JP2856214B2 - Voice secret device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] The present invention relates to an improvement of a voice confidential communication apparatus that performs frequency division by performing digital signal processing after thinning out an input sampling signal to obtain a low-speed sampling signal. The purpose of this invention is to divide a predetermined frequency band by cutting out a voice band every other band, thereby reducing noise caused by overlap with adjacent filter characteristics, and providing a voice confidential device capable of ensuring good decoded voice quality. When the number is n, the thinning interval of the input sampling signal is set to n and thinning is performed every nth sampling, and the sampling frequency of the input sampling signal is 1 / n.
, Decimating so as to output n sampled low-speed signals at the frequency of, and converting each of the n low-speed sampling signals into a first complex polyphase digital filter unit;
Using a first phase shift unit and a first inverse fast Fourier transformer, a small band is divided so that a small band signal obtained by n division is converted into a complex signal for every other small band. Dividing means, transposing means for replacing the frequency positions of n complex signals obtained at the output of the small band signal dividing means in accordance with a predetermined encryption rule, and transposed n complex signals obtained at the output of the transposing means Each of the signals
A second inverse fast Fourier transformer, a second phase shifter, a second complex polyphase digital filter, and an interpolating means for interpolating each of the signals obtained at the output of the complex polyphase digital filter. frequency band synthesizing means for synthesizing each of the n-divided small band signals.

[Industrial applications]

The present invention relates to an improvement of a voice confidential device for converting an analog voice signal into a confidential signal, and more particularly to an improvement of a voice confidential device for converting an analog voice signal into a digital signal by performing digital signal processing and substituting a frequency band for permutation.

That is, the present invention relates to a communication esoteric device, and in particular,
The present invention relates to a frequency band division of an input sampling signal which is performed by a digital filter and a thinning-out process, the band signals are exchanged, and the respective bands are synthesized to improve a signal-to-distortion ratio.

[Conventional technology]

FIG. 9 is a block diagram of a conventional voice secret device. This conventional apparatus is disclosed in Japanese Unexamined Patent Publication No.
No. 134524.

In a voice confidential system in which a signal of a predetermined frequency band including a voice band is divided and replaced to be polarized, a predetermined frequency band including a voice band is divided into 32 units (generally n divisions).

In FIG. 9, a conventional voice confidential device includes a decimation processing unit 91 for decimating an input sampling signal to 1/64 (generally 1 / 2n) in order to convert a voice signal into a complex signal of each of n frequency bands, The elements 92-1 to 92-63, the small band signal division unit 900 including the polyphase filters 93-0 to 93-63 and the inverse fast Fourier transformer 94, and 32 points for changing the frequency band of the complex signal (generally n
Point) transposition unit 95, an inverse fast Fourier transformer 96, a polyphase filter that extracts signals in each frequency band from each transposed complex signal
Each frequency band signal synthesizing means 901 comprises 97-0 to 97-63, delay elements 98-1 to 98-63, and an interpolation section 98 for synthesizing each sampling signal.

The input sampling signal is thinned out to 1 / 2n by the thinning-out processing section 91 to become 2n low-speed sampling signals. The frequency band of the complex signal output from the small band signal dividing means 900 is replaced by the transposition means 95, and the frequency band synthesizing means
By 901, the signals of the respective frequency bands after the replacement are combined.

[Problems to be solved by the invention]

FIG. 10 shows a complex signal output unit in the conventional device of FIG.
9 is a waveform diagram illustrating a thinning operation performed by a thinning processing unit and a thinning processing unit. FIG.

In the figure, (1) represents the spectrum of the input audio signal, which is a folded signal in a band of 4 KHz. Consider the case where the 4KHz band is divided into 32. Shown in (2)
Ho (Z) is the frequency characteristic of the polyphase reference filter, and is a digital filter that passes the small band 0. Hi (Z) shown in (3) is a digital filter that passes a small band i. (4) shows an output signal when the input sampling signal is filtered at the frequency using the digital filter having the characteristic shown in (3).

As shown in FIGS. 10 (2) and (3), the frequency characteristic of the actual digital filter does not show the rising / falling in a completely rectangular shape in a required small band like an ideal filter. Rise / fall is shown in an inclined manner. For this reason, the signal shown in FIG. 10 (5) obtained by thinning out the small band signal at 1 / 2n sample period has an overlapped portion shown by oblique lines.

Therefore, when the signals shown in FIG. 10 (5) obtained by thinning out the small band signals are combined using digital filters having the same characteristics, there is a problem that noise due to the above-mentioned hatched portions remains.

SUMMARY OF THE INVENTION An object of the present invention is to provide a voice confidential speech device capable of reducing noise due to overlapping portions of adjacent bands of a decimated signal and ensuring good decoded speech quality by setting the decimating cycle to 1 / n.

[Means for solving the problem]

FIG. 1 is a block diagram showing the principle of the present invention. In the figure, an input signal X (Z) is a signal of a predetermined frequency band including an audio signal, and this predetermined band is divided into n and given. 1
Is a thinning-out part that thins out the input sampling signal every nth sampling interval, and outputs n low-speed sampling signals at a frequency of 1 / n of the sampling frequency of the input sampling signal. Each of the signals is converted into a first complex polyphase digital filter unit 2-0 to 2- (n-1), a first phase shift unit 3-0 to
3- (n-1) and small band complex signal output means for converting into a complex signal of any of small bands divided into n using the first inverse fast Fourier transformer 4, and 5 is a small band complex signal output. Transposition means for transposing the output positions of the n complex signals obtained at the output of the means 21 in accordance with a predetermined encryption rule. 31 transposes each of the n transposed complex signals obtained at the output of the transposition means 5. , A second inverse fast Fourier transformer 6,
Second phase shift units 7-0 to 7- (n-1) and second complex polyphase digital filter units 8-0 to 8-
This is a frequency band signal synthesizing means for synthesizing n-divided small band signals using (n-1) and an interpolation part for synthesizing each sampling signal.

(Operation)

FIG. 2 is an explanatory diagram of the operation of the present invention. In the figure, (a) is a diagram showing a predetermined band of the input signal X (Z) and a band obtained by folding the predetermined band at the Nyquist frequency. , 1,2,
... n-1. The small band of the inversion band is
… Indicated by ▲ ▼. (B), (c), (d)
Are the first polyphase digital filters (2
It is a figure which shows the frequency characteristic of the filter comprised by the polyphase sub-filter of -0,2-1, and 2- (n-1) and the phase shift parts 3-0, -3- (n-1). However, the second polyphase digital filter includes a polyphase subfilter 2-0, 2-1,... 2- (n-1) and a phase shift portion 3-0, 3-1,. And inverse FFT 1, W
^-1 , W ^-2 , W- ^(n-1) . Similarly, the n-th
The polyphase digital filter is composed of a polyphase subfilter 2-0,2-1,... 2- (n-1), a phase shift portion 3-0,3-1,. , W
^-(n-1) , W ^{-2 (n-1)} ... W- ^{(n-1) (n-1)} .

(E), (f), and (g) are output signals of the first, second, and n-th polyphase digital filters, respectively. As is apparent from the figure, the filter characteristic for cutting out the small band is cut out every other number of samples equal to the number of band divisions so as to cut out every other small band. Even if the rising and falling are not steep, noise due to overlapping of the complex signals in adjacent bands does not occur.

〔Example〕

FIG. 3 is a principle diagram of an embodiment of the present invention. In the figure, the principle of the 32-part replacement will be described.

Also in the present invention, similarly to the invention described in Japanese Patent Application No. 63-134524, attention was paid to a conventionally known T-MUX technology. The T-MUX technology is used in the field of communication systems applied to telephone services for mutual conversion between frequency-multiplexed signals and time-division multiplexed signals. (See, for example, “Application of Digital Signal Processing,” edited by the Institute of Electronics, Communication and Communication Engineers of Japan, published on July 10, 1983, 3rd edition, pages 121 to 134).

In T-MUX, TDM-FDM converter or FDM-TDM
A converter can be used to perform mutual conversion between a time division multiplexed (TDM) signal and a frequency division multiplexed (FDM) signal. For example, in FDM-TDM, SSB conversion is inserted in the middle and FDM-TDM is used.
This is realized by the configuration of SSB and SSB-TDM.

In FIG. 3, an audio input X (Z) is regarded as n small-band frequency-multiplexed signals, a decimation process is performed by an FDM / SSB conversion unit 30, and a 32-point transposition unit 31 outputs an FDM / SSB conversion unit. Is transposed, and the SSB / FDM conversion unit 32 performs an interpolation process, which is the reverse operation of the thinning process, to obtain a secret output Z (Z). With this configuration, in the FDM / SSB converter 30, each of the 32 small bands of the input signal is converted into a digital filter (H ⁰ ) 33.
−0 to (H ³¹ ) 33−31 to separate the complex signal Y ⁰
(Z) to Y ³¹ (Z) are obtained, and the 32: 1 thinning processing units 34-0 to 3-3 are obtained.
The complex signals Y ₀ (Z) to Y ₃₁ (Z) are obtained by thinning one out of the 32 consecutive input sample signals in 4-31. The order of the complex signals Y ₀ (Z) to Y ₃₁ (Z) is based on the 32-point transposition unit 31.
And is input to the SSB / FDM conversion unit 32. SSB
In the / FDM conversion unit 32, the reverse operation of the FDM / SSB conversion unit 30
Get secret story output.

As a specific configuration example of the SSB-FDM conversion unit 30,
A so-called 4KHz sample α type realized by Bellanger by the combination of FFT and polyphase filter is known. The SSB-FDM converter described in Japanese Patent Application No. 63-134524 utilizes this 4 KHz sample α type. In the present invention, an improved 8KHz β type signal processing system is used.

The SSB signal used in the T-MUX is a single sideband signal (single-sideband signal) in the field of analog modulation, and must be distinguished from a scrambler using the T-MUX. That is, SS in the present invention
The B signal corresponds to a repetition signal of a small band signal obtained by dividing the input audio signal.

FIG. 4 shows a frequency characteristic of a complex polyphase digital filter and a complex signal Y ₀ (Z) in the case of performing a 4-part replacement for simplicity instead of the 32-part replacement in the apparatus of FIG.
~Y shows ₃ (Z).

In the figure, each sample of the input sampling sequence X (Z) is represented by a spectrum having an audio band of 0 to 4 KHz as shown in (1). (2) shows the frequency characteristic of the polyphase reference filter. (3) to (6) reverse the characteristics of the polyphase reference filter to those of the phase shift unit.
This is a polyphase filter for each small band created using the phase term of the FFT unit. Digital filter H ⁰ (Z) to H ³ from each filter is input to the (Z), a small band 0,2 of the input audio signal having the frequency characteristic shown in input sampling signal (3) to (6), , Are output. Next, when the sampling sequence of the output of these small band signals is thinned out every four samples, a complex signal which repeats the frequency component every other small band in each channel is obtained as shown in (7) to (10). Can be Since the decimated signal provides a complex signal every other small bandwidth, adjacent channels do not overlap in each channel of the complex signal, and no noise is generated due to overlapping bands.

 Here, consider the FDM-SSB conversion unit 30 in FIG.

To obtain a complex signal Y ₀ to Y ₃ illustrated in FIG. 4 (7) to (10), and using the FIR filter 17 tap as polyphase reference filter. The filter of the frequency characteristic H (Z) is expressed as _{H (Z) = h 0 +} h 1 Z -1 + h 2 Z -2 + ... + h 16 Z -16 ... (1). This is called a polyphase reference filter characteristic. The complex signal Y ₀ is obtained by a filter having a frequency characteristic obtained by shifting the above-mentioned reference filter characteristic by 小 small band. This is because the i-th term (i = 0,1,2,
…, 16) Is obtained by multiplying Therefore, the frequency characteristic of the filter corresponding to channel 0 for obtaining the complex signal Y ₀ (Z)
H ₀ (Z) is It is expressed as

Similarly, the complex signal Y ₁ (Z) is obtained by a filter corresponding to channel 1 having a frequency characteristic obtained by shifting the reference filter characteristic by 2.5 bands. This is further added to the terms of the above equation (2). Therefore, the frequency characteristic H ₁ (Z) of the filter for obtaining the complex signal Y ₁ (Z) is H ₁ (Z) = h ₀ + h ₁ E ^-1 W ^-1 Z ^-1 + h ^{_{2 E -2 W -2 Z -2 +}} ... h 16 E -16 W -16 Z -16 ... (3) = h 0 + h 4 E -4 W -4 Z -4 + ... h 8 E -8 W - ⁸ Z ^-8 + h ₁₂ E ^-12 W ^-12 Z ^-12 + h ₁₆ E ^-16 W ^-16 Z
^-16 + h ₁ E ^-1 W ^-1 Z ^-1 +… h ₁₃ E ^-13 W ^-13 Z ^-13 + h ₂ E ^-2 W ^-2 Z ^-2 +… h ₁₄ E ^-14 W ^-14 Z ^-14 + H ₃ E ^-3 W ^-3 Z ^-3 +… h ₁₅ E ^-15 W ^-15 Z ^-15 It is expressed as

Similarly, the frequency characteristic of the filter for obtaining the complex signal Y ₂ (Z) is further added to each term on the right side of the above equation (3). , And the frequency characteristic of the filter for obtaining the complex signal Y ₃ (Z) is further expressed by the equation of the frequency characteristic of H ₂ (Z). Is obtained by multiplying

Focusing on the portions of the frequency characteristics H ₀ (Z) and H ₁ (Z) enclosed by the rectangle on the right side, it can be seen that these signal processing configurations can be realized by the configuration shown in FIG.

FIG. 5 shows a β-type T-MUX constructed based on the above-mentioned invention.
It is a block diagram which shows the FDM / SSB conversion part in (4 division). In the figure, a complex polyphase sub-filter (S ₀ ) 2-0 to (S ₃ ) 2-3 is a polyphase comprising a delay element Z- ⁴ for delaying four samples, a coefficient unit and an adder. It is a sub-filter. Filter (S ₀ ) 2
At −0, four delay elements Z ^-4 and coefficients h ₀ , jh ₄ , −h ₈ , −j
It has five coefficient units having h ₄ and h ₀ , passes the input signal IN through this filter, and the real part Re = h ₀ −h ₈ Z ⁻⁸ + h ₀ Z ^-16 and the imaginary part Im = jh ₄ Z ^-4 −jh ₄ Z ^-12
4 points IF through phase shift unit E ⁰ for shifting by band
Input to FT51. In the complex polyphase subfilter (S ₁ ) 2-1, three delay elements Z ^-4 and coefficients h ₁ , jh ₅ , −
h ₇ , −jh ₃ are provided, and the input signal IN
Is passed through this filter, and the real part Re = Z ^-1 (h ₁ −h ₇ Z ^-8 ) and the imaginary part Im = Z ⁻¹ (jh ₅ Z ^-4 −jh ₃ Z ^-12 ) are separately added, Phase shift unit E ^{-1 to} shift this by 1/2 of the small band
And input to the 4-point IFFT51. In the complex polyphase sub-filter (S ₂ ) 2-2, the input signal IN is similarly passed through this filter, and Re = Z ⁻² (h ₂ −h ₆ Z ⁻⁸ ) and Im = Z ⁻²
(Jh ₆ Z ⁻⁴ −jh ₂ Z ⁻¹² ) is calculated, shifted by half of the small band by the phase shift unit E- ² , and input to the four-point IFFT 51. Similarly, the complex polyphase sub-filter (H ₃ ) 2-3 also filters the input signal IN to obtain Re = Z ^-3 (h ₃ −h ₅ Z
⁻⁸ ) and Im = Z ⁻³ (jh ₇ −jh ₁ Z ⁻¹² ), and then the signal is shifted by half of the small band in the phase shift unit E- ³ and input to the IFFT 51.

The four-point IFFT 51 performs an inverse fast Fourier transform on the input complex signals of the four channels to perform the phase shift of W− ⁱ , and also outputs the complex output signals Y ₀ (Z ⁴ ) to Y _{0.
Get 3} (Z ⁴ ). These are the channels ch0, ch2, ▲
Signals corresponding to ▼ and ▲ ▼. That is, signals of subbands 0, 2, and are obtained at the output of IFFT 51.

Of the portion enclosed by the rectangle in the above equation (2), the equation in the first row is the polyphase sub-filter (S ₀ ) in the configuration of FIG.
Represents the structure of the 2-0, wherein the second row represents the structure of the polyphase sub-filter S ₁ and the phase shift unit E ^-1, 3
Expression lines polyphase sub-filter S ₂ and the phase shift unit
It represents the structure of E ^-2, wherein the fourth row represents the structure of the polyphase sub-filter S ₃ and the phase shift unit E ^-3.

Similarly, each row of the portion surrounded by the rectangle in the above equation (3) has the same configuration as that of the equation (2) up to the phase shift units E ^{0 to} E ^-3 , but W ⁻¹ , W ⁻² , W ^−. The multiplication of ³ is performed in the 4-point IFFT51 and 2
Performs band shift operation.

Based on the above consideration of FIGS. 3 to 5,
Next, examples of the present invention will be described.

The present invention uses the FDM-SSB conversion transposition processing of FIG.
Focusing on the fact that a confidential device can be configured by the configuration of the conversion,
The configuration shown in FIG. 6 was obtained.

Here, in the configuration of FIG. 3, the calculation amount handled by the digital filter is calculated based on the entire voice band, and the calculation amount and the speed are not improved.

Therefore, in the preferred embodiment of the present invention, as shown in FIG. 6, the thinning-out processes 34-0 to 34-31 and the filter operation processes 33-0 to 33-31 in FIG. 3 are interchanged. As a result, the amount of calculation handled by each filter is 1/32 in this example, and the total amount of calculation can be reduced to 1/32 of the configuration shown in FIG.
Further, by performing the thinning processing first, the calculation speed can be reduced to a large range (1/32). This is similar to the α-type.

FIG. 6 is a block diagram of a voice confidential communication device according to an embodiment of the present invention. In the figure, an input signal X (Z) is a signal of a frequency band of 4 KHz including a voice band, and
Think separately. The thinning-out processing unit 61 thins out the input sampling signal in the reverse order every 32 bits with the thinning interval set to 32, and 1/32 of the sampling frequency of the input sampling signal.
And outputs 32 low-speed sampling signals of the frequency. The small-band complex signal dividing unit 71 converts each of the 32 low-speed sampling signals into a delay element 72-1 to 72-31, a complex polyphase sub-filter unit 73-0 to 73-31, and a phase shift unit 74-0.
Using 74-31 and 32 point inverse fast Fourier transformer 75, 32
The signal is converted into any of the divided small-band complex signals. The 32-point transposing means 76 replaces the frequency arrangement of the 32 complex signals obtained at the output of the small-band complex signal dividing means 71 in accordance with a predetermined encryption rule. The frequency band signal synthesizing means 81 converts each of the 32 transposed complex signals obtained by the 32-point transposition processing section 76 into a 32-point inverse fast Fourier transformer 82, a phase shift section 83-0 to 83-31 complex poly. Phase digital filter units 85-0 to 85-31 and delay element 85-
Using 1-85-31 and an interpolation processing unit that synthesizes each signal,
Signals in any of the n-divided small bands are synthesized.

The delay element (Z ^-1 ) 72-1 indicates a delay of one sample, and (Z ^-31 ) 72-31 indicates a delay of 31 samples.

In the apparatus shown in FIG. 6, the phase shift unit (E ⁰ ) 74-0 to (E ⁻³¹ )
The operation from 74 to 31 is as follows.

Each of the low-speed sampling signals decimated by the decimating processing unit 61 in the reverse order of the input is a delay element (Z ^-1 ) 72-1 to (Z
^-31 ) Delay is provided by 72-31. Delay element (Z- ⁱ )
72-i (i = 1, 2,..., 31) is a signal that coincides on the time axis to provide a delay of i samples.

Complex polyphase digital filters (H ₀₎ 73-0~
(H ₃₁ ) 73-31 are filters for cutting out the small bands 0 to 31 of the input signal, respectively.
It is shown in FIG.

FIG. 7A shows the configuration of a filter (H ₀ ) 73-0 when a 128-tap filter is used as a complex polyphase digital filter and the number of divisions is 32.
From the figure, the real part Re = h ₀ −h ₆₂ R ^-64 and the imaginary part Im = jh ₃₂ Z ^-32
-Jh ₉₆ Z- ⁹⁶ is output. In FIG. 6, this complex output is indicated by a double line. Structure of FIG. 7 (a) is a further description since H ⁰ and similarity shown in FIG. 5 is omitted. Other filter (H ₁₎ 73-1~
(H ₃₁ ) 73-31 can also be realized by a configuration similar to the configuration shown in FIG.

Each of the phase shift units (E ⁰ ) 74-0 to (E ⁻³¹ ) 74-31 shifts the frequency characteristic of the polyphase reference filter by 0.5 band. Then, by subjecting these phase-shifted signals to inverse FFT processing by the inverse fast Fourier transformer 75, the signals of each small band are first obtained in ascending order of the signals of the even bands, and then the signals of the odd bands are inverted on the frequency axis. Obtained in descending order by type. These signals are transposed, input to a polyphase sub-filter having the same characteristics, and subjected to interpolation processing to obtain a secret signal. This signal processing method is called a 250 Hz sample β type signal processing method.

As is clear from the above description, the filter characteristic for cutting out the audio band is configured to cut out every other band, and the thinning cycle is performed for each sample number equal to the band division number. Therefore, even if the filter characteristics are not steep, noise does not occur due to overlapping of complex signals in adjacent bands.

The configuration of the complex polyphase sub-filter 84-0 included in the frequency band signal synthesizer 81 is shown in FIG. 7 (b). As is clear from the figure, the input signal of this filter is a complex signal, and its real part is inverted from h ₀ -h ₆₄ Z ^-64 , and its imaginary part is inverted from jh ₃₂ Z ^-32 -jh ₉₆ Z ^-96 Are output, and these are added and output as a real number signal. The configuration of the other filters (H ₁ ) 84-1 to (H ₃₁ ) 84-31 can also be inferred from the above description, and thus the description is omitted.

With the above configuration and operation, the signal-to-distortion ratio (SDR) of the output signal from the interpolation unit 86 is improved in sound quality by about 5 to 8 dB as compared with the conventional α-type T-MUX confidential system as shown in FIG. doing. It also increases as the number of taps of the complex polyphase digital filter increases.

〔The invention's effect〕

As is apparent from the above description, according to the present invention, the filter characteristic for cutting out the small band is configured to cut out every other small band, and the cycle of performing the thinning-out is set for each sample number equal to the band division number. Therefore, in a voice confidential speech device that performs digital signal processing and performs frequency division permutation, noise due to overlapping of complex signals in adjacent small bands can be greatly reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram of the principle of the present invention, FIG. 2 is an explanatory diagram of the operation of the present invention, FIG. 3 is a principle diagram of an embodiment of the present invention, and FIG. 4 is a principle of frequency division in the embodiment of the present invention. FIG. 5 is a diagram showing a configuration of an FDM / SSB conversion unit in a β-type T-MUX applied to an embodiment of the present invention. FIG. 6 is a block diagram showing a voice secret device according to an embodiment of the present invention. FIG. 7, FIG. 7 is a diagram showing a configuration example of a complex polyphase digital filter applied to the embodiment of the present invention, FIG. 8 is a graph showing an improvement in signal versus distortion ratio according to the embodiment of the present invention, FIG. FIG. 10 is a block diagram showing a conventional voice secret device, and FIG. 10 is a diagram showing a problem in the device of FIG. In the figure, 1... A thinning-out unit, 2-0 to 2- (n-1)... A complex polyphase digital filter, 3-0 to 3- (n-1). ... A first inverse fast Fourier transformer, 5... Transposition means, 6... A second inverse fast Fourier transformer, 7-0 to 7- (n−1). 0 to 8- (n-1): second complex polyphase digital filter, 9: interpolator, 21: small band complex signal dividing means, 31: frequency band signal synthesizing means.

──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Naoya Torii 1015 Uedanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Inside Fujitsu Limited (72) Ryota Akiyama 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Fujitsu Limited (72) 56) References JP-A-1-302931 (JP, A) JP-A-63-124637 (JP, A) Mitsuhiro Higashi, Fumio Amano, Ryota Akiyama “Study on Speech Secretion System Using Transmultiplexer Technology” 1989 IEICE Spring National Convention Lecture Papers, Volume 1, p. 1-276 (Accepted by the Patent Office Archives on March 20, 1999) Mitsuhiro Higashi, Takayuki Hasebe, Naoya Torii, Ryota Akiyama “Study on the confidential speech system using TMUX technology” IEICE Transactions, Vol. J73-A, No. 7, p. 1261 -1269 (July 25, 1990) (58) Fields investigated (Int. Cl. ⁶ , DB name) H04K 1/00-3/00 H04L 9/00-9/38 G09C 1/00-5 / 00 JICST file (JOIS)

Claims (2)

(57) [Claims] An audio privacy apparatus for dividing and replacing a signal of a predetermined frequency band including a voice band to convert the signal into a polarized signal, wherein when the number of divisions of the predetermined frequency band is n, the thinning interval of the input sampling signal is reduced. n is decimated every n number of times, and is 1/1 of the sampling frequency of the input sampling signal.
Decimation is performed so as to output n sampled low-speed signals at a frequency of n, and each of the n low-speed sampled signals is converted into a first complex polyphase digital filter unit {2-0 to 2- (n -1)}, the n-divided small band signal obtained by using the first phase shift unit {3-0 to 3- (n-1)} and the first inverse fast Fourier transformer (4). , A small band signal dividing means (21) that divides the signal into complex signals for every other small band, and n complex signals obtained at the output of the small band signal dividing means (21). Transposing means (5) for transposing the frequency position in accordance with a predetermined encryption rule; and a second inverse fast Fourier transform of each of the n transposed complex signals obtained at the output of said transposing means (5). Vessel (6),
Second phase shift section {7-0-7- (n-1)} and second complex polyphase digital filter section {8-0-0}
8- (n-1)} and a frequency band for synthesizing each of the n-divided small band signals using interpolation means (9) for interpolating each of the signals obtained at the output of the complex polyphase digital filter section. A voice confidential speech device comprising a synthesizing means (31). 2. If the band from 0 Hz to the Nyquist frequency is divided into n and each sub-band number is 0, 1, 2,... (N-1), the sub-band signal dividing means (21) 0,2,4,…, n−
2. The voice secret device according to claim 1, wherein each small band signal is divided and output in the order of 2, ▲ ▼, ▲ ▼,... (In this case, the small band x is inverted on the frequency axis). .