A homomorphic encryption audio watermarking system in a cloud computing environment

Abstract

The application provides a homomorphic encryption audio watermarking system and method in a cloud computing environment, relates to the technical field of audio watermarking, and is characterized in that the user end, the privacy server and the cloud server cooperate with each other, the user end applies to the privacy server for obtaining a public key and a private key, and the public key is notified to the cloud server, the user end encrypts audio and embeds a watermark by using the public key, and uploads the audio to the cloud server for storage. The server distributes the watermark to the user end, the user end decrypts the encrypted audio with the watermark and extracts the watermark, the user end can first decrypt and then extract the watermark, or directly extracts the watermark in the ciphertext domain without decryption, the watermark has high security and privacy, the extraction efficiency is high, the watermark authentication efficiency is high, and the multimedia data is provided with multiple security protections.

Description

Technical Field

[0001] This invention relates to the technical field of audio watermarking, and more specifically, to a homomorphic encrypted audio watermarking system and method in a cloud computing environment. Background Technology

[0002] In today's open, interconnected, and information-rich era, people's audio and other multimedia data are inadvertently swept up in the internet wave. As this audio data is disseminated online, it may carry a large amount of our private information and may also be stolen by others. Whether intentional or unintentional, the use of other people's audio-visual works may infringe on their copyrights.

[0003] The emergence of digital watermarking and steganography has provided strong technical support for privacy and copyright protection. However, in some special application scenarios (such as the military and medical fields), higher requirements have been placed on data protection, with a desire to protect data without disclosing the original data. With the advent of homomorphic encryption technology and the exploration of its applications, it has become possible to implement watermarking algorithms based on homomorphic encryption.

[0004] Homomorphic encryption is an emerging cryptographic technique. Compared to ordinary encryption techniques, it possesses operational transferability, enabling the mapping of certain operations in the plaintext domain to certain operations in the ciphertext domain. Common examples include additive homomorphism and multiplicative homomorphism. While many watermarking algorithms combining encryption and watermarking techniques exist, such as the existing method for exchanging cryptographic watermarks based on homomorphic properties in the same operation domain, this method leverages the homomorphic properties of cryptographic algorithms to achieve mutual mapping between the ciphertext and plaintext domains of multimedia information data. Specifically, it can convert the digital watermark embedding algorithm operations in the plaintext domain into corresponding operations in the ciphertext domain, allowing the ciphertext domain to perform the plaintext domain watermark embedding algorithm operations. Simultaneously, utilizing the single-plaintext, multiple-ciphertext characteristic of cryptographic algorithms, the plaintext domain watermark information is mapped to the ciphertext domain to achieve ciphertext domain watermark embedding and extraction. However, some issues still warrant attention during the watermark embedding and extraction process. For example, encrypting the multimedia data first and then embedding the watermark in the ciphertext may result in the ciphertext being undecryptable. Embedding the watermark first and then encrypting it means the watermark information cannot be directly extracted from the ciphertext field; it can only be extracted after decryption. These issues make the watermark extraction process redundant, inefficient, and insecure. Summary of the Invention

[0005] To address the issues of low watermark security and privacy, and low extraction efficiency in current audio watermark embedding and extraction processes, this invention proposes a homomorphic encrypted audio watermarking system and method in a cloud computing environment. This system enables exchangeable encrypted watermarks on audio, providing multifaceted security protection for multimedia data.

[0006] To achieve the above-mentioned technical effects, the technical solution of the present invention is as follows:

[0007] A homomorphic encrypted audio watermarking system in a cloud computing environment includes:

[0008] The user-side is used to request homomorphic encryption public and private keys from the privacy server, encrypt audio to obtain encrypted watermarked audio, decrypt the encrypted watermarked audio, and perform authentication.

[0009] A privacy server provides homomorphic encryption public and private keys to the user's client.

[0010] The cloud server is used to store encrypted watermarked audio uploaded by users and also to distribute the encrypted watermarked audio to users.

[0011] Preferably, the homomorphic encryption is Paillier additive homomorphic.

[0012] This application also proposes a homomorphic encrypted audio watermarking method based on the aforementioned homomorphic encrypted watermarking system in a cloud computing environment, comprising the following steps:

[0013] S1. The user client requests and obtains the public key pk and the private key sk from the privacy server;

[0014] S2. The user transmits the public key pk to the cloud server, and uses the public key pk to embed encrypted watermarks into each sampling point of the audio before uploading it to the cloud server.

[0015] S3. The user terminal uses the private key sk to decrypt the encrypted watermarked audio and extract the watermark, and then performs authentication; the encrypted watermarked audio is downloaded by the user terminal from the cloud server.

[0016] Preferably, the audio is in WAV file format, with a sampling bit depth of 8 bits or 16 bits. The 8-bit sampling bit depth uses PCM pulse code modulation, is unsigned, and the sampling data range is 0–255. An embedded watermark value of d = 1 is selected. The specific preprocessing method is as follows: for sampling points with a value less than 1, all sampling values ​​are set to 1; for sampling points with a value greater than 245, the sampling value is set to 245; the sampling values ​​of the remaining sampling points remain unchanged.

[0017] The 16-bit sampling depth uses PCM pulse code modulation and is a signed number. The sampling data range is -32768 to 32767. The watermark value d = 1 is selected. The specific preprocessing method is as follows: for sampling points with a sampling value less than -32767, the sampling value is set to -32767; for sampling points with a sampling value greater than 32703, the sampling point is set to 32703; the sampling values ​​of the remaining sampling points remain unchanged.

[0018] Preferably, each audio sampling point includes floating-point numbers and negative numbers. When the user terminal uses the public key pk to embed an encrypted watermark of the floating-point numbers in each audio sampling point, the floating-point numbers in each sampling point are multiplied by a large integer Q, then rounded down, and finally encrypted. The large integer Q is a precision control coefficient. When the user terminal uses the public key pk to embed an encrypted watermark of the negative numbers in each audio sampling point, a modular arithmetic processing method is used. The process is as follows: the latter half of the original encryption range [0, n) of the public key pk is encrypted. Indicates the range of negative number encryption Make the encryption range become n represents the largest positive integer that can be encrypted.

[0019] Based on the above technical means, since the Paillier encryption method is only applicable to positive integer encryption, a transformation process is performed when encrypting floating-point numbers and negative numbers in each audio sampling point.

[0020] Preferably, when sampling audio, two sets of sampling points are randomly selected from the audio data segment: sampling point set A and sampling point set B, and sampling points are randomly selected from the two sets of sampling points.

[0021] Suppose that the number of sampling points in both sets of sampling points is N; the sampling value of the i-th sampling point in set A is a. i The sampled value of the i-th sampled point in the sampled point set B is b. i Let θ be the difference between the average value of the sampled data in sampled data set A and the average value of the sampled data in sampled data set B;

[0022] When using a public key (PK) to embed an encrypted watermark at each sampling point of the audio, the process of first embedding the watermark and then encrypting includes:

[0023] S201. Increase the sampled values ​​of all sampled points in sampled point set A by d, and decrease the sampled values ​​of all sampled points in sampled point set B by d;

[0024] S202. Let a im b im These are the sampled values ​​from sample point set A and sample point set B after watermark embedding, respectively:

[0025] aim =a i +d

[0026] b im =b i -d

[0027] S203. Encrypt the embedded watermark, let a ime b ime The sampled value obtained by embedding the watermark and then encrypting it is expressed as:

[0028]

[0029] Where r1 and r2 are random parameters, a ime ≥b ime .

[0030] Based on the aforementioned technical means, due to the additive homomorphic property of the Paillier encryption algorithm, the operations of embedding watermarks and encryption can be interchanged, ultimately resulting in a ciphertext carrier with the same watermark information. To ensure successful subsequent watermark extraction, appropriate random parameters r1 and r2 are selected to guarantee that a ime ≥b ime .

[0031] Preferably, when sampling audio, two sets of sampling points are randomly selected from the audio data segment: sampling point set A and sampling point set B, and sampling points are randomly selected from the two sets of sampling points.

[0032] Suppose that the number of sampling points in both sets of sampling points is N; the sampling value of the i-th sampling point in set A is a. i The sampled value of the i-th sampled point in the sampled point set B is b. i Let θ be the difference between the average value of the sampled data in sampled data set A and the average value of the sampled data in sampled data set B;

[0033] When using a public key (PK) to embed a watermark into each sample point of audio, the process involves first encrypting the audio and then embedding the watermark, which includes:

[0034] S21. Let a ie b ie The encrypted value:

[0035]

[0036]

[0037] S22. Embed a watermark in the ciphertext field, let a iem b iemTo embed the watermark value after encryption, the embedded value d is first encrypted, resulting in the encrypted value d. e :

[0038]

[0039] The sampling points in the sampling point set A satisfy:

[0040] a iem =a ie d e mod n 2

[0041] The sampling points in the sampling point set B satisfy:

[0042] b iem =b ie (d e ) -1 mod n 2

[0043] Among them, r1, r2, r d All are random parameters, a iem ≥b iem .

[0044] Preferably, the user terminal receives the encrypted watermarked audio from the cloud service, decrypts the encrypted watermarked audio, and then extracts the watermark. Alternatively, the user terminal receives the encrypted watermarked audio from the cloud service without decryption, and when extracting the watermark in the ciphertext field, the plaintext value obtained after decryption is the same.

[0045] Preferably, the user receives encrypted watermarked audio from the cloud service, decrypts the encrypted watermarked audio, and then extracts the watermark, assuming a... id b id The plaintext value obtained by decrypting the ciphertext in sampling point set A and sampling point set B satisfies:

[0046]

[0047] The user client authenticates the decrypted audio and watermark. The process is as follows:

[0048] S401. Calculate the difference s between the average value of the sampled data in sampled data set A and the average value of the sampled data in sampled data set B after decryption. The expression is:

[0049]

[0050] Where N represents the total number of sampled data;

[0051] S402. Use the average difference s as the watermark detection value. If s≈θ+2d, it indicates that the watermark information is contained; otherwise, it does not contain watermark information.

[0052] Preferably, the user terminal receives encrypted watermarked audio from the cloud service, does not decrypt the encrypted watermarked audio, and extracts the watermark from the ciphertext field. Let a id b id The plaintext value obtained by decrypting the ciphertext in sampling point set A and sampling point set B satisfies:

[0053] a id =D(a) iem )=D(a ime )

[0054] b id =D(b) iem )=D(b ime )

[0055] The user client authenticates the decrypted audio and watermark. The process is as follows:

[0056] S41. During the watermark embedding process, a is satisfied. iem ≥b iem Under the premise of extracting the watermark from the ciphertext field, using s as the watermark detection information, the expression is:

[0057]

[0058] in,

[0059] S42. If s≈1, it indicates that the watermark information is present; otherwise, it does not contain watermark information.

[0060] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0061] This invention proposes a homomorphic encrypted audio watermarking system and method in a cloud computing environment. Through cooperation between the user terminal, a privacy server, and a cloud server, the user terminal requests a public and private key from the privacy server and shares the public key with the cloud server. The user terminal uses the public key to encrypt the audio and embed a watermark, then uploads it to the cloud server for storage. The server distributes the watermark to the user terminal, which decrypts the encrypted watermarked audio and extracts the watermark. The user terminal can either decrypt first and then extract the watermark, or extract the watermark directly from the encrypted domain without decryption. The watermark offers high security and privacy, high extraction efficiency, and high watermark authentication efficiency, providing multifaceted security protection for multimedia data. Attached Figure Description

[0062] Figure 1This diagram illustrates the structure of the homomorphic encrypted audio watermarking system in a cloud computing environment proposed in Embodiment 1 of the present invention.

[0063] Figure 2 A flowchart illustrating the homomorphic encrypted audio watermarking method in a cloud computing environment proposed in Embodiment 2 of the present invention;

[0064] Figure 3 This represents a waveform diagram of an audio signal as described in Embodiment 3 of the present invention;

[0065] Figure 4 This represents the waveform diagram after watermarking and encryption of audio using the method proposed in this application, as shown in Embodiment 3 of the present invention.

[0066] Figure 5 This diagram shows the waveform of audio after encryption and watermark embedding using the method proposed in this application, as described in the embodiments of the present invention.

[0067] Figure 6 This indicates that the method proposed in this application is used to first decrypt the encrypted watermarked audio and then extract the waveform of the watermarked audio. Detailed Implementation

[0068] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent.

[0069] To better illustrate this embodiment, some parts of the accompanying drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions;

[0070] It is understandable to those skilled in the art that some well-known details may be omitted from the accompanying drawings.

[0071] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0072] The positional relationships depicted in the accompanying drawings are for illustrative purposes only and should not be construed as limiting this patent.

[0073] Example 1

[0074] This embodiment proposes a homomorphic encrypted audio watermarking system in a cloud computing environment. The system's structure diagram is shown below. Figure 1 ,include:

[0075] The user-side is used to request homomorphic encryption public and private keys from the privacy server, encrypt audio to obtain encrypted watermarked audio, decrypt the encrypted watermarked audio, and perform authentication.

[0076] A privacy server provides homomorphic encryption public and private keys to the user's client.

[0077] The cloud server is used to store encrypted watermarked audio uploaded by users and also to distribute the encrypted watermarked audio to users.

[0078] In this embodiment, the homomorphic encryption is Paillier additive homomorphic encryption. Overall, the user client requests a Paillier public key (pk) and private key (sk) from the privacy server and provides the public key (pk) to the cloud server. The user client uses the public key (pk) to encrypt the audio and embed a watermark, then uploads it to the cloud server for storage. The server distributes the watermark to the user client, which decrypts the encrypted watermarked audio and extracts the watermark to complete authentication. The watermark offers high security and privacy, high extraction efficiency, and high authentication efficiency, providing multifaceted security protection for multimedia data.

[0079] Example 2

[0080] This embodiment proposes a homomorphic encrypted audio watermarking method based on the homomorphic encrypted watermarking system in the cloud computing environment described in Embodiment 1. The flowchart of this method is shown below. Figure 2 This includes the following steps:

[0081] S1. The user client requests and obtains the public key pk and the private key sk from the privacy server;

[0082] S2. The user transmits the public key (PK) to the cloud server, and uses the public key (PK) to embed encrypted watermarks at each sampling point of the audio before uploading it to the cloud server.

[0083] S3. The user terminal uses the private key sk to decrypt the encrypted watermarked audio and extract the watermark, and then performs authentication; the encrypted watermarked audio is downloaded by the user terminal from the cloud server.

[0084] In this embodiment, the audio is in WAV file format, and the sampling point bit depth is 8 bits or 16 bits. The 8-bit sampling bit depth uses PCM pulse code modulation, which is an unsigned number, and the sampling data range is 0 to 255. The watermark value d = 1 is selected. The specific preprocessing method is as follows: for sampling points with a sampling value less than 1, the sampling value is set to 1; for sampling points with a sampling value greater than 245, the sampling value is set to 245; the sampling values ​​of the remaining sampling points remain unchanged.

[0085] The 16-bit sampling depth uses PCM pulse code modulation and is a signed number. The sampling data range is -32768 to 32767. The watermark value d = 1 is selected. The specific preprocessing method is as follows: for sampling points with a sampling value less than -32767, the sampling value is set to -32767; for sampling points with a sampling value greater than 32703, the sampling point is set to 32703; the sampling values ​​of the remaining sampling points remain unchanged.

[0086] Since the Paillier encryption method only applies to positive integers, a transformation process is required when encrypting floating-point numbers and negative numbers in each audio sample point. Specifically: when the user terminal uses the public key pk to embed an encrypted watermark of floating-point numbers in each audio sample point, the floating-point numbers in each audio sample point are multiplied by a large integer Q, then rounded down, and finally encrypted. The large integer Q is a precision control coefficient. When the user terminal uses the public key pk to embed an encrypted watermark of negative numbers in each audio sample point, a modulo operation method is used. The process is as follows: the latter half of the original encryption range [0, n) of the public key pk is encrypted. Indicates the range of negative number encryption Make the encryption range become n represents the largest positive integer that can be encrypted.

[0087] In this embodiment, when sampling audio, two sets of sampling points are randomly selected from the audio data segment: sampling point set A and sampling point set B, and sampling points are randomly selected from the two sets of sampling points.

[0088] Suppose that the number of sampling points in both sets of sampling points is N; the sampling value of the i-th sampling point in set A is a. i The sampled value of the i-th sampled point in the sampled point set B is b. i Let θ be the difference between the average value of the sampled data in sampled data set A and the average value of the sampled data in sampled data set B;

[0089] Due to the additive homomorphic property of the Paillier encryption algorithm, the operations of embedding watermarks and encryption can be interchanged, resulting in a ciphertext carrier with the same watermark information. When embedding encrypted watermarks at each sampling point of audio using the public key pk, the process of first embedding the watermark and then encrypting includes:

[0090] S201. Increase the sampled values ​​of all sampled points in sampled point set A by d, and decrease the sampled values ​​of all sampled points in sampled point set B by d;

[0091] S202. Let a im b im These are the sampled values ​​from sample point set A and sample point set B after watermark embedding, respectively:

[0092] a im =a i +d

[0093] b im =b i -d

[0094] S203. Encrypt the embedded watermark, let a ime b imeThe sampled value obtained by embedding the watermark and then encrypting it is expressed as:

[0095]

[0096] Where r1 and r2 are random parameters, a ime ≥b ime To ensure the smooth extraction of the watermark later, appropriate random parameters r1 and r2 are selected to ensure that a ime ≥b ime Due to the randomness of r1 and r2, it is guaranteed that a ime ≥b ime It is feasible.

[0097] When using a public key (PK) to embed a watermark into each sample point of audio, the process involves first encrypting the audio and then embedding the watermark, which includes:

[0098] S21. Let a ie b ie The encrypted value:

[0099]

[0100]

[0101] S22. Embed a watermark in the ciphertext field, let a iem b iem To embed the watermark value after encryption, the embedded value d is first encrypted, resulting in the encrypted value d. e :

[0102]

[0103] The sampling points in the sampling point set A satisfy:

[0104] a iem =a ie d e mod n 2

[0105] The sampling points in the sampling point set B satisfy:

[0106] b iem =b ie (d e ) -1 mod n 2

[0107] Among them, r1, r2, r d All are random parameters, a iem ≥b iemTo ensure that the watermark can be directly extracted from the ciphertext domain later, appropriate random parameters r1, r2, and r3 are selected. d , guarantee a iem ≥b iem Because r1, r2, r d The randomness of the selection ensures iem ≥b iem It is feasible.

[0108] In this embodiment, due to the additive homomorphic property of the Paillier encryption algorithm, the user terminal receives the encrypted watermarked audio from the cloud service, decrypts the encrypted watermarked audio, and then extracts the watermark. Alternatively, the user terminal receives the encrypted watermarked audio from the cloud service without decryption, and when extracting the watermark in the ciphertext field, the plaintext value obtained after decryption is the same.

[0109] In this process, the user receives encrypted watermarked audio from the cloud service, decrypts the encrypted watermarked audio, and then extracts the watermark, assuming a... id b id The plaintext value obtained by decrypting the ciphertext in sampling point set A and sampling point set B satisfies:

[0110]

[0111] The user client authenticates the decrypted audio and watermark. The process is as follows:

[0112] S401. Calculate the difference s between the average value of the sampled data in sampled data set A and the average value of the sampled data in sampled data set B after decryption. The expression is:

[0113]

[0114] Where N represents the total number of sampled data;

[0115] S402. Use the average difference s as the watermark detection value. If s≈θ+2d, it indicates that the watermark information is contained; otherwise, it does not contain watermark information.

[0116] In this process, the user receives encrypted watermarked audio from the cloud service. The encrypted watermarked audio is not decrypted; instead, the watermark is extracted from the ciphertext field. Let a... id b id The plaintext value obtained by decrypting the ciphertext in sampling point set A and sampling point set B satisfies:

[0117] a id =D(a) iem )=D(a ime )

[0118] b id =D(b) iem)=D(b ime )

[0119] The user client authenticates the decrypted audio and watermark. The process is as follows:

[0120] S41. During the watermark embedding process, a is satisfied. iem ≥b iem Under the premise of extracting the watermark from the ciphertext field, using s as the watermark detection information, the expression is:

[0121]

[0122] in,

[0123] S42. If s≈1, it indicates that the watermark information is present; otherwise, it does not contain watermark information.

[0124] Example 3

[0125] This embodiment verifies the effectiveness of the method proposed in this invention through specific experiments. Figure 3 This is an audio waveform diagram to be embedded with an encrypted watermark. Figure 4 To use the method proposed in this application to embed a watermark into the waveform of the audio before encryption, Figure 5 To obtain the waveform of the audio after encryption and watermarking using the method proposed in this application, Figure 6 The image shows the waveform after decrypting the encrypted watermarked audio and then extracting the watermark using the method proposed in this application. The result of directly extracting the watermark is consistent with the result of decrypting the encrypted watermarked audio and then extracting the watermark, meaning that the homomorphic encrypted audio watermark extraction result designed in this invention is essentially consistent with the plaintext field.

[0126] This embodiment tested 15 WAV audio files with different parameters, and the test results are shown in Table 1. The columns in the table represent the following: Preprocessing: The watermark detection value (plaintext field) of the original file after preprocessing; me_final: The watermark detection value (ciphertext field) of the file obtained after preprocessing, embedding the watermark, and then encrypting. me indicates Mark And Encode (embed the watermark first, then encrypt); me_decode: The watermark detection value (plaintext field) of the file obtained by decrypting the ciphertext file obtained by embedding the watermark first and then encrypting. me indicates Mark And Encode (embed the watermark first, then encrypt); em_final: The watermark detection value (ciphertext field) of the file obtained after preprocessing, encrypting first, and then embedding the watermark. em indicates Encode And Mark (encryption first, then embedding the watermark); em_decode: The watermark detection value (plaintext field) of the file obtained by decrypting the ciphertext file obtained by encrypting first and then embedding the watermark. em indicates Encode And Mark (encryption first, then embedding the watermark).

[0127] Table 1

[0128]

[0129]

[0130] The experimental results demonstrate the correctness and security of the homomorphic encrypted audio watermark designed in this invention, proving that the algorithm achieves commutativity between encryption and watermark embedding, and can directly extract the watermark from the ciphertext domain.

[0131] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A homomorphic encrypted audio watermarking system in a cloud computing environment, characterized in that, include: The user-side is used to request homomorphic encryption public and private keys from the privacy server, encrypt audio to obtain encrypted watermarked audio, decrypt the encrypted watermarked audio, and perform authentication. To encrypt the audio, the audio is first sampled. When sampling the audio, two sets of sampling points are randomly selected from the audio data segment: sampling point set A and sampling point set B, and sampling points are randomly selected from the two sets of sampling points. Assume that the number of sampling points in both sets of sampling points is 1. N ;The first sampling point in the set A i The sampled values ​​of each sampling point are a i The sampling point set B contains the first... i The sampled values ​​of each sampling point are b i Let the difference between the average value of the sampled data in sampled data set A and the average value of the sampled data in sampled data set B be denoted as . θ ; Using public key pk Encrypted watermarks are embedded at each sampling point of the audio. The process involves embedding the watermark first, followed by encryption. S201. Increase the sampled values ​​of all sampled points in sampled point set A. d Reduce the sampled values ​​of all sampled points in sampled point set B. d ; S202. Let... a im 、b im These are the sampled values ​​from sample point set A and sample point set B after watermark embedding, respectively: a im = a i + d b im = b i –d S203. Encrypt the embedded watermark, set a ime 、b ime The sampled value obtained by embedding the watermark and then encrypting it is expressed as: ； in, r 1 and r 2 represents random parameters. a ime ≥b ime ; Alternatively, using a public key pk When embedding an encrypted watermark at each sampling point of the audio, the process of first encrypting the audio and then embedding the watermark includes: S21. Let... a ie , b ie The encrypted value: S22. Embed a watermark in the ciphertext field, and set... a iem , b iem The value is encrypted first and then the watermark is embedded. First, the embedded value is... d Encrypt and embed values d The obtained encryption value is d e : The sampling points in the sampling point set A satisfy: The sampling points in the sampling point set B satisfy: in, r 1. r 2. r d All parameters are random. a iem ≥b iem ; A privacy server provides homomorphic encryption public and private keys to the user's client. The cloud server is used to store encrypted watermarked audio uploaded by users and also to distribute the encrypted watermarked audio to users.

2. The homomorphic encrypted audio watermarking system in a cloud computing environment according to claim 1, characterized in that, The homomorphic encryption is Paillier additive homomorphism.

3. A homomorphic encrypted audio watermarking method based on the homomorphic encrypted watermarking system in a cloud computing environment as described in claim 2, characterized in that, Includes the following steps: S1. The user requests and obtains a public key from the privacy server. pk and private key sk ; S2. The user will send the public key. pk Transmitted to the cloud server and using the public key. pk After embedding encrypted watermarks at each sampling point of the audio, it is uploaded to the cloud server; When sampling audio, two sets of sampling points are randomly selected from the audio data segment: sampling point set A and sampling point set B, and sampling points are randomly selected from the two sets of sampling points. Assume that the number of sampling points in both sets of sampling points is 1. N ;The first sampling point in the set A i The sampled values ​​of each sampling point are a i The sampling point set B contains the first... i The sampled values ​​of each sampling point are b i Let the difference between the average value of the sampled data in sampled data set A and the average value of the sampled data in sampled data set B be denoted as . θ ; Using public key pk The process of embedding encrypted watermarks at each sampling point of the audio, followed by encryption, includes: S201. Increase the sampled values ​​of all sampled points in sampled point set A. d Reduce the sampled values ​​of all sampled points in sampled point set B. d ; S202. Let... a im 、b im These are the sampled values ​​from sample point set A and sample point set B after watermark embedding, respectively: a im = a i + d b im = b i –d S203. Encrypt the embedded watermark, set a ime 、b ime The sampled value obtained by embedding the watermark and then encrypting it is expressed as: ； in, r 1 and r 2 represents random parameters. a ime ≥b ime ; Alternatively, using a public key pk When embedding an encrypted watermark at each sampling point of the audio, the process of first encrypting the audio and then embedding the watermark includes: S21. Let... a ie , b ie The encrypted value: S22. Embed a watermark in the ciphertext field, and set... a iem , b iem The value is encrypted first and then the watermark is embedded. First, the embedded value is... d Encrypt and embed values d The obtained encryption value is d e : The sampling points in the sampling point set A satisfy: The sampling points in the sampling point set B satisfy: in, r 1. r 2. r d All parameters are random. a iem ≥b iem ; S3. The user terminal uses the private key sk to decrypt the encrypted watermarked audio and extract the watermark, and then performs authentication; the encrypted watermarked audio is downloaded by the user terminal from the cloud server.

4. The homomorphic encrypted audio watermarking method implemented by the homomorphic encrypted watermarking system in a cloud computing environment according to claim 3, characterized in that, The audio is in WAV file format, with a sampling bit depth of 8 bits or 16 bits. The 8-bit sampling bit depth uses PCM pulse code modulation, is unsigned, and the sampling data range is 0. 255, select the embedded watermark value d =1, the specific preprocessing method is as follows: for sampling points with a sample value less than 1, set the sample value to 1; for sampling points with a sample value greater than 245, set the sample value to 245; the sample values ​​of the remaining sampling points remain unchanged; The 16-bit sampling depth uses PCM pulse code modulation, is a signed number, and the sampling data range is [missing information]. - 32768 32767, Select the embedded watermark value d =1, the specific preprocessing method is as follows: for sampling points with a sample value less than -32767, the sample value is set to -32767; for sampling points with a sample value greater than 32703, the sample value is set to 32703; the sample values ​​of the remaining sampling points remain unchanged.

5. The homomorphic encrypted audio watermarking method implemented by the homomorphic encrypted watermarking system in a cloud computing environment according to claim 4, characterized in that, The audio samples include floating-point numbers and negative numbers, where the user terminal uses a public key. pk When embedding an encrypted watermark on the floating-point numbers at each audio sampling point, the floating-point numbers at each audio sampling point are multiplied by a large integer. Q Then, round down to the nearest integer, and finally encrypt it to get the large integer. Q For precision control coefficients; the user terminal uses the public key pk When embedding encrypted watermarks for negative numbers at each audio sampling point, a modulo operation method is used. The process is as follows: The public key... pk Original encryption range The latter half of the encryption range Indicates the range of negative number encryption This makes the encryption range become , n Represents the largest positive integer that can be encrypted.

6. The homomorphic encrypted audio watermarking method implemented by the homomorphic encrypted watermarking system in a cloud computing environment according to claim 5, characterized in that, The user receives encrypted watermarked audio from the cloud service, decrypts the encrypted watermarked audio, and then extracts the watermark. Alternatively, the user receives encrypted watermarked audio from the cloud service without decryption, and when extracting the watermark in the ciphertext field, the plaintext value obtained after decryption is the same.

7. The homomorphic encrypted audio watermarking method implemented by the homomorphic encrypted watermarking system in a cloud computing environment according to claim 5, characterized in that, The user receives encrypted watermarked audio from the cloud service, decrypts the encrypted watermarked audio, and then extracts the watermark, setting... a id , b id The plaintext value obtained by decrypting the ciphertext in sampling point set A and sampling point set B satisfies: ； The user client authenticates the decrypted audio and watermark. The process is as follows: S401. Calculate the difference between the average value of the sampled data in sampled data set A and the average value of the sampled data in sampled data set B after decryption. s， The expression is: in, N Indicates the total number of sampled data; S402. Difference between average values s As a watermark detection value, if s ≈ θ +2 d If the symbol indicates that the watermark information is present, then the watermark information is not present.

8. The homomorphic encrypted audio watermarking method implemented by the homomorphic encrypted watermarking system in a cloud computing environment according to claim 5, characterized in that, The user receives encrypted watermarked audio from the cloud service. The encrypted watermarked audio is not decrypted; instead, the watermark is extracted from the ciphertext field. a id , b id The plaintext value obtained by decrypting the ciphertext in sampling point set A and sampling point set B satisfies: The user client authenticates the decrypted audio and watermark. The process is as follows: S41. During the watermark embedding process, the following conditions are met: a iem ≥b iem Under the premise of extracting watermark from encrypted domain, in order to s As the watermark detection information, the expression is: in, S42. If s ≈1 indicates that the watermark information is present; otherwise, the watermark information is not present.

Images