Super-large-scale audio routing method

Abstract

The invention relates to the technical field of audio and video output, and discloses a super-large-scale audio routing method, which comprises the following steps of: distributing a global unique identifier for audio equipment in a target scene; determining an initial audio transmission path of the audio equipment in the target scene in combination with similarity calculation, and performing fine tuning optimization by using a semantic-level routing algorithm to generate a target audio transmission path; determining an audio transmission protocol, and automatically generating an audio routing instruction; encrypting the audio data by using a sound wave fingerprint dynamic encryption technology, and issuing the encrypted audio data to audio equipment in the target scene; and the audio device receives the encrypted audio data and decrypts the encrypted audio data to complete transmission of the audio data. According to the invention, real-time audio mixing processing of 512 paths of audios and route exchange of 2048 paths of signals can be realized, low-delay and high-reliability scheduling of high-density audio signals in a super-large-scale scene is supported, and strict requirements of broadcast and TV live broadcast, large-scale venue sound reinforcement and the like on real-time performance and processing scale are met.

Description

Technical Field

[0001] This invention relates to the field of audio and video output technology, and more specifically, to a method for ultra-large-scale audio routing. Background Technology

[0002] In the current field of audio transmission, with the continuous development of technology, the application scenarios of audio devices are becoming increasingly diverse, ranging from simple home audio-visual entertainment to large-scale concerts, live events, and professional radio stations, leading to increasingly higher requirements for audio transmission. Existing audio routing technologies can, to a certain extent, meet the audio data transmission needs in conventional scenarios. For example, in some small venues, basic audio signal transmission from the source device to the target device can be achieved through traditional wired connections or common wireless transmission protocols, and some technologies can also perform simple processing and distribution of audio data. In some corporate conference rooms, audio transmission between a limited number of devices can be achieved by using ordinary network routers in conjunction with common audio protocols.

[0003] However, existing technologies reveal numerous shortcomings and deficiencies when facing ultra-large-scale audio routing scenarios. On one hand, for ultra-large-scale audio device deployments, traditional technologies struggle to quickly and accurately assign unique identifiers to a vast number of audio devices. Furthermore, in terms of device status monitoring and management, they cannot efficiently update device status information in real time, leading to chaotic device management and situations where connection errors or device disconnections cannot be detected and addressed promptly. On the other hand, when determining audio transmission paths, existing methods lack effective utilization of historical routing experience and a deep understanding of the semantics of audio content. In complex ultra-large-scale scenarios, it is difficult to comprehensively determine the optimal audio transmission protocol based on multiple factors such as real-time requirements, transmission scale, network environment, and device compatibility. Moreover, regarding the secure transmission of audio data, existing encryption technologies cannot adequately guarantee the security and integrity of audio data during transmission. Simultaneously, if encryption technology is insufficient, audio data may face the risk of being stolen or tampered with.

[0004] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0005] In view of the problems in related technologies, this invention proposes a large-scale audio routing method to overcome the aforementioned technical problems existing in the existing related technologies.

[0006] Therefore, the specific technical solution adopted by the present invention is as follows: A method for ultra-large-scale audio routing includes the following steps: S1. Using a random number generation algorithm, assign a globally unique identifier to the audio device in the target scene, and automatically register metadata in the network after the audio device is started; detect the online status of the audio device by periodically sending heartbeat packets, and update the registry periodically; S2. Based on the self-evolving routing knowledge base, the initial audio transmission path of the audio device in the target scene is determined by similarity calculation, and the initial audio transmission path is fine-tuned and optimized by the semantic-level routing algorithm to generate the target audio transmission path. S3. Determine the audio transmission protocol based on real-time requirements, transmission scale, network environment and device compatibility, and automatically generate audio routing instructions based on the target audio transmission path, audio transmission protocol and globally unique identifier, combined with signal flow direction. S4. The audio data is encrypted using acoustic fingerprint dynamic encryption technology. Based on the audio routing command, the encrypted audio data is sent to the audio device in the target scene. The audio device receives the encrypted audio data and decrypts it to complete the transmission of the audio data.

[0007] Preferably, the process of determining the initial audio transmission path of the audio device in the target scene based on a self-evolving routing knowledge base and similarity calculation, and then fine-tuning and optimizing the initial audio transmission path using a semantic-level routing algorithm to generate the target audio transmission path includes the following steps: S21. Construct a self-evolving routing knowledge base and encrypt and store historical routing data using blockchain technology to form a historical case library; based on a similarity algorithm, calculate the similarity between the audio device characteristic information in the target scene and the historical case library, and generate an initial audio transmission path based on the similarity calculation results; S22. Based on lightweight audio processing, speech activity detection and spectral feature extraction technology, the initial audio transmission path is dynamically fine-tuned, and combined with real-time network status optimization to form an optimized path with parameter adaptation, thus obtaining the target audio transmission path.

[0008] Preferably, the steps of constructing a self-evolving routing knowledge base and encrypting and storing historical routing data using blockchain technology to form a historical case library; calculating the similarity between audio device characteristic information in the target scene and the historical case library based on a similarity algorithm, and generating an initial audio transmission path based on the similarity calculation results include the following steps: S211. Construct a self-evolving routing knowledge base and encrypt and store historical routing data using blockchain technology to form a historical case library. The historical routing data includes the scenario characteristics, strategy parameters, and performance indicators of each routing session. S212. Extract the feature information of the number, type, network access method, topology, service type and audio stream characteristics of audio devices in the target scene, and calculate the similarity between the feature information in the target scene and the feature information in the historical case library. S213. Based on the similarity calculation results of the feature information, select the best historical case from the historical case library, extract the corresponding routing strategy through the best historical case, and generate the initial audio transmission path in combination with the target scene parameters.

[0009] Preferably, the process of dynamically fine-tuning the initial audio transmission path based on lightweight audio processing, speech activity detection, and spectral feature extraction technologies, and combining this with real-time network status optimization to form a parameter-adapted optimized path, to obtain the target audio transmission path, includes the following steps: S221. Based on lightweight audio processing technology, analyze the real-time semantics of audio streams in the target scene, and use voice activity detection and spectrum feature extraction technology to identify audio content features and generate corresponding semantic tags to clarify the service priority and transmission requirements of different audio streams. S222. The initial audio transmission path is dynamically fine-tuned using semantic tags, and the path node allocation is optimized in combination with the real-time network status to form an optimized path with parameter adaptation, thus obtaining the target audio transmission path.

[0010] Preferably, the step of analyzing the real-time semantics of audio streams in the target scene based on lightweight audio processing technology, and identifying audio content features using voice activity detection and spectral feature extraction technology to generate corresponding semantic tags, in order to clarify the service priority and transmission requirements of different audio streams, includes the following steps: S2211. Obtain the original audio stream data in the target scene, use adaptive gain control and noise suppression algorithm to remove environmental noise from the original audio stream data, and segment the original audio stream data through a sliding window. S2212. Using a dual-threshold energy detection algorithm, calculate the short-time energy and zero-crossing rate of each audio frame, and identify speech and non-speech segments according to a preset dynamic threshold; based on a deep learning model, identify the speech start point and end point, and generate a speech activity timestamp. S2213. Based on Fast Fourier Transform, each frame of speech segment is converted to the frequency domain, and the Mel frequency cepstral coefficients are calculated to capture the timbre features of the audio. The parameters of spectral entropy and spectral flatness are extracted to characterize the audio complexity. S2214. Based on the Mel frequency cepstral coefficient feature vector, combined with a pre-trained lightweight neural network, the audio content is classified, the corresponding classification label is output, and the importance level of the audio is evaluated by combining the speech activity detection results and energy features. S2215. Analyze the continuity and time sensitivity of the audio stream, generate latency tolerance labels, integrate classification labels, importance levels and latency tolerance labels to obtain semantic labels, so as to clarify the service priority and transmission requirements of different audio streams.

[0011] Preferably, the step of determining the audio transmission protocol based on real-time requirements, transmission scale, network environment, and device compatibility, and automatically generating audio routing instructions based on the target audio transmission path, audio transmission protocol, and globally unique identifier, combined with signal flow direction, includes the following steps: S31. Comprehensively evaluate the real-time requirements, transmission scale, network environment and equipment compatibility of the target scenario, determine the audio transmission protocol of the target scenario, and ensure that the selected audio transmission protocol is compatible with each audio device in the system to avoid situations where the protocol is not supported. S32. Based on the target audio transmission path, combined with the audio transmission protocol and the globally unique identifier of each device, automatically generate audio routing instructions according to the direction of signal flow from the source device through the processing node to the target device.

[0012] Preferably, the audio transmission protocol of the target scenario includes any one of the AES67 protocol, MADI protocol, and DANTE protocol; the audio routing instructions include information on physical port mapping, protocol parameter configuration, and quality of service settings.

[0013] Preferably, the method of encrypting audio data using acoustic fingerprint dynamic encryption technology and sending the encrypted audio data to the audio device in the target scene based on audio routing instructions includes the following steps: S41. Calculate the Mel-spectral centroid of each audio frame and use the calculation result as the key seed; S42. Based on the key seed value, select the corresponding offset from the chaotic sequence generated by the quantum random number generator, and perform a cyclic shift operation on the audio sampling points according to the offset to achieve audio data encryption. S43. Use the offset as a key encryption parameter and embed it into the metadata field of the audio routing instruction to generate an audio routing instruction containing the encryption parameter. S44. Based on the audio routing instruction containing encryption parameters, combined with the audio transmission protocol and the target audio transmission path, the encrypted audio data is sent to the audio device in the target scene. S45. The target audio device receives the encrypted audio data and audio routing instructions, and extracts the encrypted key parameters from the metadata of the audio routing instructions. S46. The target audio device uses the extracted encryption key parameters to perform a reverse cyclic displacement operation on the encrypted audio sampling points to obtain the original audio content and complete the transmission of audio data.

[0014] Preferably, the step of calculating the Mel-spectral centroid of each audio frame and using the calculation result as a key seed includes: Each audio frame is converted from the time domain to the frequency domain, the frequency domain signal is processed by the Mel filter bank, and the centroid position of the Mel spectrum is calculated to generate a key seed that is specific to the audio content.

[0015] Preferably, the step of selecting a corresponding offset from the chaotic sequence generated by the quantum random number generator based on the key seed value, and performing a cyclic shift operation on the audio sampling points according to the offset to achieve audio data encryption includes the following steps: S421. Using modular arithmetic, the barycenter of the Mel spectrum of the audio frame is mapped to the index space of the chaotic sequence of the quantum random number generator, and the index is preprocessed using the salt value in the globally unique identifier. S422. Based on the preprocessed index, extract several consecutive random numbers from the chaotic sequence pre-generated by the quantum random number generator, and generate the final offset through XOR operation. S423. Divide the audio frames into blocks of a preset length according to the time sequence, encrypt each block independently, and introduce a random increment in the offset between each block. S424. Modulate the encrypted sampling points using the phase and synchronize the offset and phase modulation parameters to the audio device in the target scene through a low bit rate channel.

[0016] Compared with existing technologies, this invention provides a method for ultra-large-scale audio routing, which has the following beneficial effects: (1) Based on one FPGA, the present invention can realize 512-channel audio real-time mixing and 2048-channel signal routing and switching. The single device processing delay is less than 1ms. It supports low-latency and high-reliability scheduling of high-density audio signals in ultra-large-scale scenarios, and meets the stringent requirements of real-time performance and processing scale for broadcasting, large venue sound reinforcement, etc.

[0017] (2) This invention can use a random number generation algorithm to give audio devices a unique global identifier. Combined with the automatic registration mechanism of device metadata and the real-time status monitoring and update function, it can achieve a double leap in the efficiency and accuracy of device management. This can make the management of audio devices more orderly, greatly reduce the situation of audio device connection errors and failure to handle disconnection in time, ensure the stable operation of the audio system, and lay a solid foundation for the smooth transmission of audio data in the future.

[0018] (3) Based on the self-evolving routing knowledge base and similarity calculation, this invention integrates semantic-level routing algorithms to quickly and accurately generate and optimize audio transmission paths. Through deep learning historical cases and detailed analysis of audio semantics, it can easily match the best transmission path in complex ultra-large-scale audio routing scenarios, significantly reduce audio transmission latency, effectively control packet loss rate, significantly improve audio transmission quality, and ensure that audio can be transmitted in real time and stably in various large-scale events, bringing users an excellent audio experience.

[0019] (4) This invention uses acoustic fingerprint dynamic encryption technology to build a security defense for audio data transmission. With the help of the unique encryption method, the audio data is almost impossible to be cracked or tampered with throughout the transmission process, which effectively protects the security and integrity of the audio data and allows the audio transmission to run stably on a safe track. Attached Figure Description

[0020] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram illustrating the principle of a large-scale audio routing method according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] According to an embodiment of the present invention, a method for ultra-large-scale audio routing is provided.

[0023] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, according to an embodiment of the present invention, a method for ultra-large-scale audio routing is provided, which implements 512-channel real-time audio mixing processing and 2048-channel signal routing and switching based on FPGA. The method includes the following steps: S1. Using a random number generation algorithm, assign a globally unique identifier to the audio device in the target scene, and automatically register metadata (including IP address, supported formats, etc.) in the network after the audio device is started; detect the online status of the audio device by periodically sending heartbeat packets, and update the registry periodically; Specifically, a random number generation algorithm is used to assign globally unique identifiers (GUIDs) to audio devices in the target scene. After the device starts up, it automatically completes the registration of metadata (such as IP address, supported formats, etc.) in the network. By periodically sending heartbeat packets to detect the online status of the devices in real time and dynamically updating the registry, efficient identification and status management of ultra-large-scale audio devices can be achieved, providing accurate device topology information for subsequent routing planning.

[0024] S2. Based on the self-evolving routing knowledge base, the initial audio transmission path of the audio device in the target scene is determined by similarity calculation, and the initial audio transmission path is fine-tuned and optimized by the semantic-level routing algorithm to generate the target audio transmission path. The process of determining the initial audio transmission path for audio devices in the target scene based on a self-evolving routing knowledge base and similarity calculation, and then fine-tuning and optimizing the initial audio transmission path using a semantic-level routing algorithm to generate the target audio transmission path includes the following steps: S21. Construct a self-evolving routing knowledge base and encrypt and store historical routing data using blockchain technology to form a historical case library; based on a similarity algorithm, calculate the similarity between the audio device characteristic information in the target scene and the historical case library, and generate an initial audio transmission path based on the similarity calculation results; Specifically, the process of constructing a self-evolving routing knowledge base and encrypting and storing historical routing data using blockchain technology to form a historical case library; calculating the similarity between audio device characteristic information in the target scene and the historical case library based on a similarity algorithm, and generating an initial audio transmission path based on the similarity calculation results includes the following steps: S211. Construct a self-evolving routing knowledge base and encrypt and store historical routing data using blockchain technology to form a historical case library. The historical routing data includes the scenario characteristics of each route (number of devices, type distribution, network topology, service type, etc.), policy parameters (transmission protocol, path bandwidth / latency, QoS configuration, etc.), and performance indicators (packet loss rate, audio quality score, device load, etc.). In this embodiment, the construction of the self-evolving routing knowledge base includes: after each routing decision, the system automatically records the scene characteristics, strategy parameters and effect indicators. The data is stored in encrypted form using blockchain technology to ensure that it is tamper-proof and supports cross-domain sharing (such as chain theater groups sharing routing experience).

[0025] S212. Extract feature information from the target scene, including the number and type of audio devices (microphones, speakers, mixers, etc.), network access method (wired / wireless), topology (star / ring / hybrid), service type (live streaming / conference / performance, etc.), and audio stream characteristics (real-time performance, number of channels, priority requirements, etc.), and calculate the similarity between the feature information in the target scene and the feature information in the historical case library; specifically including: First, numerical data such as the number of devices and channels are standardized (e.g., Z-score or Min-Max scaling) and then similarity is calculated using Euclidean distance or Manhattan distance. Next, for enumerated data such as device type (microphone / speaker, etc.) and network access method (wired / wireless), cosine similarity is calculated using Jaccard similarity or one-hot encoding. Then, topological structures (star / ring, etc.) can be mapped to graph features, and similarity is calculated using graph edit distance or subgraph isomorphism algorithms. Business types (live streaming / conference, etc.) are treated as text features, first converted to semantic vectors using TF-IDF or word vectors (e.g., Word2Vec), and then cosine similarity is calculated. Finally, the similarity calculation results are dynamically adjusted by combining historical case performance metrics stored in the blockchain (e.g., packet loss rate, audio quality). Historical cases with high packet loss rates are given lower weights, while recent cases or those with excellent performance are given higher weights, achieved through a time decay factor. S213. Based on the similarity calculation results of the feature information, select the best historical case from the historical case library, extract the corresponding routing strategy through the best historical case, and generate the initial audio transmission path by combining the target scene parameters (device IP, real-time bandwidth, etc.).

[0026] S22. Based on lightweight audio processing, speech activity detection and spectral feature extraction technology, the initial audio transmission path is dynamically fine-tuned, and combined with real-time network status optimization to form an optimized path with parameter adaptation, thus obtaining the target audio transmission path.

[0027] Specifically, the process of dynamically fine-tuning the initial audio transmission path based on lightweight audio processing, speech activity detection, and spectral feature extraction technologies, and combining this with real-time network status optimization to form a parameter-adapted optimized path, to obtain the target audio transmission path, includes the following steps: S221. Based on lightweight audio processing technology, analyze the real-time semantics of audio streams in the target scene, and use voice activity detection and spectrum feature extraction technology to identify audio content features and generate corresponding semantic tags to clarify the service priority and transmission requirements of different audio streams. Specifically, the process of analyzing the real-time semantics of audio streams in a target scene based on lightweight audio processing technology, and identifying audio content features using voice activity detection and spectral feature extraction technologies to generate corresponding semantic tags, in order to clarify the service priority and transmission requirements of different audio streams, includes the following steps: S2211. Deploy a lightweight audio acquisition module in the audio source device (such as a microphone or mixing console) or network node to acquire the raw audio stream data in the target scene in real time. Use adaptive gain control and noise suppression algorithms (such as spectral subtraction) to remove environmental noise from the raw audio stream data. Then, use a sliding window (usually 20-50ms frame length) to segment the raw audio stream data and generate processing units suitable for semantic analysis. S2212. Using a dual-threshold energy detection algorithm, calculate the short-time energy and zero-crossing rate of each frame of audio, and identify speech and non-speech segments according to a preset dynamic threshold (e.g., energy above the high threshold is considered speech, and energy below the low threshold is considered silence); based on a deep learning model, identify the start and end points of speech and generate speech activity timestamps. When calculating the short-time energy and zero-crossing rate of each frame of audio using the dual-threshold energy detection algorithm, the audio signal is first processed by dividing it into frames. The short-time energy is calculated for each frame by weighting the audio samples using a window function and then summing the squares to measure the signal energy intensity. At the same time, the zero-crossing rate is calculated, and the number of symbol changes between adjacent samples is counted to reflect the signal frequency characteristics. Then, two energy thresholds are set: when the short-time energy is higher than the high threshold, it is determined to be a speech segment; when it is lower than the low threshold, it is determined to be a silence segment; when it is between the two thresholds, it is further judged by combining features such as the zero-crossing rate, thereby achieving effective differentiation between speech and non-speech segments. S2213. Based on Fast Fourier Transform, each frame of speech segment is converted to the frequency domain, and the Mel frequency cepstral coefficients are calculated to capture the timbre features of the audio. Parameters such as spectral entropy and spectral flatness are extracted to characterize the audio complexity. Feature parameters are optimized for specific application scenarios, such as adding pitch detection in music scenarios and strengthening formant extraction in speech scenarios. When calculating Mel frequency cepstral coefficients (MFCCs), the audio is first framed and Fourier transformed to the frequency domain. Then, the frequencies are mapped according to the characteristics of human hearing through a Mel filter bank. After taking the logarithm of the Mel spectrum, the MFCCs are obtained through discrete cosine transform. This can capture the timbre characteristics of the audio, such as the formants of human voices and the harmonic structure of musical instruments. When extracting spectral entropy, the spectral amplitude is normalized to a probability distribution and the entropy value is calculated. This value is used to characterize the uniformity of the spectral energy distribution. The higher the entropy value, the higher the audio complexity. When extracting spectral flatness, the ratio of the geometric mean to the arithmetic mean of the spectrum is used to reflect the randomness of the spectrum. The closer the flatness is to 1, the higher the audio complexity. These parameters can comprehensively reflect the audio complexity. S2214. Based on the Mel frequency cepstral coefficient feature vector, combined with a pre-trained lightweight neural network (such as TinyML model), the audio content is classified and the corresponding classification labels are output (such as human voice, music, ambient sound, alarm sound, etc.). Combined with the speech activity detection results and energy features, the importance level of the audio is evaluated (such as key speech, background music, low priority monitoring sound, etc.). S2215. Analyze the continuity and time sensitivity of the audio stream, generate latency tolerance labels, such as ultra-low latency (<10ms), standard latency (50ms), and tolerable latency (>100ms), integrate the classification labels, importance levels and latency tolerance labels to obtain semantic labels (such as high-priority human voice_ultra-low latency) to clarify the service priority and transmission requirements of different audio streams.

[0028] S222. Use semantic tags to dynamically fine-tune the initial audio transmission path (increase the delay weight and reserve dedicated bandwidth for high-priority streams, and allow the reuse of idle links for low-priority streams), and combine real-time network status to optimize path node allocation, forming an optimized path with parameter adaptation, and obtaining the target audio transmission path.

[0029] Step S2 above is based on the collaborative operation of a self-evolving routing knowledge base and a semantic-level routing algorithm. This step realizes intelligent routing optimization from the reuse of historical experience to real-time semantic awareness. It forms a reusable case library by encrypting and storing historical routing data through blockchain, and quickly generates initial paths by combining similarity algorithms, which significantly shortens the path planning time in ultra-large-scale scenarios. At the same time, it uses lightweight audio processing technology to analyze audio semantic features in real time and dynamically fine-tune path parameters to adapt to the priority and transmission requirements of different audio streams. This effectively improves the real-time guarantee capability of high-priority audio streams and the overall network resource utilization rate, and realizes low-latency and high-reliability routing scheduling for ultra-large-scale audio devices. It meets the stringent requirements of real-time and stability of audio transmission in complex scenarios such as broadcasting and large venue sound reinforcement.

[0030] S3. Determine the audio transmission protocol based on real-time requirements, transmission scale, network environment and device compatibility, and automatically generate audio routing instructions based on the target audio transmission path, audio transmission protocol and globally unique identifier, combined with signal flow direction. The step of determining the audio transmission protocol based on real-time requirements, transmission scale, network environment, and device compatibility, and automatically generating audio routing instructions based on the target audio transmission path, audio transmission protocol, and globally unique identifier, combined with signal flow direction, includes the following steps: S31. Comprehensively evaluate the real-time requirements, transmission scale, network environment and equipment compatibility of the target scenario, determine the audio transmission protocol of the target scenario, and ensure that the selected audio transmission protocol is compatible with each audio device in the system to avoid situations where the protocol is not supported. Specifically, the audio transmission protocol of the target scene includes any one of the following: AES67 protocol, MADI protocol, and DANTE protocol; Based on a comprehensive assessment of the target scenario's real-time requirements, transmission scale, network environment, and device compatibility, the audio transmission protocols for the target scenario are determined to include: Real-time requirements assessment: For ultra-low latency scenarios (latency requirement ≤10ms, such as live performance monitoring, broadcast monitoring), MADI (physical layer microsecond latency) or AES67 (IP layer latency 5-10ms) should be given priority. Transmission scale assessment: For single-channel or small-scale multi-channel (≤16 channels), AES67 is preferred; for medium-scale multi-channel (17-64 channels), MADI protocol is selected; for large-scale multi-channel (≥65 channels), MADI (coaxial supports 64 channels, fiber supports 128 channels) or a hybrid architecture combining MADI and IP protocols (such as MADI carrying the backbone and AES67 handling control signals) is adopted directly. Network environment assessment: For dedicated audio networks (such as fiber optic ring networks and coaxial cables), MADI protocol should be used first; for general IP networks (LAN / WAN), AES67 should be selected based on bandwidth and stability (low bandwidth, high efficiency); for wireless networks or mobile environments, optimized AES67 wireless configuration should be adopted (such as reducing multicast traffic). Device compatibility assessment: If all devices support MADI interfaces (such as professional mixing consoles and multi-channel audio interfaces), use the MADI protocol directly; if the devices only support IP interfaces (such as computers and network speakers), select AES67; if there are heterogeneous devices (such as a mix of MADI mixing consoles and IP speakers), achieve interoperability through a protocol converter (such as a MADI-to-AES67 gateway), and prioritize the main protocol supported by the converter. S32. Based on the target audio transmission path, combined with the audio transmission protocol and the globally unique identifier of each device, automatically generate audio routing instructions according to the direction of signal flow from the source device through the processing node to the target device; Specifically, the audio routing instructions include information on physical port mapping, protocol parameter configuration, and quality of service settings.

[0031] Step S3 above is based on a multi-dimensional evaluation mechanism considering real-time requirements, transmission scale, network environment, and device compatibility. This step achieves dynamic and accurate matching of audio transmission protocols and automated generation of routing instructions. By constructing a protocol selection decision model through a hierarchical evaluation strategy (real-time, scale, network, device), it can prioritize the adaptation of MADI / AES67 protocols for ultra-low latency scenarios, adopt MADI or hybrid architectures for large-scale multi-channel scenarios, and achieve interoperability through protocol converters in heterogeneous device environments. This significantly improves the adaptation efficiency of protocols and scenarios and avoids transmission interruption problems caused by protocol incompatibility. At the same time, it automatically generates routing instructions containing physical port mapping, protocol parameters, and QoS settings based on the target path, protocol, and unique device identifier. This achieves full-process automation from protocol selection to device control, ensuring accurate control of signal flow in ultra-large-scale audio networks. It effectively improves the real-time performance, stability, and device compatibility of audio transmission in scenarios such as broadcasting and large venues, while reducing the complexity of manual configuration and the cost of protocol adaptation.

[0032] S4. The audio data is encrypted using acoustic fingerprint dynamic encryption technology. Based on the audio routing command, the encrypted audio data is sent to the audio device in the target scene. The audio device receives the encrypted audio data and decrypts it to complete the transmission of the audio data.

[0033] The process of encrypting audio data using acoustic fingerprint dynamic encryption technology, and then sending the encrypted audio data to audio devices in the target scene based on audio routing instructions, includes the following steps: The audio devices receive and decrypt the encrypted audio data to complete the audio data transmission. S41. Calculate the centroid of the Mel spectrum for each audio frame and use the calculation result as a key seed; specifically, calculating the centroid of the Mel spectrum for each audio frame and using the calculation result as a key seed includes: converting each audio frame from the time domain to the frequency domain, processing the frequency domain signal through a Mel filter bank, calculating the centroid position of the Mel spectrum, and generating a key seed with audio content specificity. The Mel spectral centroid can be understood as the "auditory centroid" of the audio spectrum energy distribution, and its calculation logic is closely related to the human ear's perception of sound frequencies. Essentially, it is the result of a weighted average of the audio spectrum energy according to the Mel frequency scale. Specifically, the calculation first converts the audio signal to the frequency domain, then processes the spectrum using a Mel filter bank to map the linear frequency axis to the Mel frequency axis that conforms to human perception. Subsequently, the energy of each Mel frequency band is multiplied by its corresponding frequency position, and the weighted energy of all frequency bands is summed and divided by the total energy to obtain the spectral centroid.

[0034] S42. Based on the key seed value, select the corresponding offset from the chaotic sequence generated by the quantum random number generator, and perform a cyclic shift operation on the audio sampling points according to the offset, thereby realizing the encryption of audio data; Specifically, the process of selecting a corresponding offset from a chaotic sequence generated by a quantum random number generator based on a key seed value, and then performing a cyclic shift operation on the audio sampling points according to the offset to encrypt the audio data includes the following steps: S421. The barycenter of the Mel spectrum of the audio frame is mapped to the index space of the chaotic sequence of the quantum random number generator using modular arithmetic, and the index is preprocessed using the salt value in the globally unique identifier; that is, the index is obfuscated by applying salt value, which comes from the hash value of the device unique identifier (GUID) to enhance collision resistance. S422. Based on the preprocessed index, extract several consecutive random numbers from the chaotic sequence pre-generated by the quantum random number generator (the number of random numbers is dynamically determined by the audio frame length), and generate the final offset through XOR operation. In addition, this embodiment also includes periodically updating the offset (e.g., regenerating it every 100ms) to achieve dynamic key rotation; S423. Divide the audio frames into blocks of a preset length according to the time sequence, encrypt each block independently, and introduce a random increment in the offset between each block to avoid pattern leakage. S424. Modulate the encrypted sampling points using the phase to further obfuscate the audio features, and synchronize the offset and phase modulation parameters to the audio device in the target scene through a low bit rate channel (such as the metadata field of the routing instruction). S43. Use the offset as a key encryption parameter and embed it in the metadata field of the audio routing instruction to generate an audio routing instruction containing encryption parameters, ensuring that these parameters can be transmitted to the target audio device along with the encrypted audio data. S44. Based on the audio routing instruction containing encryption parameters, combined with the audio transmission protocol and the target audio transmission path, the encrypted audio data is sent to the audio device in the target scene. S45. The target audio device receives the encrypted audio data and audio routing instructions, and extracts the encrypted key parameters from the metadata of the audio routing instructions. S46. The target audio device uses the extracted encryption key parameters to perform a reverse cyclic displacement operation on the encrypted audio sampling points to obtain the original audio content and complete the transmission of audio data.

[0035] Step S4 above utilizes acoustic fingerprint dynamic encryption technology to achieve highly efficient and secure protection throughout the entire audio data transmission process. Based on a key seed generated by Mel-spectrum centroid and bound to the audio content, combined with a chaotic sequence provided by a quantum random number generator, a dynamic encryption system of "one audio, one key" is constructed, effectively resisting attacks targeting fixed keys. Salt obfuscation is performed using a unique device identifier, coupled with periodic key rotation and inter-block random increment mechanisms, significantly improving key space complexity and anti-analysis capabilities. The encryption process employs low-complexity cyclic shift and phase modulation operations, with a processing latency of less than 1ms, ensuring no impact on real-time audio transmission. Encryption parameters are synchronized through routing command metadata, achieving self-synchronized decryption and simplifying system deployment. Furthermore, redundant coding and hardware acceleration design ensure transmission stability and anti-attack capabilities, guaranteeing the integrity and confidentiality of audio data even in complex network environments, meeting the stringent security and real-time requirements of professional audio scenarios.

[0036] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium. When the program is executed, it includes the steps described in the above methods. The storage medium may be, for example, ROM / RAM, magnetic disk, optical disk, etc.

[0037] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for ultra-large-scale audio routing, characterized in that, Includes the following steps: S1. Using a random number generation algorithm, assign a globally unique identifier to the audio device in the target scene, and automatically register metadata in the network after the audio device is started; detect the online status of the audio device by periodically sending heartbeat packets, and update the registry periodically; S2. Based on the self-evolving routing knowledge base, the initial audio transmission path of the audio device in the target scene is determined by similarity calculation, and the initial audio transmission path is fine-tuned and optimized by the semantic-level routing algorithm to generate the target audio transmission path. S3. Determine the audio transmission protocol based on real-time requirements, transmission scale, network environment and device compatibility, and automatically generate audio routing instructions based on the target audio transmission path, audio transmission protocol and globally unique identifier, combined with signal flow direction. S4. Use acoustic fingerprint dynamic encryption technology to encrypt audio data, and send the encrypted audio data to the audio device in the target scene based on audio routing instructions. The audio device receives the encrypted audio data and decrypts it to complete the transmission of the audio data.

2. The ultra-large-scale audio routing method according to claim 1, characterized in that, The process of determining the initial audio transmission path for audio devices in the target scene based on a self-evolving routing knowledge base and similarity calculation, and then fine-tuning and optimizing the initial audio transmission path using a semantic-level routing algorithm to generate the target audio transmission path includes the following steps: S21. Construct a self-evolving routing knowledge base and encrypt and store historical routing data using blockchain technology to form a historical case library; based on a similarity algorithm, calculate the similarity between the audio device characteristic information in the target scene and the historical case library, and generate an initial audio transmission path based on the similarity calculation results; S22. Based on lightweight audio processing, speech activity detection and spectral feature extraction technology, the initial audio transmission path is dynamically fine-tuned, and combined with real-time network status optimization to form an optimized path with parameter adaptation, thus obtaining the target audio transmission path.

3. The ultra-large-scale audio routing method according to claim 2, characterized in that, The process of constructing a self-evolving routing knowledge base and encrypting and storing historical routing data using blockchain technology to form a historical case library; calculating the similarity between audio device characteristic information in the target scene and the historical case library based on a similarity algorithm, and generating an initial audio transmission path based on the similarity calculation results includes the following steps: S211. Construct a self-evolving routing knowledge base and encrypt and store historical routing data using blockchain technology to form a historical case library. The historical routing data includes the scenario characteristics, strategy parameters, and performance indicators of each routing session. S212. Extract the feature information of the number, type, network access method, topology, service type and audio stream characteristics of audio devices in the target scene, and calculate the similarity between the feature information in the target scene and the feature information in the historical case library. S213. Based on the similarity calculation results of the feature information, select the best historical case from the historical case library, extract the corresponding routing strategy through the best historical case, and generate the initial audio transmission path in combination with the target scene parameters.

4. The ultra-large-scale audio routing method according to claim 2, characterized in that, The process of dynamically fine-tuning the initial audio transmission path based on lightweight audio processing, speech activity detection, and spectral feature extraction technologies, and combining real-time network status optimization to form a parameter-adapted optimized path, to obtain the target audio transmission path, includes the following steps: S221. Based on lightweight audio processing technology, analyze the real-time semantics of audio streams in the target scene, and use voice activity detection and spectrum feature extraction technology to identify audio content features and generate corresponding semantic tags to clarify the service priority and transmission requirements of different audio streams. S222. The initial audio transmission path is dynamically fine-tuned using semantic tags, and the path node allocation is optimized in combination with the real-time network status to form an optimized path with parameter adaptation, thus obtaining the target audio transmission path.

5. The ultra-large-scale audio routing method according to claim 4, characterized in that, The process, based on lightweight audio processing technology, analyzes the real-time semantics of audio streams in the target scene and uses speech activity detection and spectral feature extraction technology to identify audio content features and generate corresponding semantic tags to clarify the service priority and transmission requirements of different audio streams. This includes the following steps: S2211. Obtain the original audio stream data in the target scene, use adaptive gain control and noise suppression algorithm to remove environmental noise from the original audio stream data, and segment the original audio stream data through a sliding window. S2212. Using a dual-threshold energy detection algorithm, calculate the short-time energy and zero-crossing rate of each audio frame, and identify speech and non-speech segments according to a preset dynamic threshold; based on a deep learning model, identify the speech start point and end point, and generate a speech activity timestamp. S2213. Based on Fast Fourier Transform, each frame of speech segment is converted to the frequency domain, and the Mel frequency cepstral coefficients are calculated to capture the timbre features of the audio. The parameters of spectral entropy and spectral flatness are extracted to characterize the audio complexity. S2214. Based on the Mel frequency cepstral coefficient feature vector, combined with a pre-trained lightweight neural network, the audio content is classified, the corresponding classification label is output, and the importance level of the audio is evaluated by combining the speech activity detection results and energy features. S2215. Analyze the continuity and time sensitivity of the audio stream, generate latency tolerance labels, integrate classification labels, importance levels and latency tolerance labels to obtain semantic labels, so as to clarify the service priority and transmission requirements of different audio streams.

6. The ultra-large-scale audio routing method according to claim 1, characterized in that, The process of determining the audio transmission protocol based on real-time requirements, transmission scale, network environment, and device compatibility, and automatically generating audio routing instructions based on the target audio transmission path, audio transmission protocol, and globally unique identifier, combined with signal flow direction, includes the following steps: S31. Comprehensively evaluate the real-time requirements, transmission scale, network environment and equipment compatibility of the target scenario, determine the audio transmission protocol for the target scenario, and ensure that the selected audio transmission protocol is compatible with each audio device in the system to avoid situations where the protocol is not supported. S32. Based on the target audio transmission path, combined with the audio transmission protocol and the globally unique identifier of each device, automatically generate audio routing instructions according to the direction of signal flow from the source device through the processing node to the target device.

7. The ultra-large-scale audio routing method according to claim 6, characterized in that, The audio transmission protocol for the target scenario includes any one of the AES67 protocol, MADI protocol, and DANTE protocol; the audio routing instructions include information on physical port mapping, protocol parameter configuration, and quality of service settings.

8. The ultra-large-scale audio routing method according to claim 1, characterized in that, The audio data is encrypted using acoustic fingerprint dynamic encryption technology, and the encrypted audio data is sent to the audio device in the target scene based on audio routing instructions. The audio device receives encrypted audio data and decrypts it to complete the audio data transmission, which includes the following steps: S41. Calculate the Mel-spectral centroid of each audio frame and use the calculation result as the key seed; S42. Based on the key seed value, select the corresponding offset from the chaotic sequence generated by the quantum random number generator, and perform a cyclic shift operation on the audio sampling points according to the offset to achieve audio data encryption. S43. Use the offset as a key encryption parameter and embed it into the metadata field of the audio routing instruction to generate an audio routing instruction containing the encryption parameter. S44. Based on the audio routing instruction containing encryption parameters, combined with the audio transmission protocol and the target audio transmission path, the encrypted audio data is sent to the audio device in the target scene. S45. The target audio device receives the encrypted audio data and audio routing instructions, and extracts the encrypted key parameters from the metadata of the audio routing instructions. S46. The target audio device uses the extracted encryption key parameters to perform a reverse cyclic displacement operation on the encrypted audio sampling points to obtain the original audio content and complete the transmission of audio data.

9. A method for ultra-large-scale audio routing according to claim 8, characterized in that, The calculation of the Mel-spectral centroid of each audio frame, and the use of the calculation result as a key seed, includes: Each audio frame is converted from the time domain to the frequency domain, the frequency domain signal is processed by the Mel filter bank, and the centroid position of the Mel spectrum is calculated to generate a key seed that is specific to the audio content.

10. A method for ultra-large-scale audio routing according to claim 8, characterized in that, The process of encrypting audio data by selecting a corresponding offset from a chaotic sequence generated by a quantum random number generator based on a key seed value and performing a cyclic shift operation on the audio sampling points according to the offset includes the following steps: S421. Using modular arithmetic, the barycenter of the Mel spectrum of the audio frame is mapped to the index space of the chaotic sequence of the quantum random number generator, and the index is preprocessed using the salt value in the globally unique identifier. S422. Based on the preprocessed index, extract several consecutive random numbers from the chaotic sequence pre-generated by the quantum random number generator, and generate the final offset through XOR operation. S423. Divide the audio frames into blocks of a preset length according to the time sequence, encrypt each block independently, and introduce a random increment in the offset between each block. S424. Modulate the encrypted sampling points using the phase and synchronize the offset and phase modulation parameters to the audio device in the target scene through a low bit rate channel.

Images