JP3175162B2 - Secret communication method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a method for confidential communication of audio signals.

(Prior art and its problems) Particularly when transmitting a highly confidential audio signal, it is necessary to prevent a third party from intercepting the signal during transmission. For this reason, confidential communication is performed in which the original voice signal is processed and transmitted on the transmitting side and restored on the receiving side.

A method using frequency inversion has been used for a long time as a signal processing method for confidential speech signals, but in recent years, with the advancement of digital signal processing technology and device technology,
A method of replacing a signal sequence in the time domain has come to be used as a method with a strong secret talk intensity. However, even in the method of performing signal sequence replacement in the time domain, the intensity of secret talk is limited and is not always satisfactory.

For example, FIG. 1 is a diagram showing an example of a signal processing waveform of a confidential communication device that performs per-block replacement of a signal sequence in a time domain. In the figure, (a) shows an input signal waveform (audio signal). This input waveform is divided into frames (A, B, C) having a certain length (here, 320 msec as a representative value) as shown in (b). Next, each of the frames A, B, and C is further subdivided into blocks (a, b, c, d) each having a certain length (here, 80 msec as a representative value) as shown in (c). (D) shows the state of replacement in block units in each frame. For example, in the A frame, a signal input in the order of blocks (a, b, c, d) is replaced with blocks (d, c, b, a). The output waveform (e) is transmitted by performing the process of changing the temporal positional relationship of the transmitted audio signal in this manner. In this case, even if the voice from the transmission signal is heard, sufficient clarity for understanding the content cannot be obtained, so that the secrecy is achieved.

FIG. 2 is a circuit diagram on the transmitting side for executing such processing. The input signal (A) is an analog-to-digital converter
The signal is converted into a digital signal in 201 and stored in the memory 202 or 203 by the switch 210. As is well known, a method of selectively using the memory 202 and the memory 203 is, for example, a method in which a signal of the immediately preceding frame A stored in the memory 203 while the signal of the frame B is written to the memory 202 is used as a block. This is a method of reading in a different order.

A digital signal in which a signal sequence is replaced in blocks in the time domain read by the switch 211 from the memory 202 or the memory 203 is converted into a digital-to-analog converter 204
Is output to the outside as an analog signal (e).
The switches 210 and 211 and the memories 202 and 203 are controlled by a read / write control circuit 206, a frame counter 207, and a block counter 20.
8, controlled by the address control circuit 205. Further, there is a synchronizing circuit 209 for controlling these circuits (205, 206, 207, 208), but the details of this circuit and each of the circuits described above are well known and will not be described.
The portion surrounded by the broken line in FIG. Further, the restoration of the signal can be realized by performing a process reverse to the process shown in FIG. 2 by a signal sequence replacing unit (not shown).

However, in the above-mentioned conventional technology, since only the lock unit of the signal sequence in the time domain is replaced, the envelope component of the audio signal cannot be removed. The envelope component of the audio signal has information for understanding the content of the audio signal. In particular, the occurrence of a well-known word can be understood to some extent as long as the envelope component of the voice remains because it is largely related to the intonation of the voice. Therefore, there is a problem that the conventional technique cannot provide a sufficient level of privacy.

(Purpose of the Invention) It is an object of the present invention to provide a confidential communication method in which confidential information, which is a problem in the conventional method, is further improved to protect confidential information.

(Structure and Operation of the Invention) The confidential communication method of the present invention is different from the first method based on the fact that the voice signal is not continuous, and the spectrum of the vowel of the voice signal continues in a relatively long steady state, and at that time, Spectrum is a second method based on focusing on low frequencies, and it is also possible to use these two methods simultaneously.

 First, the first method will be described.

FIG. 3 shows waveforms of various parts based on the signal processing process of the present invention. (H) is an input signal waveform. This input waveform has a certain length as shown in (i).
320 msec) (A, B ...). Next, each of these frames is a block (a, b, c, d) of a certain length (here, 80 msec as a representative value) such as (nu).
Is subdivided into Furthermore, in order to make the length suitable for audio signal analysis, it is divided into sub-blocks (1, 2, 3, 4) each having a fixed length (here, 20 msec as a representative value) such as (R). The input signal (h) is analyzed for an audio signal on a sub-block basis to determine the presence or absence of an audio signal.

Here, the noise has a small fluctuation in the spectrum envelope and has no pitch component, the voiced sound has a small fluctuation in the spectrum envelope and the pitch component, and the unvoiced sound has a large fluctuation in the spectrum envelope and has no pitch component. To take advantage of the known property that

The determination result is indicated by [V] when an audio signal exists as in (ヲ) and by [N] when it is not, as shown in (ヲ). As a result of the determination, a result indicating that no voice signal is present is obtained, and noise is generated as shown in (a) only when the determination continues for two or more sub-blocks. Here, (W) is digital data, but is represented by analog notation. Next, this noise is added to the input signal (h) to obtain a signal (f) expressed in analog. This signal sequence (f) is input to a conventional permutation unit, and is replaced with each other in blocks of a, b, c, and d in each frame to obtain a signal sequence expressed in analog as shown in (ta). The state of the replacement is shown in FIG.

The signal sequence of (iv) is transmitted after digital-to-analog conversion, and is input to a conventional replacement unit after analog-to-digital conversion on the receiving side, and is reversely replaced with the original arrangement. This is shown in FIG. Next, the same analysis as on the transmitting side is performed for each sub-block with respect to the digital signal (S) in which the signal sequence has been subjected to reverse permutation, and the presence or absence of a voice signal is determined. The result of this determination is shown in FIG. Here, attention is paid to the A frame, the c block, and the fourth sub block. In this sub-block, since the addition of noise has just started, the fluctuation of the spectrum envelope with the previous sub-block is large, and it is determined that a voice signal (no voice) exists. For this reason, as shown in (d), when it is determined that no audio signal exists, the output is set to “0” over the sub-block and the sub-block determined to have the previous audio signal. To Therefore, by multiplying (S) and (D), the signal output (N) restored on the receiving side becomes a reproduction of the input signal (H) on the transmitting side.

FIG. 4 is a block diagram showing an example of a circuit for performing the above signal processing.

(A) shows a configuration example on the transmission side. The input signal (h)
It is digitized by analog-to-digital conversion 401. The spectrum analyzer 402 converts the signal into the frequency domain and then performs a spectrum analysis, and outputs information on the continuity of the spectrum envelope and the presence / absence of a pitch component. The audio sub-block detector 403 receiving these information determines whether or not an audio signal is included in sub-block units, and causes the noise generator 404 to generate noise in the sub-block determined to be [N]. On the other hand, the signal delayed by the appropriate time by the delay circuit 405 and the output of the noise generator 404 are added to the adder 406.
, And the output (f) is input to the replacement section of the conventional circuit to replace the signal sequence, and then the digital-to-analog conversion and transmission are performed.

(B) shows a configuration example on the receiving side. The signal returned to the original signal sequence output from the conventional replacement unit on the receiving side (SO)
As in the transmission side, the spectrum analysis unit 407 performs analysis in the frequency domain on a sub-block basis, and outputs information on the continuity of the spectrum envelope and the presence / absence of a pitch component. From this information, the additional noise sub-block detector 408 determines that the sub-block determined to be a noise section in the frequency domain and the sub-block of the speech immediately before the sub-block determined to be the noise section are the additional noise sub-blocks. From the determination result, the switch control circuit 409 controls the switch 412 so as to open the switch 412 in the case of the additional noise sub-block and to short-circuit otherwise.

Here, a signal that has been returned to the original signal sequence delayed by an appropriate time by the delay circuit 410 is input to the switch 412, and a signal in which additional noise is clipped by the switch 412 is obtained. This signal is converted by the digital-to-analog converter 411 into an analog signal (n), and becomes an audio output.

 Next, a second method of the present invention will be described.

FIG. 5 shows the waveform of each part based on the signal processing process of the present invention. The input signal waveform (Y) is divided into frames having a certain length (here, 320 msec) as shown in (MA). Here, only one frame (A) is shown. Subsequently, each frame is subdivided into blocks (a, b, c, d) having a certain length (here, 80 msec), such as (q). Next, the audio signal is divided into sub-blocks (1, 2, 3, 4) having a fixed length (here, 20 msec) as shown in FIG.

The input signal (Y) is analyzed for each sub-block to determine whether the input signal is a vowel or a non-vowel of a voice signal. Here, it is known that a vowel has a pitch component and that a component other than a vowel does not have a pitch component. The judgment result is indicated by (V) in the case of (v) for a vowel and [NC] otherwise.

Noise is generated as in (d) only when a result indicating that the vowel is a vowel is obtained in the determination result (v). Here, the noise is a narrow-band signal (noise) superimposed on a low spectral density portion of the sub-block. (D) is a digital signal, which is represented by an analog signal. Next, this noise is added to the input signal (Y) to obtain a signal expressed in analog as shown in (T). The spectrum (frequency f to power p characteristic) of each of the sub-blocks of the signal of (T) is shown in (A). Next, the signal of (T) is input to a conventional replacement unit, and the signals are replaced with each other in units of blocks of a, b, c, and d in the frame to obtain a signal sequence expressed in analog as shown in (G).
The state of the replacement is shown in FIG. The signal sequence (g) is transmitted after digital-to-analog conversion, and after analog-to-digital conversion on the receiving side, is input to a conventional replacement unit, and is reversely replaced with the original arrangement. This situation is shown in FIG. Next, the vowels and other signals are also determined for the digital signal (me) having undergone the reverse permutation of the signal sequence by analysis in units of sub-blocks similarly to the transmission side. The saw determination result is shown in (M).
The signal output restored on the receiving side by removing the added band signal by the narrow band rejection filter (BEF) only when it is determined to be a vowel based on the result (M).
Can reproduce the input signal (Y) on the transmission side.

FIG. 6 is a circuit block diagram for performing the processing of the second method.

(A) is a configuration example on the transmission side. In the figure, an input signal (Y) is converted into a digital signal by an analog-to-digital converter 601. The vowel sub-block detector 602 analyzes this signal and outputs a result (K) of determining whether the vowel is not a vowel based on the presence or absence of the pitch. In the subblock in which this information (K) is determined to be [V], the narrowband signal generator 60
3. A narrow band signal (d) is generated by (3). On the other hand, delay circuit 6
The input signal subjected to proper time replacement by 04 and the output (D) of the narrowband signal generator 603 are added by the adder 605, and the output (T) is input to the conventional replacement unit to replace the signal sequence. After that, the digital-to-analog conversion and transmission are performed.

(B) is a configuration example on the receiving side. Signal returned to the original signal sequence output from the conventional replacement unit on the receiving side (me)
The vowel sub-block detection circuit 606 performs analysis in the same manner as on the transmitting side, and outputs a determination result (i) as to whether the analyzed sub-block is a vowel section or not. Based on this information (m), the narrow band elimination filter control circuit 607
If it is a vowel, a narrow band filter 609 is inserted, and if it is not a vowel, control is made to bypass the narrow band elimination filter 609. Here, the delay circuit 608 is provided in the narrow band elimination filter 609.
The signal which has been returned to the original signal sequence delayed by an appropriate time by the
By means of 9, a signal from which the narrow-band signal added on the transmitting side is removed is obtained. This signal is converted to digital-to-analog
By 0, it is converted to an analog signal (S) and becomes an audio output.

As described above, the first method is a method of adding noise to a portion (sub-block) where no audio signal exists, and then replacing the signal sequence in block units in the time domain. In this method, a narrow-band signal is added to a vowel part (sub-block), and a signal sequence is replaced in block units in the time domain. Accordingly, since the spectrum envelope information of the confidential speech can be different from the spectrum envelope information of the original signal, the confidential speech intensity can be increased.

Furthermore, if the first method and the second method are used at the same time, the intensity of the confidential speech can be further increased as compared with the case of using either of the methods.

In the above description, examples in which the transmission path is an analog transmission path have been described. However, in the case of a digital transmission path, it is needless to say that transmission is performed without re-converting to an analog signal.

(Effects of the Invention) As described above in detail, according to the present invention, since the spectrum envelope information of the confidential speech can be different from the spectrum envelope information of the original signal, the confidential speech intensity is increased. be able to.

In addition, since there is no need to transmit information related to noise inserted to increase the intensity of the confidential talk to the receiving side, there is an advantage that the amount of information to be transmitted does not increase compared to the related art.

[Brief description of the drawings]

1 is a diagram showing an example of a processing waveform of the prior art, FIG. 2 is a diagram of an example of a circuit of the prior art, FIG. 3 is an example of a processing waveform of the first embodiment of the present invention, and FIG. Processing circuit diagram of the embodiment of FIG.
FIG. 7 is a diagram showing an example of a processing waveform according to the second embodiment of the present invention, and FIG. 6 is a processing circuit diagram of the second embodiment of the present invention. 200: Signal sequence replacement unit, 201, 401, 601 AD converter, 20
2,203 …… Memory, 204,411,610 …… DA converter, 205 ……
Address control circuit, 206 ... Read / write control circuit, 207 ...
Frame counter, 208 …… Block counter, 209 ……
Synchronous circuit, 210, 211, 412 Switch, 402, 407 Spectrum analyzer, 403 Voice detector, 404 Noise generator, 405, 410, 604, 608 Delay circuit, 406, 605 Adder, 408 Additional noise sub-block detector , 409: Switch control circuit, 602, 606: Vowel sub-block detector, 6
03 Narrow-band signal generator, 607 BEF control circuit, 609
... BEF.

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-58-198934 (JP, A) JP-A-63-209338 (JP, A) JP-A-61-263342 (JP, A) JP-A-2- 7636 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G09C 1/00-5/00 H04K 1/00-3/00 H04L 9/00

Claims (3)

(57) [Claims] 1. A confidential communication method in which a voice signal converted into a digital signal is replaced in a block of a signal sequence in a time domain and then transmitted to a transmission line, wherein each of the blocks of the digital signal is subjected to voice signal analysis. The sub-block is divided into a plurality of sub-blocks having a length suitable for performing the sub-block, and the presence or absence of an audio signal is sequentially determined by performing an audio signal analysis for each of the divided sub-blocks. As a result of this determination, it is determined that there is no audio signal A secret communication method characterized in that a noise is added to a sub-block only when two or more sub-blocks are consecutive, and then the block unit replacement of the signal sequence in the time domain is performed. 2. A confidential communication method in which a voice signal converted into a digital signal is replaced on a transmission line after performing block-wise replacement of a signal sequence in a time domain, wherein each of the digital signal blocks is subjected to voice signal analysis. It is subdivided into a plurality of sub-blocks of a length suitable for performing, and the audio signal analysis is performed in units of the sub-sub-blocks, and it is sequentially determined whether or not a vowel is based on the continuity of the spectrum envelope and the presence or absence of a pitch component. Determining a vowel, adding noise to the sub-block only when the sub-block is determined to be a vowel, and then performing per-block replacement of the signal sequence in the time domain. . 3. A confidential communication method in which a voice signal converted to a digital signal is replaced on a transmission line after performing block-wise replacement of a signal sequence in a time domain, wherein each of the digital signal blocks is subjected to voice signal analysis. The sub-blocks are divided into a plurality of sub-blocks having a length suitable for performing the sub-blocks. The sub-blocks are subjected to audio signal analysis to sequentially determine the presence / absence of an audio signal, and to sequentially determine whether or not they are vowels. As a result of this determination, when two or more sub-blocks determined to have no voice signal are continuous and when it is determined to be a vowel, noise is added to the sub-block, and then the block unit of the signal sequence in the time domain is used. Confidential communication method characterized by performing the substitution.