hz,.

As conditions for the simulation of the digital filter, the clock frequency of the microcomputer is 2 MHz, the data sampling rate is 1.760563 kHz (2000000/1136 = 4 × 440.14), and the sampling time is 0.1 second (sec).
In this case, since the number of samples is half of 880 Hz, as a result of simulation, a Q value of about 50 can be obtained by sampling for 0.1 seconds.

Accordingly, when the predicted sound frequency is 440 Hz and the correct sound frequency is 880 Hz, the low-pass filter 13 having a cut-off frequency of 880 Hz is disposed at the front stage of the digital filter (the front stage of the microcomputer 14 in this embodiment) as in this embodiment. By providing this, it is possible to remove both the frequency three times or more of 440 Hz and the frequency three times or more of 880 Hz.
That is, 440 Hz and 880 Hz can be discriminated (discriminated) only by changing the data sampling frequency while keeping the low-pass filter fixed.
That is, with a simple configuration, it is possible to discriminate between the predicted sound of the time signal and the correct sound only by changing the sampling frequency of the data.

In the time signal sound detection apparatus 10 having the above configuration, for example, a broadcast signal transmitted from a radio broadcast station is input to a reception circuit 12 via a reception antenna 11.
In the receiving circuit 12, a desired channel is selected from the received broadcast signal, and an audio signal S 12 (f (t)) including a time signal sound signal of the channel is output to the filter 13. The filter 13 receives the audio signal from the receiving circuit 13 and attenuates a frequency component more than 3 times the predicted sound frequency (440 Hz) and a frequency component more than 3 times the correct sound frequency (880 Hz) from the audio signal. To be supplied.
In the microcomputer 14, first, among the time series sampling data of the forecast sound in the voice signal, the data appearing in the vicinity period twice as high as the frequency of the forecast sound is alternately multiplied and multiplied by the inverse sign coefficient. The presence / absence of the predicted sound frequency component is determined based on the change in integration.
Here, if it is determined that there is a forecast sound, the data appearing in the neighborhood period twice as high as the frequency of the correct alarm sound is alternately multiplied by the opposite sign, and the presence or absence of the correct alarm sound frequency component is determined by the change in the integration. Is judged. When it is determined that both the forecast sound and the correct report sound are present, it is determined as a time signal.
If the microcomputer 14 determines that it is a time signal, the microcomputer 14 performs time correction processing of an internal clock (not shown), for example, assuming that the hour signal has been detected.

  As described above, according to the present embodiment, the broadcast signal received by the receiving antenna 11 is input, the desired channel is selected, and the audio signal S12 (f (t)) including the time signal sound signal of the channel is selected. ) And a voice signal from the receiver circuit 12, and from the voice signal, a frequency that is at least three times the predicted sound frequency (440 Hz) and a frequency that is at least three times the correct sound frequency (880 Hz) The filter 13 that is attenuated and supplied to the microcomputer 14 and the time-series sampling data of the predicted sound of the audio signal, multiply the data that appears in the vicinity of twice the frequency of the predicted sound by alternately applying the coefficient of the opposite sign. Then, the presence or absence of the forecast sound frequency component is judged from the change in the integrated value, and if it is judged that the forecast sound is present, the opposite sign is alternately applied to the data appearing in the neighborhood cycle twice the frequency of the correct report sound. And the microcomputer 14 that determines the presence of the forecast sound and the correct report sound when it is determined that both the forecast sound and the correct report sound are present. Obtainable.

That is, according to this embodiment, it is possible to distinguish between 440 Hz and 880 Hz simply by changing the data sampling frequency while the low-pass filter is fixed.
That is, with a simple configuration, it is possible to discriminate between the predicted sound of the time signal and the correct sound only by changing the sampling frequency of the data.
Therefore, the circuit can be simplified, and the detection sensitivity of the time signal sound can be greatly improved.

It is a block diagram which shows the structural example of the conventional time signal sound detection apparatus. It is a block diagram which shows the example of a principal part structure of the conventional correct report extraction circuit. It is a block diagram which shows one Embodiment of the time signal sound detection apparatus which concerns on this invention. It is a figure which shows an example of the waveform of the time signal which concerns on this embodiment. It is a figure for demonstrating the signal processing which concerns on this embodiment, Comprising: It is a figure which shows a cosine waveform, Comprising: It is a figure for demonstrating the meaning which multiplies COS ((omega) t) in Fourier cosine transformation. It is a figure for demonstrating the signal processing which concerns on this embodiment, Comprising: It is a figure which shows the pulse train E (t) of the magnitude | size E and the width | variety (tau). It is a figure for demonstrating the signal processing which concerns on this embodiment, Comprising: It is a figure which shows the function D (t). It is a figure for demonstrating the basic signal processing of the microcomputer which concerns on this embodiment. It is a figure for demonstrating the suitable signal processing of the microcomputer which concerns on this embodiment. It is a figure which shows the simulation result of the filter characteristic of the 880Hz digital filter which concerns on this embodiment. It is a figure which shows the simulation result of the filter characteristic of the 440Hz digital filter which concerns on this embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Time signal sound detection apparatus, 11 ... Reception antenna, 12 ... Reception circuit, 13 ... Filter (low-pass filter), 14 ... Microcomputer (microcomputer), 141 ... Analog / digital (A / D) converter, 142, 142A, 142B: Signal processing unit.

Claims (6)

A time signal sound detection device for detecting a time signal sound of a desired frequency from an audio signal of a broadcast signal,
Of the time-series sampling data of the forecast sound of the sound signal, the data appearing in the vicinity of twice the frequency of the forecast sound is multiplied by the inverse sign coefficient alternately, and the forecast sound frequency component If it is determined whether there is a forecast sound, the data that appears in the vicinity of twice the frequency of the correct sound is multiplied by the opposite sign alternately, and the presence of the correct sound frequency component is determined by the change in the total. A time signal sound detection device having a processing circuit that determines a time signal when it is determined that both the forecast sound and the correct sound sound are present. The time signal sound detection device according to claim 1, further comprising a filter that attenuates both a frequency that is three times or more of a predicted sound frequency and a frequency that is three times or more of a normal sound frequency before the processing circuit. The time signal sound detection device according to claim 1, wherein the processing circuit samples the audio signal at a period of a frequency which is a multiple of the desired frequency and can compensate for a phase difference between the input audio signal and the sampling. The time signal sound detection device according to claim 3, wherein the processing circuit samples an audio signal at a period of four times the frequency. The processing circuit integrates the odd-numbered signal processing unit that alternately adds the inverse-sign coefficient to the odd-numbered data in the sampling data and the odd-numbered data and alternately applies the inverse-sign coefficient to the even-side data. An even-numbered signal processing unit,
The time signal sound detection device according to claim 3 or 4, wherein the presence / absence of a desired frequency component is determined based on an integration result of the odd-numbered signal processing unit and an integration result of the even-numbered signal processing. The time signal sound detection device according to any one of claims 1 to 4, wherein the processing circuit includes a digital filter.

Images