CN101059957A - An audio coding selective cryptographic method - Google Patents

Abstract

The invention discloses a speech coding selective encryption method, which includes step A, firstly perform initialization, select the security level required for encryption, and determine the adjustment parameters of the chaotic flow encryption method; step B, determine the selected one according to the frame rate and security level The bits of the parameters in the byte sequence number table and the bit position table are encrypted, and the encryption operation is performed by using a chaotic stream encryption method. Step C may also be included. After receiving the encrypted data, the decryption party first performs initialization and then performs a decryption operation. It selects and determines the data bits of speech codes that need to be encrypted according to the security level, improves the real-time performance of voice calls, reduces system energy consumption, and meets business requirements of different levels.

Description

A Speech Coding Selective Encryption Method

technical field

The invention relates to the technical field of speech coding. In particular, it relates to a selective encryption method of speech coding, especially a selective encryption method of G.723.1 speech coding.

Background technique

With the rapid growth of communication service demands, in order to ensure the transmission of communication service data, people vigorously carry out the research work of various data compression technologies. The compressed digital transmission of voice signals has been the direction that people have been working hard for many years.

The prior art generally adopts low-rate speech coding technology for speech transmission. Low-rate speech coding is to reduce the coding rate while minimizing distortion, so as to reduce the bandwidth occupied during transmission. Compared with analog transmission, it can save bandwidth and facilitate integration with the Internet (Internet Protocol, IP).

G.723.1 is a voice coding standard developed by the International Telecommunication Union (ITU) for low-bit-rate multimedia communications. This speech coding scheme is a part of the ITU-T H.324 standard series, which can compress speech or other audio signal components of multimedia equipment at a very low bit rate, with two bit rates of 6.3kbps and 5.3kbps. The excitation signal at high code rate (6.3kbs) is Multipulse Maximum Likelihood Quantization (MP-MLQ); the excitation signal at low code rate (5.3kbs) is algebraic codebook excitation linear prediction (Algebraic- Code-Excited Linear-Prediction, ACELP). The frame length of G.723.1 speech signal is 30ms, 240 sampling values. The encoder employs linear prediction-synthesis analysis coding to minimize the perceptually weighted error signal.

In the encoding process, one frame is input at a time, and each frame is high-pass filtered to remove the DC component, and then divided into 4 subframes, each with 60 sampling points. Carry out short-term predictive analysis on the speech signal with the linear predictive analysis method (Linear Predictive, LP), calculate the 10th order filter coefficient of its linear predictive coding (Linear Predictive Coding, LPC) for each subframe with the speech signal after adding the window, The LPC coefficients of these 4 subframes will be used to establish a short-term perceptual weighting filter, which is applied to the entire frame and obtains a perceptual weighting signal. The LPC filter coefficients of the last subframe will also be converted into Line Spectrum Pairs (LSP) coefficients, and then quantized using a predictive split vector quantizer.

As shown in Figure 1, it is a schematic diagram of G.723.1 encoding. G.723.1 utilizes the short-term correlation between speech samples and the long-term correlation of adjacent speech segments, and encodes the residual signals after removing the two correlations.

First carry out short-term speech analysis encoding: 1) 240 points of the speech signal after high-pass filtering and the last 120 points of the previous frame are synthesized into 360 samples, if the current frame is the first frame of the speech signal, then the preceding 120 points Each sample point is all 0; 2) Divide the sample points into 4 overlapping sections, each section is 180 long, and use the Hamming window function to multiply and weight to reduce the Gibbs effect brought by the segmentation; 3) Pass Calculate the autocorrelation function, etc., and obtain the linear prediction coefficient. From the short-term correlation of the speech signal, it can be seen that the prediction coefficient of the speech signal will not change greatly within the frame, so only the last 10 prediction coefficients (the LPC parameters of the last subframe in each frame) are used in this encoder To approximately replace the prediction coefficient of the speech of this frame.

Secondly, the LPC parameters of each subframe are converted into line spectrum pair (LSP) parameters, quantized and encoded by a Predictive Split Vector Quantizer (PSVQ) and transmitted: the line spectrum pair (LSP) residual The vector (the difference between the LSP vector and the LSP prediction vector after the long-term DC component is removed, is a 10-dimensional vector) is divided into 3 sub-vectors, the dimensions are 3, 3, and 4, and then 8-bit codebook quantization is performed on each sub-vector. In this way, three 8-bit codebook vectors are generated, with a total of 24-bit codebooks.

In order to improve the perceptual quality of quantization, the high-pass filtered speech signal needs to pass through formant perceptual weighting filter and formant noise shaping filter to filter the speech signal to generate the initial target signal. The parameters of the formant-aware weighting filter are composed of unquantized LPC coefficients of each subframe; the parameters of the formant noise shaping filter are obtained by performing open-loop pitch period estimation on every two subframes.

Carry out weighted filtering and noise filtering on the voice signal, during which the signal is also analyzed for a long time (that is, the pitch component, that is, the periodic component), that is, the open-loop pitch estimation and the closed-loop pitch prediction are performed successively to obtain the long-term parameter coding of the speech, and finally the Speech Removal of Long-Term Correlation. Wherein, the pitch period (adaptive codebook) of the even subframe is encoded with 7 bits, and the pitch period of the odd subframe is encoded with 2 bits of difference.

The margin signal is obtained by subtracting the long-term contribution from the target vector. Two code rates of 6.3kbit/s or 5.3kbit/s can be selected for encoding (fixed codebook) of the residual signal. The former (MP-MLQ algorithm) utilizes the small signal in the residual signal to have little effect on the quality of the synthesized speech, so the residual signal can be clipped to leave the one with a larger amplitude for encoding; the latter (ACELP algorithm) is used in the codebook. The stored codeword is used to replace the residual signal, and the code vector with the smallest mean square error between the residual signal and the residual signal is searched for transmission. The difference between the two algorithms lies in the number of coded pulses used to replace the residual signal: Algebraic Codebook Excited Linear Prediction (ACELP) uses slightly fewer pulses than MP-MLQ.

As shown in Figure 2, it is a schematic diagram of G.723.1 decoding. It first extracts the LSP codebook index value from the received code stream, and obtains the LSP parameters of each subframe after LSP decoding and interpolation, and converts them into LPC parameters to form LPC Synthesis filter. Then extract the pitch period, pitch gain codebook index and excitation pulse information of each subframe from the received code stream, and obtain the excitation signal e(n) through pitch decoding and excitation decoding respectively, perform pitch filtering on it and then pass through the synthesis filter The reconstructed speech is obtained, and the reconstructed speech is passed through a post-formant filter and a gain adjustment unit to obtain the final output of the decoder.

From the above process, it can be seen that the parameters to be transmitted for a frame signal processed by the encoder include: channel parameters, that is, LSP parameters, used to construct an LPC synthesis filter at the decoding end; excitation parameters, that is, pitch period parameters and long-term prediction Gain parameter, pulse position parameter and gain parameter of random codebook.

The traditional voice coding and encryption methods encrypt the entire G.723.1 compressed voice stream.

However, compared with text messages, voice streams have a large amount of data and high real-time requirements. Directly using traditional encryption algorithms such as AES (Advanced Encryption Standard) and 3DES (3Data Encryption Standard) algorithms to encrypt them as a whole will cause obvious delays , reducing its real-time performance. Not only that, it will significantly increase the computing load of the system, occupy more resources, and consume more energy. This is unbearable in many occasions, such as mobile communications, where low energy consumption is very demanding. In addition, the overall encryption of the voice data stream is inconvenient to meet the business requirements of different levels.

Contents of the invention

The problem to be solved by the present invention is to provide a selective encryption method for speech coding, which selects and determines the data bits of speech coding that need to be encrypted according to the security level.

A kind of speech coding selective encryption method provided for realizing the present invention, comprises the following steps:

Step A, first perform initialization, select the security level required for encryption, and determine the adjustment parameters of the chaotic flow encryption method;

Step B, according to the frame rate and security level, determine the bits of parameters in the encrypted byte sequence number table and bit position table, and use the chaotic stream encryption method to perform encryption operations.

The speech coding selective encryption method may also include the following steps:

In step C, after the decryption party receives the encrypted data, it first performs initialization and then performs a decryption operation.

Described step A comprises the following steps:

Step A1, the user selects the security level required for encryption;

Step A2, determining the adjustment parameters of the chaotic flow encryption method;

Step A3, get the effective binary bits of the initial value in the adjustment parameter, convert the byte data, store the converted data in an array as a key stream table, and perform XOR operation with the bits to be encrypted in the plain text of the voice data stream to get the ciphertext.

Said step B comprises the following steps:

Step B1, judge whether it is a high-rate frame, a low-rate frame, or a silent frame according to the first two bits of a received voice frame;

Step B2, according to different frames, go to the corresponding steps for processing:

If it is a high-rate frame, select the byte sequence number table corresponding to the high rate and the bit position table corresponding to the required security level, and go to step B3;

If it is a low-rate frame, select the byte sequence number table corresponding to the above low rate and the bit position table corresponding to the required security level, and go to step B3;

If it is a silent frame, no encryption is required, and it returns to step B1 to process the next frame;

Step B3, after determining the selected encrypted byte sequence number table and bit position table according to the frame rate and security level, perform the encryption operation of the current frame;

Step B4, after completing the encryption operation of the current frame, if the key stream is used up, use the adjustment parameters in the chaotic stream encryption method to iterate the current chaotic sequence value to obtain a new key stream, and update the key stream table; If there are speech frames, go back to step B1 and process the next frame.

Described step C comprises the following steps:

Step C1, receiving and decrypting the control parameters to obtain the encryption security level, and decrypting to obtain the adjustment parameters of the chaotic flow encryption method;

Step C2, take the effective binary digits of the adjustment parameters, convert them into byte data, store the converted data into the byte array as the key stream table, and perform XOR operation with the bits to be encrypted in the plain text of the voice data stream, to obtain plaintext;

Step C3, execute the specific decryption operation process of the current frame;

Step C4, after completing the decryption operation of the current frame, if the key stream is used up, use the adjustment parameters to iterate the current chaotic sequence value to obtain a new key stream, and update the key stream table; if there are still speech frames, turn to Process the next frame.

The chaotic flow encryption method is a chaotic flow encryption method characterized by a one-dimensional nonlinear iterative method of Logistic mapping.

The one-dimensional nonlinear iterative method of the Logistics map is an improved one-dimensional nonlinear iterative method of the Logistics map, and the iterative method is shown in the following formula:

G(x)＝(β+1)(1+1/β) ^β ×(1-x) ^β

Among them, β∈(1, 4), x ₀ ∈ (0, 1), the initial value of x is x ₀ , through iteration of this formula, x ₁ , x ₂ , x ₃ , ... x _n ... .

In the step B, the encryption byte sequence number table and the bit position table selected for use are determined according to the security level required for the frame rate and encryption, which is:

Determine the sensitivity arrangement order of the bit bits of the parameters in the speech frame, so as to divide the bits into different categories, and encrypt the bits of different categories to obtain different security levels.

The speech coding is the speech coding of the G.723.1 standard speech frame.

The sensitivity arrangement order of the bit bits of the parameters in the speech frame is five categories, respectively CLASS1, CLASS2, CLASS3, CLASS4, and CLASS5, and their importance decreases in order.

When the security level is Level 1, the bits in CLASS1 are encrypted, and the 48bit bits in the voice frame are encrypted in the high rate mode of the G.723.1 standard voice frame; the voice is encrypted in the low rate mode of the G.723.1 standard voice frame 38bit bit encryption in the frame;

When the security level is Level 2, the bits in CLASS1 and CLASS2 are encrypted, and the 62bit bits in the voice frame are encrypted in the high rate mode of the G.723.1 standard voice frame; 52bit encryption in voice frames;

When the security level is Level 3, encrypt the bits in CLASS1, CLASS2 and CLASS3, and encrypt 74 bits in the voice frame in the high rate mode of the G.723.1 standard voice frame; in the low rate mode of the G.723.1 standard voice frame In this mode, the 64bit bits in the voice frame are encrypted;

When the security level is Level 4, the bits in CLASS1, CLASS2, CLASS3 and CLASS4 are encrypted, and the 86bit bits in the voice frame are encrypted in the high-speed mode of the G.723.1 standard voice frame; in the G.723.1 standard voice frame Encrypt 76 bits in voice frames in low rate mode.

The beneficial effects of the present invention are: the speech coding selective encryption method of the present invention improves the real-time performance of the speech conversation on the one hand. Selective encryption is performed in voice frames, that is, between frames, only the high/low rate frames carrying voice information are encrypted, and a large number of SID frames (silent frames) are not encrypted; within frames, according to the needs of the security level , to encrypt those bits with high sensitivity (that is, having a great influence on reconstructed speech). Compared with the traditional overall encryption of voice media content, this undoubtedly speeds up the encryption speed; on the other hand, it can reduce system energy consumption and resource occupation. By selectively encrypting speech frames, these measures can not only improve efficiency, but also reduce system energy consumption and reduce the amount of resources occupied. Reducing energy consumption is a very critical issue in some application environments where energy resources are very tight, such as the limited energy carried by mobile devices in the field of wireless communications. Reducing the occupation of resources can also reduce the load of network processing nodes. Further, it can facilitate meeting business demands of different levels. Selective encryption can not only meet better technical requirements, but also has practical application value in many Internet businesses. For example, for websites that provide various audio services, the voice materials can be provided to users with different encryption strengths. If they are provided for free to users for trial listening, a certain encryption strength can be used for the voice to perform obfuscation. Many applications like this, in short, selective encryption provides feasibility and good technical support for different levels of business needs.

Description of drawings

Fig. 1 is the speech coding principle diagram of prior art G.723.1 standard speech frame;

Fig. 2 is the speech decoding principle diagram of prior art G.723.1 standard speech frame;

Fig. 3 is the flow chart of speech coding selective encryption method of the present invention;

Fig. 4 is a schematic diagram of the selective encryption process of interlingual coding in the present invention;

FIG. 5 is a flowchart of the initialization process of step S100 in FIG. 4;

Fig. 6 is that step S200 in Fig. 4 selects and confirms bit position, carries out the flow chart of encryption operation process;

Fig. 7 is a flowchart of the decryption process of step S300 in Fig. 4;

Fig. 8 is a schematic diagram of the encryption process of the third byte when the high-rate frame security level of the G.723.1 standard voice frame is 1 according to an embodiment of the present invention;

Fig. 9 is the schematic diagram of the 10th byte encryption process of the G.723.1 standard voice frame shown in Fig. 8;

Fig. 10 is a schematic diagram of the 11th byte encryption process of the G.723.1 standard voice frame shown in Fig. 8;

Fig. 11 is the schematic diagram of the 12th byte encryption process of the G.723.1 standard voice frame shown in Fig. 8;

Fig. 12 is the schematic diagram of the 14th byte encryption process of the G.723.1 standard voice frame shown in Fig. 8;

FIG. 13 is a schematic diagram of the encryption process starting from the third byte of the next frame of the G.723.1 standard voice frame shown in FIG. 8 .

Detailed ways

In order to make the purpose, technical solution and advantages of the present invention clearer, a voice coding selective encryption method of the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

Generally speaking, the measurement standard of speech encryption effect includes both objective and subjective evaluation, but for low-rate narrowband compressed speech signals, the parameters of objective evaluation, such as signal-to-noise ratio, etc., are difficult to truly reflect its performance, while Subjective evaluation is closer to the actual situation, so it usually pays more attention to subjective evaluation. Several main measurement criteria in subjective evaluation include the following points:

1. Speech intelligibility (semantics)

2. Voice gender judgment

3. Plaintext attack

4. Mute/non-mute distinction

This is the standard that is expected to be achieved in the subjective evaluation of speech. That is to say, in the embodiment of the present invention, it should be encrypted so that people can no longer make effective discrimination behaviors about the above points after listening to the speech.

It should be noted that the above four standards correspond to the highest encryption security requirements. In the embodiment of the present invention, the encryption effects obtained by different security levels are not the same, and not all of them can meet the above four standards, and the method in the present invention is a compromise between encryption efficiency and security strength. Find a balance. The above standard is only a general standard and principle that we hope to achieve, and has no direct relationship with the security level of the present invention, etc., and there is no sequence.

Digital voice, image, and video compressed data streams all have a property of nonuniform perceptual importance, that is, errors occurring in some bits of the data stream are less frequent than errors occurring in other bits. The impact is much more significant.

So far, the most important application of this conclusion is the Unequal Error Protection (UEP) mechanism used in wireless channel multimedia data transmission.

Therefore, different from encrypting the entire multimedia stream, the selective encryption method for voice coding of the present invention adopts a selective encryption method, which is different from the way of encrypting all data in the traditional encryption method, and the selective encryption method only affects a part more on perception Encrypt important bits, while the rest are directly transmitted in the channel without protection, that is, by keeping the encrypted data format information and control information unchanged, only the actual data is encrypted, so as to maintain the similarity of the encrypted data stream. Capacitance, it reduces the computing load while ensuring the required security strength, which saves more resources and energy consumption, greatly improves the transmission speed, and has better real-time performance. The performance can meet the real-time requirements. At the same time, the selective encryption method can be required to be combined with a specific data format. According to different security level requirements, different sensitive data can be encrypted to meet different needs.

The embodiment of the present invention uses the selective encryption of G.723.1 speech coding to illustrate the speech coding selective encryption method of the present invention, but the present invention is not limited to only applicable to G.723.1 speech coding, and it can be applicable to other standards Speech coding.

As shown in Fig. 3 and Fig. 4, wherein, Fig. 3 speech coding selective encryption method flowchart of the present invention; Fig. 4 is the speech coding selective encryption process schematic diagram of the present invention; A kind of speech coding selective encryption of the present invention is described in detail below method:

Step S100, first perform initialization, select the security level required for encryption, and determine the adjustment parameters of the encryption iteration method;

As shown in Figure 5, it specifically includes the following steps:

Step S110, the user selects the security level Level required for encryption;

Step S120, determining the adjustment parameters x ₀ and β of the chaotic flow encryption method represented by the one-dimensional nonlinear iterative method of the Logistics map;

The user selects these two parameters x ₀ and β, and these two parameters are equivalent to the key. x ₀ takes a value between 0 and 1, β takes a value between 1...~4, and there is no requirement for the number of digits after the decimal point. Preferably, in view of the number of digits that the computer can handle, the two parameters are in the Take no more than 10 digits afterward. The embodiment of the present invention illustrates that the two parameters take values x ₀ =0.3 and β=3.78. At the same time, it should be noted that in the computer VC programming environment, these two parameters can be stored as double-precision floating-point numbers, so as to select 64 effective binary digits.

Step S130, get the 64 effective binary digits after the decimal point of the initial value _x0 in the adjustment parameter, convert it into 8 byte type (byte) data, store the converted array into an array as the key stream table, in order to The XOR operation is performed with the bits to be encrypted in the plaintext of the voice data stream to obtain the ciphertext.

Described conversion is prior art, and it divides 64 binary digits into 8 groups, and each group of 8 bits just forms a byte type data;

The key stream table is obtained by storing the converted data into a byte type (byte) type array keyList[] with a length of 8, and it is used to carry out an XOR operation with the bits to be encrypted in plain text of the voice data stream to obtain ciphertext.

Step S200, determine the selected encryption byte sequence number table and bit position table according to the frame rate (frame length) and security level, and execute the encryption operation.

In the process of speech coding encryption, after the user selects the security level Level required for encryption, the user must select and determine the bits of the speech coding to be encrypted according to the format characteristics of the speech coding of the specific encrypted object G.723.1.

As shown in Figure 6, it specifically includes the following steps:

Step S210, at first according to the first two bits of a received frame of voice frame, judge whether it is a high rate frame (24 bytes of frame length, 6.3kbps), a low rate frame (20 bytes of frame length, 5.3kbps), or SID frame(silent frame);

Step S220, according to different frames, turn to corresponding steps for processing;

Case1: if it is a high-rate frame, then select the byte sequence number table corresponding to the high-rate to be encrypted and the bit position table corresponding to the required security level, and go to step S230;

Case2: if it is a low-rate frame, then select the byte sequence number table corresponding to the above low-rate to be encrypted and the bit position table corresponding to the required security level, and go to step S230;

Case3: if it is a SID frame (silent frame), then no encryption is required, and it turns back to step S210 to process the next frame;

Step S230, after determining the selected encryption byte sequence number table and bit position table according to the frame rate (frame length) and security level, perform the encryption operation of the current frame.

Step 240, after the encryption operation of the current frame is completed, if the key stream is used up, use the adjustment parameters in the chaotic stream encryption method to iterate the current chaotic sequence value to obtain a new key stream, and update the key stream table; If there are speech frames, go back to step 210 and process the next frame.

Further, the following steps may also be included:

In step S300, after receiving the encrypted data, the decryption party first performs initialization and then performs a decryption operation.

As shown in Figure 7, it specifically includes the following steps:

Step S310, receiving and decrypting the control parameters to obtain the encrypted security level level, and decrypting to obtain the adjustment parameters x ₀ and β of the chaotic flow encryption method;

Step S320, take the 64 effective binary digits after the decimal point of x ₀ , convert them into 8 byte data, and store the converted data into a byte array keyList[] with a length of 8 as a key stream table for use with The bits to be encrypted in the plaintext of the voice data stream are XORed to obtain the plaintext;

Step S330, execute the specific decryption operation process of the current frame;

Step S340, after completing the decryption operation of the current frame, if the key stream is used up, use the adjustment parameters to iterate the current chaotic sequence value to obtain a new key stream, and update the key stream table; if there are voice frames, turn to Process the next frame.

The specific decryption process for performing the selective encryption method of the present invention corresponds to the execution of the encryption operation process, and those skilled in the art can complete the decryption process according to the encryption process described in the specific embodiment of the present invention. Therefore, in the embodiment of the present invention , will not be described in detail one by one.

The specific method process of the adjustment parameters X ₀ and β of the chaotic flow encryption method characterized by the one-dimensional nonlinear iterative method of determining the Logistics mapping in the step S120 is described in detail below:

Preferably, considering the characteristics, efficiency and security of selective encryption, the chaotic stream encryption algorithm is adopted in the embodiment of the present invention, and the XOR operation is in bytes, but the use of the chaotic key stream is improved, combined with the The feature of the part of the G.723.1 frame structure that needs to be encrypted can properly reuse the chaotic key stream without greatly affecting its security performance, thereby improving efficiency.

The advantages of stream encryption are small error expansion and high real-time performance. Its confidentiality depends on the randomness of the key sequence generated by the key generator. When the key sequence is very close to the random sequence, its security is very high. However, in the traditional stream encryption method, the pseudo-random sequence generator used is a linear congruential generator or a linear feedback shift register, and their security is poor and easy to be cracked. Therefore, the embodiment of the present invention adopts a chaotic stream encryption method.

As an implementable way, a simple and widely studied autonomous one-dimensional discrete dynamic system is the logistic mapping method, which uses a one-dimensional nonlinear iterative method to characterize the chaotic flow encryption method, as shown in equation (10) .

F(x _n )＝λx _n (1-x _n ) (10)

Among them, n=0, 1, 2, ..., x ₀ and λ are adjustment parameters, when x ₀ ∈ (0, 1), λ ∈ (3.5699456..., 4), the Logistics mapping works in a chaotic state, That is, the sequence {x _k , k=0, 1, 2, 3...} generated by the initial condition x ₀ under the action of the Logistics mapping method is aperiodic and non-convergent, and is very sensitive to the initial value and parameters .

There are two problems with this chaotic flow encryption method, one is the fixed point of the flow encryption method, that is, multiple iterations approach a certain fixed value, and the other is the "stable window", that is, points gathered in a certain interval, in the window The resulting iterative sequence does not provide the necessary security as a keystream.

The embodiment of the present invention adopts an improved one-dimensional nonlinear iterative method of the Logistics map (Logistics map) to avoid the problems of the prior art. The iterative method of the embodiment of the present invention is shown in formula (11):

G(x)＝(β+1)(1+1/β) ^β ×(1-x) ^β (11)

Among them, β∈(1, 4), x ₀ ∈ (0, 1), the initial value of x is x ₀ , through iteration of this formula, x ₁ , x ₂ , x ₃ , ... x _n ... .

Theoretically, the sequence generated by formula (11) is aperiodic, but due to the limitation of the word length of the computer, the chaotic sequence obtained by the computer simulation is only the approximation of the objective chaos, so there is a "cyclic window" problem, that is, after For several iterations, the iteration value appears periodic. According to the test results, when double-precision floating-point operations are used, the average number of iterations available >=2*107.

The specific method process of determining the encryption byte sequence number table and bit position table selected by frame rate (frame length) and encryption required security level Level in step S200 in detail below:

In order to illustrate the encryption byte sequence number table and bit position table selected by frame rate (frame length) and encryption required security level Level of the present invention, first illustrate the G.723.1 standard frame structure, and analyze its parameters according to the bit frame structure Importance, and then select the bits to be encrypted according to the importance of the parameters according to the security level Level, that is, how to determine the sensitivity order of the bits of the parameters in the voice frame, so as to divide the bits into different categories and encrypt different categories The bits get different security levels.

As shown in Table 1, the bit allocation frame structure after the low-rate frame 5.3kbps mode G.723.1 speech coding:

Table 1 Frame structure table of bit allocation after G.723.1 speech coding in low-rate frame 5.3kbps mode parameter Codeword first subframe second subframe third subframe fourth subframe Total per frame (bit) LPC index LPC twenty four Adaptive codebook delay ACL0, ACL1, ACL2, ACL3 7 2 7 2 18 Joint encoding of all gains GAIN0, GAIN1, GAIN2, GAIN3 12 12 12 12 48 pulse position POS0, POS1, POS2, POS3 12 12 12 12 48 pulse symbol PSIG0, PSIG1, PSIG2, PSIG3 4 4 4 4 16 parity bit GRID0, GRID1, GRID2, GRID3 1 1 1 1 4 total 158

As shown in Table 2, the bit allocation frame structure after the high-rate frame 6.3kbps mode G.723.1 speech coding:

Table 2 Frame structure table of bit allocation after G.723.1 speech coding in high-rate frame 6.3kbps mode parameter Codeword first subframe second subframe third subframe fourth subframe Total per frame (bit) LPC index LPC twenty four Adaptive codebook delay ACL0, ACL1, ACL2, ACL3 7 2 7 2 18 Joint encoding of all gains GAIN0, GAIN1, GAIN2, GAIN3 12 12 12 12 48 pulse position POS0, POS1, POS2, POS3 20 18 20 18 73 pulse symbol PSIG0, PSIG1, PSIG2, PSIG3 6 5 6 5 twenty two parity bit GRID0, GRID1, GRID2, GRID3 1 1 1 1 4 total 189

By analyzing the encoding process of G.723.1 and comparing the frame structure, the role of each parameter in the frame can be determined:

1) The LPC index is the LPC parameter, which is used to construct the LPC synthesis filter at the decoding end. It is very critical. The LPC coefficient is closely related to the intelligibility (ie semantics) of the final speech;

2) The joint gains GAIN0, GAIN1, GAIN2, and GAIN3 are the joint coding of the gains of the adaptive codebook and the fixed codebook, and the gain parameters affect the human ear's ability to distinguish speech conversations and silent periods;

3) ACL0, ACL1, ACL2, and ACL3 respectively represent the adaptive codebook delays of the first, second, third, and fourth subframes. They represent the long-term pitch in speech excitation, that is, the periodic pulse component. This component affects human Ear-to-sound gender judgments;

4) Pulse sign (PSIG0, PSIG1, PSIG2, PSIG3) and pulse position (POS0, POS1, POS2, POS3) respectively represent the sign and position of the coded pulse of the fixed codebook (aperiodic pulse component in speech excitation). The fixed codebook is the approximation of the difference between the target vector and the long-term pitch contribution in the process of encoding the excitation signal, and it is the residual signal of the total excitation, so its importance is lower than that of the adaptive codebook. In the adaptive codebook On the basis of the already effective encryption, the encryption priority of the relevant parameters of the fixed codebook can be considered later.

Therefore, it is very important to clarify the three parameters of LPC parameters, joint gain and adaptive codebook through the joint consideration of the measurement standard of speech encryption effect and the role of G.723.1 parameters. When selecting encryption parameters, it is necessary to consider first. Fixed codebook Encryption of parameters can be considered as a post-processing.

At the same time, in order to further illustrate the selection of bits that need to be encrypted according to the security level, the embodiment of the present invention combines the arrangement order of the error sensitivity of the frame bits in Annex C of G.723.1 to analyze and illustrate the coded bit frame structure of G.723.1 The importance of medium parameters, and select the bits to be encrypted according to the security level.

After understanding how the different bits in the G.723.1 frame structure affect the reconstruction of the speech signal at the decoder, combine the arrangement order of the error sensitivity of the frame bits given in Annex C of G.723.1 , and according to the measurement standard of voice encryption effect, etc., according to the conditions of system resources and the requirements of different security levels, select different bits for encryption, and select the bits to be encrypted according to the security level.

Annex C of G.723.1 released by ITU-T is about channel coding in wireless communication, and because the channel bit error rate of wireless communication is very high, Annex C is proposed to be particularly sensitive to bit errors in the G.723.1 frame structure ( That is to say, errors have a very large impact on the reconstructed voice) bits for CRC check coding to enhance its error resistance performance. For this reason, Annex C gives the arrangement order of the bit error sensitivity of the improved frame structure of G.723.1 (slightly different from the standard frame format, which has stronger error resistance to channel errors). This is also one of the main arguments in favor of selective encrypted references.

In Annex C of G.723.1, the frame structure of G.723.1 is slightly modified to adapt to the transmission of wireless channels. It is a decompressed representation of the original structure and has stronger robustness against channel errors.

The high-speed (6.3kbps) transmission speech parameter frame format of the G.723.1 speech coded frame after the Annex C standard decompression of G.723.1 is shown in Table 3 below:

Table 3 G.723.1 speech coded frames after decompression

High-speed (6.3kbps) transmission voice parameter frame format table byte ordinal in channel bit 1 R_LPC_B5...R_LPC_B0, VAD, RATE 2 R_LPC_B13...R_LPC_B6 3 R_LPC_B21...R_LPC_B14 4 ACL0_B5...ACL0_B0, R_LPC_B23, R_LPC_B22 5 ACL2_B4...ACL2_B0, ACL1_B1, ACL1_B0, ACL0_B6 6 AGAIN0_B3...AGAIN0_B0, ACL3_B1, ACL3_B0, ACL2_B6, ACL2_B5 7 AGAIN1_B3...AGAIN1_B0, AGAIN0_B7...AGAIN0_B4 8 AGAIN2_B3...AGAIN2_B0, AGAIN1_B7...AGAIN1_B4 9 AGAIN3_B3...AGAIN3_B0, AGAIN2_B7...AGAIN2_B4 10 FGAIN0_B3...FGAIN0_B0, AGAIN3_B7...AGAIN3_B4 11 FGAIN2_B1, FGAIN2_B0, FGAIN1_B4...FGAIN1_B0, FGAIN0_B4 12 FGAIN3_B4...FGAIN3_B0, FGAIN2_B4...FGAIN2_B2 13 MSBPOS_B3...MSBPOS_B0, GRID3_B0, GRID2_B0, GRID1_B0, GRID0_B0 14 MSBPOS_B11...MSBPOS_B4 15 POS0_B6...POS0_B0, MSBPOS_B12 16 POS0_B14...POS0_B7 17 POS1_B6...POS1_B0, POS0_B15 18 POS2_B0, POS1_B13...POS1_B7 19 POS2_B8...POS2_B1 20 POS3_B0, POS2_B15...POS2_B9 twenty one POS3_B8...POS3_B1 twenty two PSIG0_B2...PSIG0_B0, POS3_B13...POS3_B9 twenty three PSIG1_B4...PSIG1_B0, PSIG0_B5...PSIG0_B3 twenty four PSIG3_B1, PSIG3_B0, PSIG2_B5...PSIG2_B0 25 UB, UB, UB, UB, UB, PSIG3_B4...PSIG3_B2

The low-rate (5.3kbps) transmission speech parameter frame format of the G.723.1 speech coded frame after the Annex C standard decompression of G.723.1 is shown in Table 4 below:

Table 4 G.723.1 speech coded frames after decompression

Low rate (5.3kbps) transmission speech parameter frame format table byte ordinal in channel bit 1 R_LPC_B5...R_LPC_B0, VAD, RATE 2 R_LPC_B13...R_LPC_B6 3 R_LPC_B21...R_LPC_B14 4 ACL0_B5...ACL0_B0, R_LPC_B23, R_LPC_B22 5 ACL2_B4...ACL2_B0, ACL1_B1, ACL1_B0, ACL0_B6 6 AGAIN0_B3...AGAIN0_B0, ACL3_B1, ACL3_B0, ACL2_B6, ACL2_B5 7 AGAIN1_B3...AGAIN1_B0, AGAIN0_B7...AGAIN0_B4 8 AGAIN2_B3...AGAIN2_B0, AGAIN1_B7...AGAIN1_B4 9 AGAIN3_B3...AGAIN3_B0, AGAIN2_B7...AGAIN2_B4 10 FGAIN0_B3...FGAIN0_B0, AGAIN3_B7...AGAIN3_B4 11 FGAIN2_B1, FGAIN2_B0, FGAIN1_B4...FGAIN1_B0, FGAIN0_B4 12 FGAIN3_B4...FGAIN3_B0, FGAIN2_B4...FGAIN2_B2 13 POS0_B3...POS0_B0, GRID3_B0, GRID2_B0, GRID1_B0, GRID0_B0 14 POS0_B11...POS0_B8, POS0_B7...POS0_B4 15 POS1_B7...POS1_B4, POS1_B3...POS1_B0 16 POS2_B3...POS2_B0, POS1_B11...POS1_B8 17 POS2_B11...POS2_B8, POS2_B7...POS2_B4 18 POS3_B7, POS3_B4...POS3_B3...POS3_B0 19 PSIG0_B3...PSIG0_B0, POS3_B11...POS3_B8 20 PSIG2_B3...PSIG2_B0, PSIG1_B3...PSIG1_B0 twenty one UB, UB, UB, UB, PSIG3_B3...PSIG3_B0

Table 5 is the subjective sensitivity sorting table of the bits of the high-speed (6.3 kbps) transmission speech frame after decompression in Annex C of G.723.1. The numbers on the corresponding bits indicate the ranking of their subjective sensitivities (the most important rank is 0, and so on).

Table 5 High rate after decompression (6.3kbps)

Subjective Sensitivity Sorting Table of Transmitted Speech Frame Bits byte ordinal in channel Subjective Sensitivity Ranking Number of Bits 1 175 180 189 190 191 192 VAD RATE 2 98 73 107 154 167 168 169 170 3 30 17 16 31 48 55 49 71 4 6 4 0 2 11 26 10 14 5 5 1 3 12 27 twenty four 60 8 6 44 66 62 82 25 61 9 7 7 45 67 63 83 78 36 50 40 8 46 68 64 84 79 37 51 41 9 47 69 65 85 80 38 52 42 10 56 99 159 185 81 39 53 43 11 161 187 20 57 100 160 186 19 12 twenty two 59 102 162 188 twenty one 58 101 13 35 54 70 72 179 178 177 176 14 13 15 twenty three 28 29 32 33 34 15 128 132 146 155 163 171 181 18 16 76 86 90 94 103 108 112 116 17 129 133 147 156 164 172 182 74 18 183 87 91 95 104 109 113 117 19 114 118 130 134 148 157 165 173 20 184 75 77 88 92 96 105 110 twenty one 115 119 131 135 149 158 166 174 twenty two 136 124 120 89 93 97 106 111 twenty three 151 141 137 125 121 144 150 140 twenty four 127 123 145 152 142 138 126 122 25 UB UB UB UB UB 153 143 139

Table 6 is the subjective sensitivity sorting table of the bits of the speech frame transmitted at a low rate (5.3 kbps) after decompression in Annex C of G.723.1. The numbers on the corresponding bits indicate the ranking of their subjective sensitivities (the most important rank is 0, and so on).

Table 6 Low rate after decompression (5.3kbps)

Subjective Sensitivity Sorting Table of Transmitted Speech Frame Bits channel Subjective Sensitivity Ranking Number of Bits Byte ordinals in 1 152 153 158 159 160 161 VAD RATE 2 69 64 70 91 145 140 147 146 3 twenty four 15 14 25 46 50 47 63 4 4 6 0 2 11 18 10 13 5 7 1 3 12 19 16 48 8 6 42 59 55 65 17 49 9 5 7 43 60 56 66 26 30 34 38 8 44 61 57 67 27 31 35 39 9 45 62 58 68 28 32 36 40 10 51 87 141 154 29 33 37 41 11 143 156 twenty one 52 88 142 155 20 12 twenty three 54 90 144 157 twenty two 53 89 13 100 128 96 104 151 150 149 148 14 132 112 116 136 108 120 124 92 15 109 121 125 93 101 129 97 105 16 102 130 98 106 133 113 117 137 17 134 114 118 138 110 122 126 94 18 111 123 127 95 103 131 99 107 19 83 79 75 71 135 115 119 139 20 85 81 77 73 84 80 76 72 twenty one UB UB UB UB 86 82 78 74

It can be seen from the decompressed frame bit sensitivity ranking table combined with Annex C of G.723.1 that the ranking order of its importance is basically consistent with the previous conclusions on the role and importance of each parameter of G.723.1 .

That is, in general, the parameters LPC coefficients, adaptive code vector index and gain are more sensitive because they have a great influence on reconstructed speech.

The bit sensitivity sorting obtained in Table 5 and Table 6 is the bit sensitivity sorting in the decompressed speech frame of G.723.1 Annex C, but what the present invention needs is the bit in the G.723.1 standard speech frame Bit-sensitivity sorting, so it is necessary to obtain the conversion relationship between the two voice frames and the corresponding relationship between bits, so as to obtain the sorting and classification of the bit-sensitivity in the G.723.1 standard voice frames.

The conversion relationship between the bit-sensitive sorting in the decompressed speech frame of G.723.1 Annex C and the bit-sensitive sorting in the G.723.1 standard speech frame, and the bit Corresponding relationship:

Combining Table 1 to Table 6 about G.723.1 parameters, and Annex C parameters of G.723.1 and subjective sensitivity ranking table of speech frame bits. It can be seen that, because the decompressed frame structure of Annex C of G.723.1 is different from the standard frame structure in the encoding sequence of each parameter, the bit-sensitive sequence of the decompressed frame cannot be directly used for The standard frame structure of G.723.1, so it is necessary to determine the mutual conversion relationship between the voice parameters of the G.723.1 standard and the voice coding parameters of the Annex C standard of G.723.1, so that the voice parameters of the G.723.1 standard and the Annex C standard of G.723.1 Speech coding parameters can be converted to each other.

1) Conversion relationship of LPC parameters:

In the G.723.1 standard encoding, the LPC parameter is to divide the LSP residual vector (the difference between the LSP vector and the LSP prediction vector after the long-term DC component is removed, which is a 10-dimensional vector) into 3 sub-vectors, and the dimensions are 3 and 3 respectively. , 4, and then perform 8-bit codebook quantization on each sub-vector, thus generating three 8-bit codebook vectors, with a total of 24-bit codebooks.

The LPC parameters in the G.723.1 standard frame structure are obtained through the following transformations to obtain the LPC parameters in the frame structure in Annex C of G.723.1:

The LPC parameters in the G.723.1 standard frame structure consist of the following three sub-vectors:

E0={LPC_B7, LPC_B6, LPC_B5, LPC_B4, LPC_B3, LPC_B2, LPC_B1, LPC_B0}

E1={LPC_B15, LPC_B14, LPC_B13, LPC_B12, LPC_B11, LPC_B10, LPC_B9, LPC_B8}

E2={LPC_B23, LPC_B22, LPC_B21, LPC_B20, LPC_B19, LPC_B18, LPC_B17, LPC_B16}

The three sub-vectors get new three sub-vectors through the mapping of formula (1):

${\mathrm{<span\; class="notranslate">\; e</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}^{\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; Reorder\; Ta</span>}{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; [</span>{\mathrm{<span\; class="notranslate">\; e</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; ]</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0,1.2</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

The obtained output is the LPC parameter index used for wireless channel transmission in Annex C of G.723.1.

Therefore, the corresponding relationship between decompressed frame parameters and standard frame parameters in Annex C of G.723.1 is:

R_LPC_B7...R_LPC_B0 is mapped from LPC_B7...LPC_B0 (the first sub-vector);

R_LPC_B15...R_LPC_B8 is obtained by mapping LPC_B15...LPC_B8 (the second sub-vector);

R_LPC_B23...R_LPC_B16 is obtained by mapping LPC_B23...LPC_B16 (the third sub-vector);

In the sensitivity ranking table of the bits of the frame structure in Annex C of G.723.1, the sensitivity of the last sub-vector of the LPC parameter (R_LPC_B23...R_LPC_B16) is much higher than the first two, so in the standard frame structure, the corresponding There are LPC_B23...LPC_B16 with high sensitivity and LPC_B15...LPC_B0 with low sensitivity.

2) Conversion relationship of gain parameters:

In G.723.1 standard coding, the joint gain of each subframe is coded with 12 bits, which is actually the joint compression coding of two independent gains (adaptive codebook related gain and fixed codebook related gain). In Annex C of G.723.1, in order to enhance the ability of the frame structure to resist channel errors, it decompresses the joint gain in the standard structure, and decompresses the gain index GAINx_By of each subframe into an 8-bit AGAINx_By and A 5-bit FGAINx_By has two gain indices.

The steps are as follows:

21) Decoding according to formula (2) to obtain the pitch period prediction of each even subframe

Among them, PIndi is the adaptive codebook index

22) According to formula (3), the gene period prediction of each odd subframe is obtained by decoding

23) Calculate the gain vector of the pitch period prediction of each subframe

231) When the rate is low, the joint gain index (GIndi) contains the information of the pitch period prediction (adaptive codebook) gain vector and the information of the pulse sequence (fixed codebook) gain index, and its calculation formula is expressed as formula (4):

Among them, PGIndi is the pitch cycle adaptive codebook correlation gain index, GSize=24;

232) When the rate is high, if Li>=58, the gain calculation is the same as formula (4). Otherwise, its calculation is as formula (5):

24) Calculate the maximum gain (MGIndi) of the pulse of each subframe according to formula (6)

From the above decompression process, the corresponding relationship between decompressed frame parameters and standard frame parameters in Annex C of G.723.1 can be obtained as follows:

The gain index AGAINx_By corresponds to the pitch period adaptive codebook related gain index PGIndi in formula (5), where the gain index FGAINx_By corresponds to the maximum gain MGIndi in formula (6), and the gain index GAINx_By in the standard frame format corresponds to the above formula Joint gain index GIndi. Therefore, AGAINx_By and FGAINx_By are decompressed from GAINx_By.

GAINx_By is a parameter in the G.723.1 standard frame structure. AGAINx_By and FGAINx_By are parameters in the decompressed frame structure in G.723.1 Annex C, which are decompressed from the GAINx_By parameter in the standard frame structure. In the present invention, according to the conversion relationship between them, to obtain the corresponding relationship of bits between GAINx_By and AGAINx_By, FGAINx_By, thereby can be sorted by the bit sensitivity of the decompressed frame given in Annex C, get Bit-sensitive ordering of standard speech frames.

From the bit sensitivity ranking table, it can be concluded that the sensitivity of the gain index AGAINx_By (y>=3) of each subframe is relatively high, the sensitivity of the gain index FGAINx_B4 of each subframe is very high, and the sensitivity of AGAINx_By (y<3 ) and the lower four bits of FGAINx_By are less sensitive.

Since AGAINx_By is the quotient of GAINx_By in the standard frame structure divided by 24, the bits of AGAINx_By (y>=3) are mainly determined by the highest 4 bits of GAINx_By, so correspondingly in the G.723.1 standard frame structure, the highest 4 bits of GAINx_By bit is more sensitive than other bits.

In addition, the highest bit of the gain index FGAINx_By is more complicated. Not only the high bit of GAINx_By affects it, but the low bit also affects its value, but the sensitivity of the lower four bits of FGAINx_By is very low, so the overall priority is given to the high bit of AGAINx_By Encryption of the affected GAINx_By bit, that is, the high bit of GAINx_By.

Therefore, the bits of the G.723.1 standard frame are divided into five categories according to sensitivity: CLASS1, CLASS2, CLASS3, CLASS4, CLASS5, and the bit sensitivity of CLASS1 is the highest, followed by CLASS2, and so on. Therefore, the encryption of the lowest security level Level1 is to encrypt the bits in CLASS1, the level of one higher security level Level2 is to encrypt the bits in CLASS1 and CLASS2, ..., the encryption of the highest security level Level4 is to encrypt the bits in the five categories The first 4 types of bits are encrypted.

Among them, CLASS is the classification of bits, and Level is the division of security levels.

In the high-speed mode (6.3kbps) of the standard frame structure of G.723.1, the subjective sensitivity ranking numbers of the bits are shown in Table 7. In particular, it should be noted that this table does not give the precise ordering of each bit, but roughly classifies them according to their sensitivities. The five classes represent their sensitivities from high to low.

                                                                                        , the standard frame structure of G.723.1 is at a high rate

In mode (6.3kbps), the subjective sensitivity order number of bits byte ordinal in channel Subjective Sensitivity Ranking Number of Bits 1 LPC_B5 (CLASS5) LPC_B4 (CLASS5) LPC_B3 (CLASS5) LPC_B2 (CLASS5) LPC_B1(CLASS5) LPC_B0 (CLASS5) VADFLAG_B0 RATEFLAG_B0 2 LPC_B13 (CLASS5) LPC_B12 (CLASS5) LPC_B11 (CLASS5) LPC_B10 (CLASS5) LPC_B9 (CLASS5) LPC_B8 (CLASS5) LPC_B7 (CLASS5) LPC_B6 (CLASS5) 3 LPC_B21 (CLASS1) LPC_B20 (CLASS1) LPC_B19 (CLASS1) LPC_B18 (CLASS1) LPC_B17 (CLASS2) LPC_B16(CLASS2) LPC_B15 (CLASS2) LPC_B14 (CLASS2) 4 ACL0_B5 (CLASS1) ACL0_B4 (CLASS1) ACL0_B3 (CLASS1) ACL0_B2 (CLASS1) ACL0_B1 (CLASS1) ACL0_B0 (CLASS1) LPC_B23 (CLASS1) LPC_B22 (CLASS1) 5 ACL2_B4 (CLASS1) ACL2_B3 (CLASS1) ACL2_B2(CLASS1) ACL2_B1(CLASS1) ACL2_B0(CLASS1) ACL1_B1 (CLASS1) ACL1_B0(CLASS2) ACL0_B6 (CLASS1) 6 GAIN0_B3 (CLASS3) GAIN0_B2 (CLASS4) GAIN0_B1 (CLASS4) GAIN0_B0 (CLASS4) ACL3_B1 (CLASS1) ACL3_B0 (CLASS2) ACL2_B6(CLASS1) ACL2_B5 (CLASS1) 7 GAIN0_B11 (CLASS1) GAIN0_B10 (CLASS) GAIN0_B9 (CLASS1) GAIN0_B8 (CLASS1) GAIN0_B7 (CLASS2) GAIN0_B6 (CLASS2) GAIN0_B5 (CLASS3) GAIN0_B4 (CLASS3) 8 GAIN1_B7 (CLASS2) GAIN1_B6 (CLASS2) GAIN1_B5 (CLASS3) GAIN1_B4 (CLASS3) GAIN1_B3 (CLASS3) GAIN1_B2 (CLASS4) GAIN1_B1 (CLASS4) GAIN1_B0 (CLASS4) 9 GAIN2_B3 (CLASS3) GAIN2_B2 (CLASS4) GAIN2_B1 (CLASS4) GAIN2_B0 (CLASS4) GAIN1_B11 (CLASS1) GAIN1_B10 (CLASS1) GAIN1_B9 (CLASS1) GAIN1_B8 (CLASS1) 10 GAIN2_B11 (CLASS1) GAIN2_B10 (CLASS1) GAIN2_B9 (CLASS1) GAIN2_B8 (CLASS1) GAIN2_B7 (CLASS2) GAIN2_B6 (CLASS2) GAIN2_B5 (CLASS3) GAIN2_B4 (CLASS3) 11 GAIN3_B7 (CLASS2) GAIN3_B6 (CLASS2) GAIN3_B5 (CLASS3) GAIN3_B4 (CLASS3) GAIN3_B3 (CLASS3) GAIN3_B2 (CLASS4) GAIN3_B1 (CLASS4) GAIN3_B0 (CLASS4) 12 GRID3_B0(CLASS5) GRID2_B0(CLASS5) GRID1_B0 (CLASS5) GRID0_B0 (CLASS5) GAIN3_B11 (CLASS1) GAIN3_B10 (CLASS1) GAIN3_B9 (CLASS1) GAIN3_B8 (CLASS1) 13 MSBPOS_B6 (CLASS1) MSBPOS_B5 (CLASS1) MSBPOS_B4 (CLASS1) MSBPOS_B3 (CLASS1) MSBPOS_B2 (CLASS5) MSBPOS_B1 (CLASS5) MSBPOS_B0 (CLASS5) UB (CLASS5) 14 POS0_B1 (CLASS5) POS0_B0 (CLASS5) MSBPOS_B12 (CLASS1) MSBPOS_B11 (CLASS1) MSBPOS_B10 (CLASS1) MSBOS_B9(CLASS1) MSBPOS_B8 (CLASS1) MSBPOS_B7 (CLASS1) 15 POS0_B9 (CLASS5) POS0_B8 (CLASS5) POS0_B7 (CLASS5) POS0_B6 (CLASS5) POS0_B5 (CLASS5) POS0_B4 (CLASS5) POS0_B3 (CLASS5) POS0_B2 (CLASS5) 16 POS1_B1 (CLASS5) POS1_B0 (CLASS5) POS0_B15 (CLASS5) POS0_B14 (CLASS5) POS0_B13 (CLASS5) POS0_B12 (CLASS5) POS0_B11 (CLASS5) POS0_B10 (CLASS5) 17 POS1_B9 (CLASS5) POS1_B8 (CLASS5) POS1_B7 (CLASS5) POS1_B6 (CLASS5) POS1_B5 (CLASS5) POS1_B4 (CLASS5) POS1_B3 (CLASS5) POS1_B2 (CLASS5) 18 POS2_B3 (CLASS5) POS2_B2 (CLASS5) POS2_B1 (CLASS5) POS1_B0 (CLASS5) POS1_B13 (CLASS5) POS1_B12 (CLASS5) POS1_B11 (CLASS5) POS1_B10 (CLASS5) 19 POS2_B11 (CLASS5) POS2_B10 (CLASS5) POS2_B9 (CLASS5) POS2_B8 (CLASS5) POS2_B7 (CLASS5) POS2_B6 (CLASS5) POS2_B5 (CLASS5) POS2_B4 (CLASS5) 20 POS3_B3 (CLASS5) POS3_B2 (CLASS5) POS3_B1 (CLASS5) POS3_B0 (CLASS5) POS2_B15 (CLASS5) POS2_B14 (CLASS5) POS2_B13 (CLASS5) POS2_B12 (CLASS5) twenty one POS3_B11 POS3_B10 POS3_B9 POS3_B8 POS3_B7 POS3_B6 POS3_B5 POS3_B4 (CLASS5) (CLASS5) (CLASS5) (CLASS5) (CLASS5) (CLASS5) (CLASS5) (CLASS5) twenty two PSIG0_B5 (CLASS5) PSIG0_B4 (CLASS5) PSIG0_B3 (CLASS5) PSIG0_B2 (CLASS5) PSIG0_B1 (CLASS5) PSIG0_B0 (CLASS5) POS3_B13 (CLASS5) POS3_B12 (CLASS5) twenty three PSIG2_B2 (CLASS5) PSIG2_B1 (CLASS5) PSIG2_B0 (CLASS5) PSIG1_B4 (CLASS5) PSIG1_B3 (CLASS5) PSIG1_B2 (CLASS5) PSIG1_B1 (CLASS5) PSIG1_B0 (CLASS5) twenty four PSIG3_B4 (CLASS5) PSIG3_B3 (CLASS5) PSIG3_B2 (CLASS5) PSIG3_B1 (CLASS5) PSIG3_B0 (CLASS5) PSIG2_B5 (CLASS5) PSIG2_B4 (CLASS5) PSIG2_B3 (CLASS5)

In the low-rate mode (5.3kbps) of the standard frame structure of G.723.1, the subjective sensitivity ranking numbers of the bits are shown in Table 8:

Table 8 Low rate in G.723.1 standard frame structure

In mode (5.3kbps), the subjective sensitivity order number of bits byte ordinal in channel Subjective Sensitivity Ranking Number of Bits 1 LPC_B5 (CLASS5) LPC_B4 (CLASS5) LPC_B3 (CLASS5) LPC_B2 (CLASS5) LPC_B1(CLASS5) LPC_B0 (CLASS5) VADFLAG_B0 RATEFLAG_B0 2 LPC_B13 (CLASS5) LPC_B12 (CLASS5) LPC_B11 (CLASS5) LPC_B10 (CLASS5) LPC_B9 (CLASS5) LPC_B8 (CLASS5) LPC_B7 (CLASS5) LPC_B6 (CLASS5) 3 LPC_B21 (CLASS1) LPC_B20 (CLASS1) LPC_B19 (CLASS1) LPC_B18 (CLASS1) LPC_B17 (CLASS2) LPC_B16(CLASS2) LPC_B15 (CLASS2) LPC_B14 (CLASS2) 4 ACL0_B5 (CLASS1) ACL0_B4 (CLASS1) ACL0_B3 (CLASS1) ACL0_B2 (CLASS1) ACL0_B1 (CLASS1) ACL0_B0 (CLASS1) LPC_B23 (CLASS1) LPC_B22 (CLASS1) 5 ACL2_B4 (CLASS1) ACL2_B3 (CLASS1) ACL2_B2(CLASS1) ACL2_B1(CLASS1) ACL2_B0(CLASS1) ACL1_B1 (CLASS1) ACL1_B0(CLASS2) ACL0_B6 (CLASS1) 6 GAIN0_B3 (CLASS3) GAIN0_B2 (CLASS4) GAIN0_B1 (CLASS4) GAIN0_B0 (CLASS4) ACL3_B1 (CLASS1) ACL3_B0 (CLASS2) ACL2_B6(CLASS1) ACL2_B5 (CLASS1) 7 GAIN0_B11 (CLASS1) GAIN0_B10 (CLASS1) GAIN0_B9 (CLASS1) GAIN0_B8 (CLASS1) GAIN0_B7 (CLASS2) GAIN0_B6 (CLASS2) GAIN0_B5 (CLASS3) GAIN0_B4 (CLASS3) 8 GAIN1_B7 (CLASS2) GAIN1_B6 (CLASS2) GAIN1_B5 (CLASS3) GAIN1_B4 (CLASS3) GAIN1_B3 (CLASS3) GAIN1_B2 (CLASS4) GAIN1_B1 (CLASS4) GAIN1_B0 (CLASS4) 9 GAIN2_B3 (CLASS3) GAIN2_B2 (CLASS4) GAIN2_B1 (CLASS4) GAIN2_B0 (CLASS4) GAIN1_B11 (CLASS1) GAIN1_B10 (CLASS1) GAIN1_B9 (CLASS1) GAIN1_B8 (CLASS1) 10 GAIN2_B11 (CLASS1) GAIN2_B10 (CLASS1) GAIN2_B9 (CLASS1) GAIN2_B8 (CLASS1) GAIN2_B7 (CLASS2) GAIN2_B6 (CLASS2) GAIN2_B5 (CLASS3) GAIN2_B4 (CLASS3) 11 GAIN3_B7 (CLASS2) GAIN3_B6 (CLASS2) GAIN3_B5 (CLASS3) GAIN3_B4 (CLASS3) GAIN3_B3 (CLASS3) GAIN3_B2 (CLASS4) GAIN3_B1 (CLASS4) GAIN3_B0 (CLASS4) 12 GRID3_B0(CLASS5) GRID2_B0(CLASS5) GRID1_B0 (CLASS5) GRID0_B0 (CLASS5) GAIN3_B11 (CLASS1) GAIN3_B10 (CLASS1) GAIN3_B9 (CLASS1) GAIN3_B8 (CLASS1) 13 POS0_B7 (CLASS5) POS0_B6 (CLASS5) POS0_B5 (CLASS5) POS0_B4 (CLASS5) POS0_B3 (CLASS5) POS0_B2 (CLASS5) POS0_B1 (CLASS5) POS0_B0 (CLASS5) 14 POS1_B3 (CLASS5) POS1_B2 (CLASS5) POS1_B1 (CLASS5) POS1_B0 (CLASS5) POS0_B11 (CLASS5) POS0_B10 (CLASS5) POS0_B9 (CLASS5) POS0_B8 (CLASS5) 15 POS1_B11 (CLASS5) POS1_B10 (CLASS5) POS1_B9 (CLASS5) POS1_B8 (CLASS5) POS1_B7 (CLASS5) POS1_B6 (CLASS5) POS1_B5 (CLASS5) POS1_B4 (CLASS5) 16 POS2_B7 (CLASS5) POS2_B6 (CLASS5) POS2_B5 (CLASS5) POS2_B4 (CLASS5) POS2_B3 (CLASS5) POS2_B2 (CLASS5) POS2_B1 (CLASS5) POS2_B0 (CLASS5) 17 POS3_B3 (CLASS5) POS3_B2 (CLASS5) POS3_B1 (CLASS5) POS3_B0 (CLASS5) POS2_B11 (CLASS5) POS2_B10 (CLASS5) POS2_B9 (CLASS5) POS2_B8 (CLASS5) 18 POS3_B11 (CLASS5) POS3_B10 (CLASS5) POS3_B9 (CLASS5) POS3_B8 (CLASS5) POS3_B7 (CLASS5) POS3_B6 (CLASS5) POS3_B5 (CLASS5) POS3_B4 (CLASS5) 19 PSIG1_B3 (CLASS5) PSIG1_B2 (CLASS5) PSIG1_B1 (CLASS5) PSIG1_B0 (CLASS5) PSIG0_B3 (CLASS5) PSIG0_B2 (CLASS5) PSIG0_B1 (CLASS5) PSIG0_B0 (CLASS5) 20 PSIG3_B3 (CLASS5) PSIG3_B2 (CLASS5) PSIG3_B1 (CLASS5) PSIG3_B0 (CLASS5) PSIG2_B3 (CLASS5) PSIG2_B2 (CLASS5) PSIG2_B1 (CLASS5) PSIG2_B0 (CLASS5)

It can be obtained from Table 7 and Table 8 that when the security level is Level1, the bits in CLASS1 are encrypted. In the high-speed mode (6.3kbps), 48 bits in CLASS1 need to be encrypted, so 48 bits in the voice frame are encrypted. ;Under the low rate mode (5.3kbps), it is necessary to encrypt 38 bits in CLASS1, so the 38 bits in the voice frame are encrypted;

When the security level is Level 2, the bits in CLASS1 and CLASS2 are encrypted, and 14 bits in CLASS2 need to be encrypted in the high-speed mode (6.3kbps), so 48+14=62bits in the voice frame are encrypted; low Also need to encrypt 14 in the CLASS2 under the rate pattern (5.3kbps), therefore 38+12=52bit in the speech frame is encrypted;

When the security level is Level 3, the bits in CLASS1, CLASS2 and CLASS3 are encrypted, and 12 bits in CLASS3 need to be encrypted in the high-speed mode (6.3kbps), so 62+12=74 bits in the voice frame are encrypted ; Also need to encrypt 12 in the CLASS3 under the low rate mode (5.3kbps), so 52+12=64bit in the voice frame is encrypted;

When the security level is Level 4, the bits in CLASS1, CLASS2, CLASS3 and CLASS4 are encrypted, and 12 bits in CLASS4 need to be encrypted in the high-speed mode (6.3kbps), so 74+12=86bit in the voice frame Bit encryption; under the low rate mode (5.3kbps), 12 bits need to be encrypted, so 64+12=76 bits in the speech frame are encrypted.

In the embodiment of the present invention, according to the sensitivity sorting of bits, 48 bits (bits) are given to the first class (CLASS1), and 14 bits (bits) are given to the second class, which is only a better division category. The most sensitive 48 bits are classified as the first category because it includes the most important bits of several important parameters.

In the embodiment of the present invention, the priority order of bit encryption is given, without limiting a certain specific application background and security strength requirements. If in a certain practical application, the security strength of encrypting the bits does not meet the application requirements of the specific environment, on this basis, increase the encryption of the bits in CLASS1, CLASS2... until the desired effect is obtained. Those skilled in the art may also perform sensitivity sorting on the bits according to the division methods commonly used in the art, but this does not exceed the scope of the present invention.

As an implementable manner, the data structure of the security level and the bit position table to be encrypted for each byte in a selective encryption frame can be expressed as follows:

/*Security Level*/

#define SECURITY_LEVEL1 1 ∥ needs to encrypt the bits in CLASS1 proposed above

#define SECURITY_LEVEL2 2 ∥ It is necessary to encrypt the bits in CLASS1~2 proposed above

#define SECURITY_LEVEL3 3 ∥ It is necessary to encrypt the bits in CLASS1~3 proposed above

#define SECURITY_LEVEL4 4 ∥ It is necessary to encrypt the bits in CLASS1~4 proposed above

/*Selectively encrypt the byte sequence number table to be encrypted in a frame*/

byte highrate_bytepos_list[]={3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14};

byte lowrate_bytepos_list[]={3, 4, 5, 6, 7, 8, 9, 10, 11, 12};

/* Selectively encrypt the bit position table to be encrypted for each byte in a frame */

byte highrate_level1_bitpos_list[]={0xf0, 0xff, 0xfd, 0x0b, 0xf0, 0x00,

0x0f, 0xf0, 0x00, 0x0f, 0xf0, 0x3f};

∥For high speed and security level 1, the selected bit position table

byte highrate_level2_bitpos_list[]={0xff, 0xff, 0xff, 0x0f, 0xfc, 0xc0,

0x0f, 0xfc, 0xc0, 0x0f, 0xf0, 0x3f};

∥For high speed and security level 2, the selected bit position table

byte highrate_level3_bitpos_list[]={0xff, 0xff, 0xff, 0x8f, 0xff, 0xf8,

0x8f, 0xff, 0xf8, 0x0f, 0xf0, 0x3f};

∥For high speed and security level 3, the selected bit position table

byte highrate_level4_bitpos_list[]={0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

0xff, 0xff, 0xff, 0x0f, 0xf0, 0x3f};

∥For high speed and security level 4, the selected bit position table

byte lowrate_level1_bitpos_list[]={0xf0, 0xff, 0xfd, 0x0b, 0xf0, 0x00,

0x0f, 0xf0, 0x00, 0x0f};

∥ When the rate is low and the security level is 1, the selected bit position table

byte lowrate_level2_bitpos_list[]={0xff, 0xff, 0xff, 0x0f, 0xfc, 0xc0,

0x0f, 0xfc, 0xc0, 0x0f};

∥ When the rate is low and the security level is 2, the selected bit position table

byte lowrate_level3_bitpos_list[]={0xff, 0xff, 0xff, 0x8f, 0xff, 0xf8,

0x8f, 0xff, 0xf8, 0x0f};

∥ When the rate is low and the security level is 3, the selected bit position table

byte lowrate_level4_bitpos_list[]={0xff, 0xff, 0xff, 0xff, 0xff, 0xff,

0xff, 0xff, 0xff, 0x0f};

∥ When the rate is low and the security level is 4, the selected bit position table

The concrete method process that step S230 of the present invention carries out encryption operation is described in detail below:

First of all, G.723.1 has three frame sizes: 24 bytes (6.4kbit/s frame), 20 bytes (5.3kbit/s frame) and 4 bytes, of which the 4-byte frame is the SID (Silent Insertion Indicator) , there is no limit on how the three frames are spaced apart. The frame size is reflected by the first two bits VADFLAG_B0 and RATEFLAG_B0 of each frame. The corresponding relationship is shown in Table 9:

Table 9 Correspondence between VADFLAG_B0 and RATEFLAG_B0 and the number of bits per frame RATEFLAG_B0 VADFLAG Bits per frame (bits) 0 0 192 (24 bytes) 0 1 160 (20 bytes) 1 0 32 (4 bytes) default default 8 (1 byte)

Therefore, it can be seen that the two bits of VADFLAG_B0 and RATEFLAG_B0 are used to distinguish the three types of frames of G.723.1 and cannot be encrypted.

According to the characteristics of this frame structure, it is necessary to reserve the frame header VADFLAG_B0 and RATEFLAG_B0 of each frame, so as to judge whether it belongs to a silent frame.

At the same time, in the G.723.1 encoding, since the effective data of the voice is included in the voice frames of 24 bytes and 20 bytes per frame, and the silence frame contains invalid data, there is no need to modify the silence when encrypting. Frames are encrypted, which reduces the amount of encrypted data.

In the embodiment of the present invention, as shown in Figure 8; Take the situation of high-speed voice frame and security level as 1 as an example to analyze and illustrate, the byte sequence number table to be encrypted is selected byte highrate_bytepos_list[] for use, and the bit position table is selected for use byte highrate_level1_bitpos_list[].

Among them, in the high-speed speech frame 71 (a frame of 24 bytes), the underlined italics indicate that this byte has bits that need to be encrypted, otherwise there is no bit.

Array 72 represents that length is a byte (byte) type array of 12, and the data stored in its first item corresponds to the third frame of the speech frame, and the first four digits are 1 to represent that the first four digits of the third frame of the speech frame need to be encrypted. Four bits are not needed, and so on.

The array keyList[8]73 table table Xi gets 64 binary digits after the decimal point, generates 8-byte byte type (byte) type data, and stores it in the array keyList[8]73.

Specifically include the following steps:

Step S231, at first receive the first frame of the speech frame, which is assumed to be a high-rate frame after being judged here, and the security level is 1, and the first item of the search array byte highrate bytepos_list[] is 3, indicating that the third word of the speech frame section has bits to encrypt;

Step S232, the first item of the array byte highrate_level1_bitpos_list[] records the bit that needs to be encrypted in the 3rd byte of the voice frame, takes out its value 11110000, and the bit is 1 to illustrate that the bit of the corresponding voice frame byte needs to be encrypted, otherwise not;

Step S233: AND 11110000 with the first item of the array keyList[] (storing the generated 64-bit key stream), XOR the result obtained with the third byte of the speech frame, and send the result back to the third word of the speech frame Festival;

Step S234, then look for the second item of the array byte highrate_bytepos_list[] to be 4, indicating that the 4th byte of the voice frame has a bit to be encrypted;

Step S235, the 2nd item of the array byte highrate_level1_bitpos_list[] records the bit that the 4th byte of the voice frame needs to be encrypted, and takes out its value 11111111;

Step S236, AND with the second item of the array keyList[], XOR the obtained result with the 4th byte of the speech frame, and send the result back to the 4th byte of the speech frame.

By analogy, until the 10th byte of the voice frame is encrypted, as shown in FIG. 9 .

Step S237, when the 11th byte of the voice frame needs to be encrypted, the keyList[] table returns to item 1 of the header, and uses this byte to perform XOR encryption with the 11th byte of the voice frame.

As shown in Figure 10, when proceeding to step S237, the 8th byte of keyList[] is also used up, and when the 11th byte of the voice frame needs to be encrypted, keyList[] table returns to the 1st byte of the header Item, use this byte to perform XOR encryption with the 11th byte of the voice frame. Because the bits to be encrypted for each byte of the voice frame are different, even if keyList[] is used repeatedly, the bits actually used each time are different.

Step S238, slide backward to continue encrypting the 12th byte, meanwhile, the keyList[] table also slides backward to the second item, and byte highrate_bytepos_list[] and byte highrate_level1_bitpos_list[] also move backward one item;

As shown in Figure 11, after the 11th byte of the voice frame is encrypted, slide backwards to continue encrypting the 12th byte. At the same time, the keyList[] table also slides backwards to the second item, byte highrate_bytepos_list[] and byte highrate_level1_bitpos_list[] also move backward one item;

By analogy, until the 14th byte of the voice frame is encrypted, the voice frame is encrypted, because there is no byte to be encrypted after the 14th byte, as shown in Figure 12.

Step S239, after the encryption of the current voice frame is completed, take the next frame of voice frame, and judge whether it is a silent frame, if it is a silent frame, then do not perform the encryption operation, and continue to take the next frame; if it is not a silent frame, then continue to the next step ;

At this time, as shown in FIG. 13 , the keyList[] table is used starting from the second item, and XOR operation is performed with the first byte to be encrypted (that is, byte 3) of the speech frame.

Subsequent operations are the same as the previous voice frame, the voice frame bytes, keyList[], byte highrate_bytepos_list[] and byte highrate_level1_bitpos_list[] all move backwards accordingly, and XOR encryption is performed byte by byte. Until the 8th item of keyList[] is used up, return to the 1st item of its header, and then continue XOR encryption with the next byte of the voice frame until the encryption of the voice frame is completed.

Then take the next non-silent voice frame, and use the keyList[] table from item 3. Similar to the above, perform XOR encryption with the voice frame byte by byte until the encryption of this frame is completed.

Then take the next non-silent voice frame, and use the keyList[] table from the 8th item, similar to the above, perform XOR encryption with the voice frame byte by byte until the encryption of this frame is completed.

At this time, 8 non-silent frames have been encrypted, and the keyList[] table is invalid and needs to be updated. Using formula (9)

G(x)＝(β+1)(1+1/β) ^β ×(1-x) ^β , (9)

Iterate over xx (the current value of x) to get x _i+1 , take the 64 effective binary digits after the decimal point of x _i+1 , convert it into 8 byte data, and store the converted data into the array keyList[] , to achieve the update of keyList[].

Continue to take a non-silent frame, and the keyList[] table will be used from item 1;

Continue to take a non-silent frame, and the keyList[] table will be used from item 2;

Continue to take a non-silent frame, and the keyList[] table will be used from item 3;

Repeat the above process until a non-silent frame is taken for the 8th time, and the keyList[] table is used from the 8th item.

In this way, 8 non-silent frames are encrypted, and _xi is iterated to obtain xi ₊₁ to update keyList[]. In this way, the next 8 non-silent frames are processed.

Until all voice frames are encrypted.

Speech coding selective encryption method of the present invention, uses selective encryption to carry out chaotic stream encryption to G.723.1 coded speech, has following advantages:

1) Improve the real-time performance of voice calls

The speech coding selective encryption method improves the encryption efficiency from the following aspects:

A1) Selectively encrypt voice frames. Between frames, only high/low rate frames carrying voice information are encrypted, and a large number of SID frames (silent frames) are not encrypted; within frames, according to the needs of the security level , to encrypt those bits with high sensitivity (that is, having a great influence on reconstructed speech). Compared with the traditional overall encryption of voice media content, this undoubtedly speeds up the encryption speed.

A2) Use the stream encryption algorithm to XOR the plaintext directly with the key stream. Compared with the block algorithm, this method is faster.

A3) In order to enhance security, the key stream used in the stream encryption algorithm is a chaotic sequence, which is obtained by the improved Logistics map mapping. In the iterative process of this method, the calculation complexity is relatively large. In order to reduce the calculation amount of the key stream, combined with the characteristics of the selected part of the G.723.1 frame structure that needs to be encrypted, the chaotic key stream is appropriately reused (specifically See examples for detailed explanation), without significantly affecting its safety performance.

2) Reduce system energy consumption and reduce resource occupation

Selectively encrypt voice frames, use stream encryption algorithm, and properly reuse chaotic key streams. These measures can not only improve efficiency, but also reduce system energy consumption and the amount of resources occupied. Reducing energy consumption is a very critical issue in some application environments where energy resources are very tight, such as the limited energy carried by mobile devices in the field of wireless communications. Reducing the occupation of resources can also reduce the load of network processing nodes.

3) It is convenient to meet the business needs of different levels

Selective encryption can not only meet better technical requirements, but also has practical application value in many Internet businesses. For example, for websites that provide various audio services, the voice materials can be provided to users with different encryption strengths. If they are provided for free to users for trial listening, a certain encryption strength can be used for the voice to perform obfuscation. Many applications like this, in short, selective encryption provides feasibility and good technical support for different levels of business needs.

The specific embodiments of the present invention have been described and illustrated above, and these embodiments should be considered as exemplary only, and are not used to limit the present invention, and the present invention should be interpreted according to the appended claims.

Claims (11)

1, a kind of speech coding selective encryption method is characterized in that, comprises the following steps: Step A, first perform initialization, select the security level required for encryption, and determine the adjustment parameters of the chaotic flow encryption method; Step B, according to the frame rate and security level, determine the bits of parameters in the encrypted byte sequence number table and bit position table, and use the chaotic stream encryption method to perform encryption operations. 2. The speech coding selective encryption method according to claim 1, further comprising the following steps: In step C, after the decryption party receives the encrypted data, it first performs initialization and then performs a decryption operation. 3. The speech coding selective encryption method according to claim 1 or 2, wherein said step A comprises the following steps: Step A1, the user selects the security level required for encryption; Step A2, determining the adjustment parameters of the chaotic flow encryption method; Step A3, get the effective binary bits of the initial value in the adjustment parameter, convert the byte data, store the converted data in an array as a key stream table, and perform XOR operation with the bits to be encrypted in the plain text of the voice data stream to get the ciphertext. 4. The speech coding selective encryption method according to claim 1 or 2, wherein said step B comprises the following steps: Step B1, judge whether it is a high-rate frame, a low-rate frame, or a silent frame according to the first two bits of a received voice frame; Step B2, according to different frames, go to the corresponding steps for processing: If it is a high-rate frame, select the byte sequence number table corresponding to the high rate and the bit position table corresponding to the required security level, and go to step B3; If it is a low-rate frame, select the byte sequence number table corresponding to the above low rate and the bit position table corresponding to the required security level, and go to step B3; If it is a silent frame, no encryption is required, and it returns to step B1 to process the next frame; Step B3, after determining the selected encrypted byte sequence number table and bit position table according to the frame rate and security level, perform the encryption operation of the current frame; Step B4, after completing the encryption operation of the current frame, if the key stream is used up, use the adjustment parameters in the chaotic stream encryption method to iterate the current chaotic sequence value to obtain a new key stream, and update the key stream table; If there are speech frames, go back to step B1 and process the next frame. 5. The speech coding selective encryption method according to claim 2, wherein said step C comprises the following steps: Step C1, receiving and decrypting the control parameters to obtain the encryption security level, and decrypting to obtain the adjustment parameters of the chaotic flow encryption method; Step C2, take the effective binary digits of the adjustment parameters, convert them into byte data, store the converted data into the byte array as the key stream table, and perform XOR operation with the bits to be encrypted in the plain text of the voice data stream, to obtain plaintext; Step C3, execute the decryption operation process of the current frame; Step C4, after completing the decryption operation of the current frame, if the key stream is used up, use the adjustment parameters to iterate the current chaotic sequence value to obtain a new key stream, and update the key stream table; if there are still speech frames, turn to Process the next frame. 6. The speech coding selective encryption method according to claim 1 or 2, characterized in that the chaotic flow encryption method is a chaotic flow encryption method represented by a one-dimensional nonlinear iterative method of Logistics mapping. 7. The speech coding selective encryption method according to claim 6, wherein the one-dimensional nonlinear iterative method of the Logistics map is an improved one-dimensional nonlinear iterative method of the Logistics map, and the iterative method is as follows Show: G(x)＝(β+1)(1+1/β) ^β x(1-x) ^β Among them, β∈(1,4), x ₀ ∈(0,1), the initial value of x is x ₀ , and x ₁ , x ₂ , x ₃ ,…x _n … can be obtained through iteration of this formula. 8. The speech coding selective encryption method according to claim 1 or 2, characterized in that, in said step B, the selected encrypted byte sequence number table and bit position table are determined according to the frame rate and the security level required for encryption ,for: Determine the sensitivity arrangement order of the bit bits of the parameters in the speech frame, so as to divide the bits into different categories, and encrypt the bits of different categories to obtain different security levels. 9. The speech coding selective encryption method according to claim 8, characterized in that the speech coding is speech coding of G.723.1 standard speech frames. 10. The speech coding selective encryption method according to claim 9, characterized in that, the order of sensitive arrangement of bits of parameters in the speech frame is five categories, namely CLASS1, CLASS2, CLASS3, CLASS4, and CLASS5 , and its importance decreases in turn. 11. The speech coding selective encryption method according to claim 10, characterized in that: When the security level is Level 1, the bits in CLASS1 are encrypted, and the 48bit bits in the voice frame are encrypted in the high rate mode of the G.723.1 standard voice frame; the voice is encrypted in the low rate mode of the G.723.1 standard voice frame 38bit bit encryption in the frame; When the security level is Level 2, the bits in CLASS1 and CLASS2 are encrypted, and the 62bit bits in the voice frame are encrypted in the high rate mode of the G.723.1 standard voice frame; 52bit encryption in voice frames; When the security level is Level 3, encrypt the bits in CLASS1, CLASS2 and CLASS3, and encrypt 74 bits in the voice frame in the high rate mode of the G.723.1 standard voice frame; in the low rate mode of the G.723.1 standard voice frame In this mode, the 64bit bits in the voice frame are encrypted; When the security level is Level 4, the bits in CLASS1, CLASS2, CLASS3 and CLASS4 are encrypted, and the 86bit bits in the voice frame are encrypted in the high-speed mode of the G.723.1 standard voice frame; in the G.723.1 standard voice frame Encrypt 76 bits in voice frames in low rate mode.

Images