CN112614498B - A method of modulating low-speed data into speech-like signals - Google Patents

Abstract

The invention discloses a method for modulating low-speed data into a speech-like signal, which comprises the steps of constructing a low-speed data stream through a speech-based excitation-formant generation model; generating a key stream by a stream cipher algorithm through a key updated at fixed time, obtaining a segmented ciphertext data stream by encryption operation with a low-speed data stream with a fixed length obtained by segmentation, obtaining a signal frame to be modulated by framing operation, and obtaining a speech-like signal by speech-like modulation operation of a speech-based excitation-formant generation model by the signal frame; and the voice-like signal received at the demodulation end is processed to obtain a low-speed data stream. The invention can modulate low-speed data on the excitation signal and a plurality of formant frequencies to generate a similar voice signal, has the parameter characteristics of a common voice signal, can transmit in a voice channel with a vocoder, and further realizes the low-speed data transmission in the voice channel of the vocoder.

Description

A method of modulating low-speed data into speech-like signals

Technical field

The present invention relates to the technical field of speech-like modulation, and in particular, to a method of modulating low-speed data into a speech-like signal.

Background technique

Voice communication is an indispensable way of daily information exchange, and incidents of eavesdropping on mobile phones and telephones occur from time to time. After a Trojan is installed on a mobile phone, the user's calls can easily be eavesdropped through the wireless network; phone voice monitoring can be achieved by paralleling the phone lines. If the analog voice can be directly encrypted, the voice information can be protected before the voice is sent, preventing the voice from being eavesdropped during subsequent processing and transmission. However, encryption and decryption of analog voice has always been one of the difficulties in information security research. The difficulty lies in how to enable the encrypted voice to still be transmitted through the voice channel. Because usually the voice characteristics are lost after voice encryption. In most voice transmission systems, in order to reduce the channel bandwidth required for voice transmission, voice is usually compressed and encoded. If the encrypted voice loses its voice characteristics, it cannot be effectively transmitted through such channels.

Existing analog voice encryption technologies mainly include time domain scrambling and frequency domain scrambling. Among them, the time domain scrambling method divides a speech signal into several segments, and encrypts the speech by randomly shuffling the order of each segment. The advantage of this method is that it is simple to operate and the decrypted speech is highly intelligible. However, the security of this method is very low, especially it cannot resist slow speech attacks; the encryption and decryption delay is very large, and the decrypted speech also contains periodic interference sounds. The frequency domain scrambling method is to transform a frame of speech into the frequency domain, and scramble it directly or with molecular bands in the frequency domain. The advantage of this method is that the security intensity is higher than that of the time domain scrambling method, but the characteristics of the encrypted speech will be lost to a certain extent, which is not conducive to transmission through the speech compression encoding channel.

Modulation technology converts digital signals into a certain form to adapt to channel transmission. The general modulation method is to change a sinusoidal carrier signal and use the typical characteristics of the carrier: amplitude, frequency and phase to carry digital data. However, this data modem cannot be used in systems that use speech codec compression techniques. The main purpose of speech coding is to reduce the bits required to represent speech while still maintaining acceptable speech quality when decoding is restored, making it sound almost the same to the original speech to the human ear. However, the human ear has a hearing "threshold" that can block out differences in signal sampling waveforms. Therefore, the sampled waveform of the speech signal restored by decoding may be significantly different from the original signal, which causes greater distortion in the frequency, phase and amplitude characteristics of the data-bearing data output by the modem. Therefore, most modems cannot work on channels using voice codecs, and a modem scheme needs to be redesigned, a voice-like modulation and demodulation scheme.

Contents of the invention

Purpose of the invention: In order to overcome the deficiencies in the existing technology, the present invention provides a method for modulating low-speed data into a speech-like signal. The invention constructs a speech-like signal modulated by a speech-based excitation-formant model. It has the parameter characteristics of general speech information, is conducive to preserving the characteristics after transmission through the vocoder speech channel, and has good ability to resist non-linear operations of speech compression and decompression.

Technical solution: In order to achieve the above-mentioned object of the invention, the present invention provides a method for modulating low-speed data into a speech-like signal, which includes the following steps:

Step 1: Construct a low-speed data stream through a speech-based excitation-formant generation model. The low-speed data stream generates a speech-like signal capable of transmitting low-speed data in the vocoder channel through the model, including encryption and decryption operations, framing and deframing. operations and voice-like modem operations;

Step 2: The regularly updated key generates a key stream through the stream cipher algorithm, and is segmented to obtain a fixed-length low-speed data stream through an encryption operation to obtain a segmented ciphertext data stream. The segmented ciphertext data stream and key The sequence number and the frame header including the blank frame and the channel detection frame are processed through a framing operation to obtain the signal frame to be modulated. The signal frame is obtained through a speech-like modulation operation based on the speech excitation-formant generation model to obtain a speech-like speech with general speech signal parameter characteristics. Signal;

Step 3: Perform frame synchronization and speech-like demodulation operations on the speech-like signal received at the demodulation end to obtain framed data. After deframing, the key serial number and encrypted data are obtained to obtain the corresponding key. The key passes through the stream The cryptographic algorithm generates a key stream, which is decrypted with the encrypted data to obtain a low-speed data stream.

Further, in the present invention: the speech-like modulation operation adopts multi-frequency point amplitude modulation, selects the pitch frequency, initial frequency, number and position of formant frequency points according to the vocoder channel, and performs spectrum-based processing according to the frame structure of the signal frame. Amplitude modulation of the mid-formant frequency point is used to obtain a speech-like signal with general speech signal parameter characteristics.

Further, in the present invention: the frame structure of the signal frame includes a blank frame for synchronous channel vocoder energy detection, a word interval of the synchronous speech system, and a channel detection frame for frequency point amplitude reference, according to the changing characteristics of the channel conditions , for the receiving end to detect the distortion of each formant frequency point in the channel.

Further, in the present invention: the frame structure of the signal frame includes a blank frame, a channel detection frame and several data frames, and the number of bits carried by each data frame is its number of formants multiplied by the number of formants modulated. The number of bits, the same signal frame contains the key sequence number of the corresponding data segment.

Further, in the present invention: the key sequence number is used to update the key used in the segmented ciphertext data stream.

Further, in the present invention: the bit rate carried by the voice-like signal is greater than or equal to 2.4Kb/s.

Further, in the present invention: the encryption and decryption operations also include:

At the sending end, the corresponding subkey group sequence is generated through the key distributed by the upper layer, and is sorted according to the sequence number. Each segment generates a random key stream with the same bits as the data according to the subkey of the key group corresponding to the sequence number. , perform XOR operation with low-speed data to obtain an encrypted data stream;

At the receiving end, the encrypted data stream and the corresponding key group sequence number are obtained, the corresponding sub-key is obtained according to the sequence number, and the key stream is generated for decryption to obtain the transmitted low-speed data.

Beneficial effects: Compared with the prior art, the beneficial effects of the present invention are:

(1) The speech-like signal modulated by the speech excitation-formant model has the parameter characteristics of general speech information, which is beneficial to preserving the characteristics after transmission through the vocoder speech channel, and has good resistance to nonlinear operations of speech compression and decompression;

(2) The stream cipher algorithm does not introduce additional bit error rates, errors will not propagate, and the strength meets the requirements;

(3) The encryption and decryption delay is small, the framing and deframing speed is fast, and it does not affect the transmission of low-speed data.

The method proposed by the present invention to modulate low-speed data into a speech-like signal not only changes intelligible speech into incomprehensible speech, but also retains the speech characteristics, so that the encrypted speech can be smoothly transmitted through the speech compression encoding channel. .

Description of drawings

Figure 1 is a schematic flow chart of the method of modulating low-speed data into a speech-like signal according to the present invention;

Figure 2 is a schematic diagram of the frame structure format of the signal frame in the present invention;

Figure 3 is a schematic spectrum diagram of the channel detection frame in the present invention;

Figure 4 is a schematic spectrum diagram of a single data frame in the present invention.

Detailed ways

The technical solution of the present invention will be further described in detail below in conjunction with the accompanying drawings:

The invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

As shown in Figure 1, it is an overall flow diagram of a method for modulating low-speed data into a speech-like signal proposed by the present invention. The method specifically includes the following steps:

Step 1: The low-speed data stream passes through the speech-based excitation-formant generation model to generate a speech-like signal that can transmit low-speed data in the vocoder channel, including encryption and decryption operations, framing and deframing operations, and speech-like modulation and demodulation. operate;

Specifically, the encryption and decryption operations are used to realize confidential transmission of low-speed data, using a stream cipher algorithm. The encryption and decryption operations also include the following steps:

The regularly updated key generates a key stream through a stream cipher algorithm, and is segmented to obtain a fixed-length low-speed data stream through an encryption operation to obtain a segmented ciphertext data stream. The segmented ciphertext data stream, key sequence number and include The frame headers of the blank frame and the key number frame undergo a framing operation to obtain the signal frame to be modulated. The signal frame undergoes a speech-like modulation operation based on the speech excitation-formant generation model to obtain a speech-like signal with general speech signal parameter characteristics;

The speech-like signal received at the demodulation end performs frame synchronization and speech-like demodulation operations to obtain framed data. After deframing, the key sequence number and encrypted data are obtained to obtain the corresponding key. The key is generated through a stream cipher algorithm. The key stream and the encrypted data undergo a decryption operation to obtain a low-speed data stream.

Specifically, at the sending end, the corresponding key group sequence is generated through the key k _s distributed by the upper layer, and is sorted according to the sequence number. Each segment generates the same bits as the data according to the subkey of the sequence number corresponding to the key group. The random key stream k _i is XORed with the low-speed data to obtain an encrypted data stream.

At the receiving end, the encrypted segmented data stream and the corresponding key group sequence number are obtained, and the corresponding subkey is obtained according to the sequence number, and the key stream k _i is generated for decryption to obtain the transmitted low-speed data. The key sequence number is used to update the key used by the segmented ciphertext data stream.

Assume that each frame transmits m bits, the data to be encrypted at the sending end is Mi ( _i =1,...,m), the encrypted data is C _i (i=1,...,m), and the encrypted data received at the receiving end is C′ _i (i=1,...,m), the decrypted data is M′ _i (i=1,...,m), it should be possible to get M′ _i =M _i , then:

Framing and deframing operations are used to frame and integrate data, including adding blank frames and channel detection frames at the transmitter, and performing data framing and adding generated key serial numbers for voice-like modulation operations; decryption at the receiver The operation provides channel detection results and generates key sequence numbers, and integrates data frames.

It includes the blank frame for synchronous channel vocoder energy detection, the word interval of the synchronous speech system, and the channel detection frame for frequency point amplitude reference. According to the changing characteristics of the channel conditions, the receiving end can detect the distortion of each formant frequency point in the channel. Referring to the diagram in Figure 2, the frame structure format of each signal frame is composed of a blank frame of T milliseconds, a channel detection frame and n data frames. Each signal frame contains a total of n+2 frames of data. Each frame time is T*(n+2) milliseconds.

Specifically, the blank frame is used for energy detection of the synchronous channel vocoder, and the channel detection frame is used by the receiving end to detect the distortion of each formant frequency point of the channel. Each data frame contains N _f formant frequency points, each The frequency point contains kbit data, then a single data frame contains N _f *k bit data, where mbit is the transmission key sequence number, then the total data amount transmitted by each signal frame is S(t)=(N _f *km)*nbit, so Its transmission rate v(t) is:

v(t)＝S(t)/T _all ＝(N _f *km)*n/T*(n+2)(bit/ms)

Further, the voice-like modulation and demodulation operation is used to ensure effective transmission of signals in the vocoder channel. At the transmitting end, the framed data is subjected to voice-based excitation-formant generation model multi-frequency amplitude modulation (MFAK)-like speech modulation, and the low-speed data is modulated into speech-like; at the receiving end, the data transmitted through the vocoder channel is The speech-like signal is demodulated into framed data. In this embodiment, the speech-like modulation operation adopts multi-frequency point amplitude modulation. The pitch frequency, initial frequency, number and position of formant frequency points are selected according to the vocoder channel, and the formant frequency points in the spectrum are performed according to the frame structure of the signal frame. Amplitude modulation to obtain a speech-like signal with general speech signal parameter characteristics.

Furthermore, the bit rate carried by the voice-like signal is greater than or equal to 2.4Kb/s.

The speech-like modulation based on the speech excitation-formant generation model includes a channel detection frame, and the distortion of each formant frequency point is detected at the receiving end as a reference for data demodulation. Referring to the diagram in Figure 3, which is a spectrum diagram of a channel detection frame, the difference between the formant frequencies is the fundamental frequency. The fundamental frequency is determined based on the sampling rate and the length of the signal frame, and the fundamental frequency should be greater than 60Hz and less than 600Hz.

Assume that the fundamental frequency is Δf, the initial frequency is f ₀ , and the number of formant frequency points is N _f , then each formant frequency is f(i):

f(i)=f ₀ +i*Δf, i=0, 1, 2,..., N _f -1

The channel detection frame contains the sum of unit sine waves of each formant frequency during speech-like modulation, and the number of formant frequency points is N _f .

Referring to the diagram in Figure 4, it is a spectrum diagram of a single data frame. The data in the data frame is amplitude modulated based on the formant frequency points in the spectrum at the modulation end. Each data frame contains N _f formant frequency points, and each frequency The points contain kbit data, and the gain modulation range is -2 ^k-1 ~ 2 ^k-1 dB.

_{Suppose _ _ _}

E(f _i )=H(f _i )*10^(X(f _i )/20)

Among them, H(fi ₎ is the spectrum amplitude of _fi in the channel detection frame of the transmitter.

The frequency domain signal E(f) of a single data frame is:

The corresponding demodulation end demodulates the data according to the unit formant amplitude in the received channel detection frame:

X'(f _i )=20*log10(E'(f _i )/H'(f _i ))

Among them, H'( _fi ) is the spectrum amplitude of _fi in the channel detection frame of the receiving end.

Step 2: The regularly updated key generates a key stream through the stream cipher algorithm, and is segmented to obtain a fixed-length low-speed data stream through an encryption operation to obtain a segmented ciphertext data stream. The segmented ciphertext data stream and key The sequence number and the frame header including the blank frame and the channel detection frame are processed through a framing operation to obtain the signal frame to be modulated. The signal frame is obtained through a speech-like modulation operation based on the speech excitation-formant generation model to obtain a speech-like speech with general speech signal parameter characteristics. Signal;

Step 3: Perform frame synchronization and speech-like demodulation operations on the speech-like signal received at the demodulation end to obtain framed data. After deframing, the key serial number and encrypted data are obtained to obtain the corresponding key. The key passes through the stream The cryptographic algorithm generates a key stream, which is decrypted with the encrypted data to obtain a low-speed data stream.

It should be noted that the above-described embodiments only express some implementations of the present invention, and their descriptions should not be construed as limiting the patent scope of the present invention. It should be noted that for those of ordinary skill in the art, several improvements can be made without departing from the concept of the present invention, and these should all fall within the protection scope of the present invention.

Claims (1)

1. A method of modulating low-speed data into a speech-like signal, characterized by: including the following steps, Step 1: Construct a low-speed data stream through a speech-based excitation-formant generation model. The low-speed data stream generates a speech-like signal capable of transmitting low-speed data in the vocoder channel through the model, including encryption and decryption operations, framing and deframing. operations and voice-like modem operations; Step 2: The regularly updated key generates a key stream through the stream cipher algorithm, and is segmented to obtain a fixed-length low-speed data stream through an encryption operation to obtain a segmented ciphertext data stream. The segmented ciphertext data stream and key The sequence number and the frame header including the blank frame and the channel detection frame are processed through a framing operation to obtain the signal frame to be modulated. The signal frame is obtained through a speech-like modulation operation based on the speech excitation-formant generation model to obtain a speech-like speech with general speech signal parameter characteristics. Signal; Step 3: Perform frame synchronization and speech-like demodulation operations on the speech-like signal received at the demodulation end to obtain framed data. After deframing, obtain the key serial number and encrypted data to obtain the corresponding key. The key passes through the stream The cryptographic algorithm generates a key stream, which is decrypted with the encrypted data to obtain a low-speed data stream; The speech modulation operation adopts multi-frequency point amplitude modulation. The pitch frequency, initial frequency, number and position of formant frequency points are selected according to the vocoder channel, and amplitude modulation based on the formant frequency point in the spectrum is performed according to the frame structure of the signal frame. , obtain a speech-like signal with general speech signal parameter characteristics; The frame structure of the signal frame includes a blank frame for synchronous channel vocoder energy detection, a word interval for the synchronous speech system, and a channel detection frame for frequency point amplitude reference. According to the changing characteristics of the channel conditions, the receiving end can detect each signal in the channel. The distortion of each formant frequency point; The frame structure of the signal frame includes a blank frame, a channel detection frame and several data frames, and the number of bits carried by each data frame is its number of formants multiplied by the number of bits modulated by each formant. The same signal frame contains the corresponding The key sequence number of the data segment; The key sequence number is used to update the key used by the segmented ciphertext data stream; The bit rate carried by the voice signal is greater than or equal to 2.4Kb/s; The encryption and decryption operations also include: At the sending end, the corresponding subkey group sequence is generated through the key distributed by the upper layer, and is sorted according to the sequence number. Each segment generates a random key stream with the same bits as the data according to the subkey of the key group corresponding to the sequence number. , perform XOR operation with low-speed data to obtain an encrypted data stream; At the receiving end, the encrypted data stream and the corresponding key group sequence number are obtained, the corresponding sub-key is obtained according to the sequence number, and the key stream is generated for decryption to obtain the transmitted low-speed data.