AI Dialogue Optimization Methods, Systems, Devices, and Media Based on MAC CRC Statistical Features

Abstract

This invention discloses an AI dialogue optimization method, system, device, and medium based on MAC CRC statistical features, relating to the field of intelligent voice interaction technology. The method includes: a MAC layer performing cyclic redundancy check (CRC) on received data packets and generating a MAC CRC check result; the MAC layer calculating at least one statistical feature parameter reflecting the characteristics of audio burst noise based on the MAC CRC check result; the MAC layer synchronizing the MAC CRC check result and the statistical feature parameter to the application layer; the application layer determining whether a data packet is suspicious based on the MAC CRC check result; if a data packet is suspicious, the application layer combining the statistical feature parameter and the real-time audio frame parameters of the data packet decides whether to initiate application-layer CRC verification; and the application layer performing corresponding processing operations on the data packet according to the CRC check decision. This invention achieves cross-layer collaborative optimization, significantly improving the real-time performance, accuracy, and system energy efficiency of voice interaction without requiring audio buffering.

Description

Technical Field

[0001] This invention relates to the field of intelligent voice interaction technology, and in particular to an AI dialogue optimization method, system, device and medium based on MAC CRC statistical features. Background Technology

[0002] With the deep integration of IoT and AI technologies, intelligent voice interaction devices are increasingly widely used in terminals such as toys and walkie-talkies. In these practical application scenarios, the real-time performance and robustness of audio transmission are particularly critical. Currently, mainstream audio processing solutions mostly employ efficient audio codecs such as Opus, and optimize them to improve voice quality. Within existing optimization frameworks, technical implementation typically relies on extracting application-layer acoustic feature parameters from the audio stream, such as Mel-frequency cepstral coefficients and pulse-code modulation energy (PCM) frame-level parameters. Based on these, application-layer cyclic redundancy checks are performed on the audio information carrying key semantics to ensure data integrity.

[0003] However, this application-layer-centric optimization approach has revealed several inherent flaws in practice. First, to ensure sufficient reliability and demonstrate the value of cyclic redundancy check (CRC), the encoder typically needs to accumulate and buffer a certain amount of audio data before initiating the verification process. This buffering mechanism inevitably introduces processing latency, thereby impairing the real-time performance of voice interaction. In dialogue scenarios requiring immediate feedback, such latency can lead to voice interruptions or sluggish responses, severely impacting the user experience. Second, while enabling CRC at the application layer in a fixed and unconditional manner can comprehensively check for data errors, it also continuously consumes considerable computing resources, generating unnecessary computational overhead. For terminal devices with limited computing power, this undoubtedly increases their processing burden, hindering energy efficiency control and cost optimization.

[0004] On the other hand, completely bypassing application-layer verification and relying solely on the native cyclic redundancy check (CRC) mechanism provided by the media access control (Media Access Control) layer in the communication protocol stack also presents challenges. The CRC layer's verification mechanism is primarily designed to detect bit errors introduced during physical transmission. Its detection capability is quite limited for audio distortions at the semantic level caused by intermittent environmental noise, such as multiple people speaking simultaneously or sudden interference. Such missed detections can lead to a series of problems, including incorrectly discarding actually valid voice data packets or performing ineffective error correction without specific targeting, ultimately manifesting as impulse noise, stuttering, or speech recognition errors during audio playback.

[0005] Therefore, the existing technological system presents a prominent contradiction: while application-layer verification methods are sophisticated, they are not real-time enough and consume significant resources; while media access control layer verification methods, although fast and direct, are not intelligent enough and cannot effectively handle semantic noise. In summary, the core issue of the current technological state is: how to construct an audio optimization method and system architecture that can effectively reuse underlying verification information and make real-time adaptive decisions based on it, thereby ensuring the real-time performance of voice interaction while balancing verification accuracy and system energy efficiency. Summary of the Invention

[0006] This invention provides an AI dialogue optimization method, system, device, and medium based on MAC CRC statistical features, aiming to solve the problem of how to ensure the real-time performance of voice interaction while taking into account the accuracy of verification and the energy efficiency of the system.

[0007] In a first aspect, embodiments of the present invention provide an AI dialogue optimization method based on MAC CRC statistical features, comprising: S1, the MAC layer performs cyclic redundancy check (CRC) on the received data packets and generates a MAC CRC check result; S2, the MAC layer calculates at least one statistical characteristic parameter reflecting the characteristics of audio burst noise based on the MAC CRC check result; S3, the MAC layer synchronizes the MAC CRC check result and the statistical feature parameters to the application layer; S4, the application layer determines whether the data packet is a suspicious data packet based on the MAC CRC check result; S5, if the data packet is a suspicious data packet, the application layer combines the statistical feature parameters with the real-time audio frame parameters of the data packet to decide whether to initiate application layer CRC verification for the data packet; S6, the application layer performs corresponding processing operations on the data packet based on the decision result of the CRC check.

[0008] A further technical solution is that the statistical characteristic parameters in step S2 include at least one of the following: The variance of the maximum consecutive failure length calculated based on MAC CRC failures within the sliding time window; The density of failed or suspicious clusters above the threshold, calculated based on MAC CRC failures within a sliding time window.

[0009] A further technical solution is that, in step S3, the MAC CRC check result and the statistical characteristic parameters are called back to the application layer through the Hardware Abstraction Layer (HAL) application programming interface (API) provided by the firmware of the communication module.

[0010] A further technical solution is that, in step S1, the MAC CRC check result is characterized by setting a suspicious flag bit in the extended field of the MAC layer; In step S4, the application layer determines whether the data packet is a suspicious data packet by reading the status of the suspicious flag bit.

[0011] A further technical solution is that, in step S5, the application layer combines the statistical feature parameters and the real-time audio frame parameters to perform lightweight feature cluster classification, and decides whether to start the application layer CRC check based on the classification results.

[0012] A further technical solution is that the processing operation in step S6 includes at least one of the following: When the decision is not to initiate application-layer CRC check or the application-layer CRC check passes, the data packet is processed normally. The data packet is discarded when the application layer CRC check fails. When the application layer CRC check fails, a retransmission request is sent to the data sender.

[0013] A further technical solution is that the communication module is a 4G communication module.

[0014] Secondly, embodiments of the present invention also provide an AI dialogue optimization system based on MAC CRC statistical features, which includes a unit for performing the above-described method.

[0015] Thirdly, embodiments of the present invention also provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0016] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the above-described method.

[0017] This invention provides an AI dialogue optimization method, system, device, and medium based on MAC CRC statistical features. The method includes: the MAC layer performing cyclic redundancy check (CRC) on received data packets and generating a MAC CRC check result; the MAC layer calculating at least one statistical feature parameter reflecting the characteristics of audio burst noise based on the MAC CRC check result; the MAC layer synchronizing the MAC CRC check result and the statistical feature parameter to the application layer; the application layer determining whether the data packet is a suspicious data packet based on the MAC CRC check result; if the data packet is suspicious, the application layer combining the statistical feature parameter and the real-time audio frame parameters of the data packet to decide whether to initiate application-layer CRC check on the data packet; and the application layer performing corresponding processing operations on the data packet according to the CRC check decision result. This invention performs CRC check at the MAC layer and generates statistical feature parameters reflecting audio burst noise, which are then synchronized to the application layer. The application layer quickly filters suspicious data packets and integrates network parameters and real-time audio frame parameters for intelligent decision-making, initiating application-layer check as needed. This method achieves cross-layer collaborative optimization, significantly improving the real-time performance, accuracy, and system energy efficiency of voice interaction without the need for audio buffering. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating an AI dialogue optimization method based on MAC CRC statistical features provided in an embodiment of the present invention; Figure 2 This is a schematic block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0020] 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, not all, of the embodiments of the present invention. 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.

[0021] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0022] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0023] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0024] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0025] This invention proposes an AI dialogue optimization system based on MAC CRC statistical features. The system is used to execute an AI dialogue optimization method based on MAC CRC statistical features proposed in this invention. The system includes: Main control base plate; The AI ​​voice board is connected to the main control baseboard via a serial port. The AI ​​voice board includes a communication module and an application processor; The communication module is configured with a MAC layer, which is used to perform cyclic redundancy check (CRC) on the received data packets and generate a MAC CRC check result; based on the MAC CRC check result, at least one statistical feature parameter reflecting the characteristics of audio burst noise is calculated; and the MAC CRC check result and the statistical feature parameter are synchronized to the application layer. The application processor is configured with an application layer, which is used to determine whether the data packet is a suspicious data packet based on the MAC CRC check result; if the data packet is a suspicious data packet, it combines the statistical feature parameters and the real-time audio frame parameters of the data packet to decide whether to start the application layer CRC check on the data packet; and performs corresponding processing operations on the data packet according to the decision result of the CRC check.

[0026] It should be noted that in the Open Systems Interconnection (OSI) reference model or related communication protocol architecture, the "MAC layer" specifically refers to the "Media Access Control layer." This layer is a core sublayer of the data link layer, located above the physical layer. Its main technical function is to act as the scheduling and management hub for data packets transmitted over the physical medium. Specifically, the MAC layer is responsible for encapsulating the data delivered from the upper layer into data frame structures that conform to the specified protocols, and for performing channel access control, such as through mechanisms like carrier sense multiple access / collision avoidance or time division multiple access, to coordinate the access order when multiple terminal devices share the same physical transmission medium, thereby avoiding or reducing data collisions. Furthermore, a fundamental and crucial function of the MAC layer is to perform cyclic redundancy check on sent and received data frames to detect errors and ensure the integrity of data during physical transmission.

[0027] In communication protocol architecture, the "application layer" is located at the top of the protocol stack. It is a logical entity that directly provides communication services to application processes to support various business implementations. Its technical function is to define the interaction protocols, data formats, and processing logic related to the business needs of end users, rather than focusing on the low-level details of data transmission. In this invention, the application layer is the core functional layer for implementing audio business logic and intelligent decision-making. It is responsible for the operation of audio codecs (such as Opus), the real-time extraction of audio frame parameters (such as MFCC and PCM energy), and the interaction with backend artificial intelligence services (such as large language models).

[0028] Specifically, please refer to Figure 1 This invention provides an AI dialogue optimization method based on MAC CRC statistical features. By constructing a cross-protocol layer collaborative processing and decision-making mechanism, it achieves significant and organically unified technical effects in improving voice interaction quality, ensuring real-time communication, and optimizing system energy efficiency. Specifically, the method includes the following steps: S1, the MAC layer performs cyclic redundancy check (CRC) on the received data packets and generates the MAC CRC check result.

[0029] In practice, the MAC layer performs a Cyclic Redundancy Check (CRC) on the received data packets and generates a verification result. This initial step forms the foundation of the entire method. Its technical effect lies in completing the first, fast, and low-level screening of data packet transmission integrity. Compared to relying entirely on the application layer for verification, this step fully utilizes the inherent, highly efficient error detection capabilities of the communication protocol stack's underlying layers, providing the most basic trustworthy status information for subsequent processing.

[0030] S2, the MAC layer calculates at least one statistical characteristic parameter reflecting the characteristics of audio burst noise based on the MAC CRC check result.

[0031] In practice, the MAC layer further calculates statistical characteristic parameters reflecting the characteristics of audio burst noise based on the CRC check result. This is a key step in transforming discrete low-level check events into network quality indicators with continuity and pattern significance. It can extract channel noise pattern characteristics closely related to the upper-layer audio perception quality from the transmission errors of the bit stream, such as the burstiness, continuity or clustering of errors, thus providing the application layer with much richer contextual information than a single check result for intelligent decision-making.

[0032] In some preferred embodiments, the statistical characteristic parameters in step S2 include at least one of the following: the variance of the maximum consecutive failure length calculated based on the MAC CRC failures within the sliding time window; and the density of failed or suspicious clusters above the threshold calculated based on the MAC CRC failures within the sliding time window.

[0033] In practical implementation, the variance of the maximum consecutive failure length serves to quantify the clustering and fluctuation intensity of errors in the transmission channel. For example, brief, isolated CRC errors may have limited impact on the audio stream, but consecutive CRC failure clusters, especially those with drastically varying lengths, strongly suggest that the channel is experiencing persistent burst interference. This interference pattern is highly correlated with burst noise in audio conversations. Calculating the variance of this length effectively captures the discreteness and instability of such error sequences, providing the application layer with a quantitative indicator of the intensity and fluctuation patterns of noise bursts.

[0034] On the other hand, the density parameter of failed or suspicious clusters exceeding the threshold aims to measure the concentration of error events in a spatial or temporal dimension. This parameter reveals whether error events are sparsely distributed or densely bursting within a specific sliding time window. High-density error clusters typically correspond to scenarios with strong, continuous external noise interference, such as equipment in a noisy industrial environment or a meeting room where multiple people are talking simultaneously. By calculating this density, the system can assess the severity of the current channel environment and the coverage of noise impact. Combining these two statistical characteristic parameters, the application layer obtains not just isolated, instantaneous MAC layer verification results, but a dynamic and information-rich "portrait" of channel error patterns. This allows the application layer to make more accurate judgments about the severity and type of interference to the current data packet based on the continuity, volatility, and concentration of error occurrences, thereby making more reasonable verification decisions.

[0035] For example, when faced with a single MAC CRC error, if both variance and density parameters indicate that the channel has been stable recently, the system may determine it as a random error and choose not to start application layer CRC; conversely, if both parameters indicate that the channel is in a highly unstable and error-dense state, the system will start application layer deep check with a higher probability.

[0036] Specifically, the variance of the maximum consecutive failure length calculated based on MAC CRC failures within the sliding time window is obtained as follows: The firmware of the communication module is configured to continuously monitor the cyclic redundancy check (CRC) results of data packets performed by the Media Access Control (MAC) layer within a predefined sliding time window. The firmware first identifies and records each consecutive MAC CRC check failure sequence occurring within this time window, and calculates the length of each such sequence, i.e., the number of consecutively failed data packets within a single sequence. Subsequently, the maximum sequence length is extracted from all recorded consecutive failure sequences, termed the maximum consecutive failure length. The firmware further maintains a dynamic sequence of the maximum consecutive failure length over the duration of the sliding time window. Finally, the variance of this dynamic sequence is calculated. This variance is used to quantify the drastic fluctuation of the maximum consecutive failure length over time, reflecting the suddenness and instability of channel errors. A higher variance indicates that the channel is in a state of intermittent, severe, sudden errors.

[0037] Furthermore, the method for obtaining the density of failed or suspicious clusters exceeding the threshold, calculated based on MAC CRC failures within the sliding time window: The firmware of the communication module is configured to identify suspicious clusters based on MAC CRC check results within a predefined sliding time window. A suspicious cluster is defined as a set of consecutive or adjacent data packets in a data packet sequence where MAC CRC checks fail or the check result value exceeds a preset threshold. The firmware first detects and delineates the boundaries of all suspicious clusters meeting the above conditions within the sliding time window. Then, it calculates the overall metric of these suspicious clusters, which can be the total number of data packets covered by all suspicious clusters or the sum of the durations of all suspicious clusters. Finally, the overall metric is compared with the total number of data packets transmitted within the sliding time window or the total window duration to obtain the density of the suspicious clusters. This density parameter aims to characterize the concentration of time or data packet resources dominated by transmission errors or suspicious data packets within a statistical time period. Its technical purpose is to reveal the severity and scope of channel interference; a higher density value indicates that the channel is suffering from continuous and dense interference.

[0038] S3, the MAC layer synchronizes the MAC CRC check result and the statistical feature parameters to the application layer.

[0039] In practice, the MAC CRC check result and its statistical characteristic parameters are synchronized to the application layer. This step breaks down the technical barrier of parameter isolation between layers in the traditional network protocol stack. The core effect is the deep integration of network transmission status and upper-layer application semantics, enabling the application layer to know the real-time health status and noise characteristics of the underlying channel when processing audio data, thus creating the possibility for context-aware optimization decisions.

[0040] In some preferred embodiments, in step S3, the MAC CRC check result and the statistical characteristic parameters are sent back to the application layer via the Hardware Abstraction Layer (HAL) application programming interface (API) provided by the firmware of the communication module.

[0041] In practical implementation, the Hardware Abstraction Layer (HAL) is essentially an interface between the underlying hardware and the upper-layer software, designed to hide the specific implementation details of particular hardware. By using this established, standardized HAL API to handle cross-layer parameter transfer, the reliability and efficiency of the data synchronization process are ensured. Application layer developers do not need to concern themselves with the specific model, register configuration, or complex differences in drivers of the underlying communication modules; they only need to follow the unified API interface specification to reliably obtain the required MAC layer parameters. This greatly reduces the complexity of system integration and software development, and minimizes the instability risks that may be introduced by directly manipulating the underlying hardware. Secondly, this interface-based design gives the entire solution excellent hardware compatibility and portability. As long as the communication module manufacturer provides a compliant HAL API implementation in its firmware, this method can run smoothly on different hardware platforms, whether using 4G modules from different manufacturers or evolving to 5G modules in the future, the application layer code can be adapted without major modifications. This portability is crucial for the large-scale commercial deployment and ecosystem building of the solution. It allows equipment manufacturers to flexibly choose their supply chain without being tied to specific hardware, while also reserving a convenient channel for future firmware upgrades and feature iterations.

[0042] In some preferred embodiments, the communication module is a 4G communication module.

[0043] In practical implementation, firstly, the communication module is explicitly defined as a 4G communication module, placing this invention within a mature and widely deployed mobile communication technology context. 4G networks, characterized by wide coverage, high speed, and relatively low latency, are the mainstream choice for wireless connectivity among numerous IoT devices and smart terminals. Optimization based on a 4G module means this invention can be directly applied to a vast existing 4G device ecosystem, possessing enormous potential application market and technology promotion value. Secondly, and more practically significant, is the firmware configuration of the communication module via OTA (Over-The-Air) upgrades. OTA upgrades are a technology that remotely updates the internal firmware of a device via a wireless network, completely changing the traditional model that required physical contact or factory returns to update device functions. For the tens of thousands of devices already deployed on the market, adding the MAC layer statistical feature calculation and HAL API exposure functions involved in this invention via OTA upgrades becomes the only economically feasible approach. Device manufacturers or service providers can push new version module firmware containing new functions to terminal devices via cloud servers, allowing the terminal devices to complete self-upgrades without manual intervention, instantly gaining the optimized capabilities of this invention. This deployment method significantly reduces the costs of system maintenance and feature iteration, shortens the development-to-commercialization cycle of new features, and makes large-scale, rapid technology deployment possible. At the same time, it provides great flexibility, allowing developers to perform differentiated firmware configurations and feature activations for different batches and models of devices.

[0044] S4, the application layer determines whether the data packet is a suspicious data packet based on the MAC CRC check result.

[0045] In practice, the application layer determines whether a data packet is suspicious based on the received MAC CRC check result. This step achieves an efficient initial screening. By reusing the check work already completed at the lower level, it quickly identifies potentially problematic data packets, avoiding sending all data packets indiscriminately into computationally intensive processing flows, thus achieving initial savings in computing resources at the beginning of the decision-making chain.

[0046] In some preferred embodiments, in step S1, the MAC CRC check result is characterized by setting a suspicious flag bit in the extended field of the MAC layer; in step S4, the application layer determines whether the data packet is a suspicious data packet by reading the status of the suspicious flag bit.

[0047] In practice, the MAC CRC check result is condensed into a binary or multi-valued suspicious flag bit and embedded in the MAC layer's inherent extended field. This is a lightweight reuse and extension of the existing protocol frame structure. This method adds almost no additional protocol header overhead, yet can carry critical status information. For the application layer, the judgment logic is simplified to the greatest extent. It does not need to parse complex underlying protocol data or perform cumbersome calculations; a single fast status bit read operation is sufficient to complete the initial classification of the packet's suspiciousness. This simplified interaction mode brings two significant benefits. First, it improves processing speed. Flag-based status judgment is one of the most basic and fastest operations in a computer system, ensuring that the system can maintain real-time response even in high-data-throughput scenarios, which is crucial for latency-sensitive voice dialogue applications. Second, it saves system resources. The application layer does not need to resident complex parsing threads or consume a large number of CPU cycles to actively probe the MAC layer status; it can obtain the key inputs needed for decision-making only through passive flag bit reading, which reduces the consumption of the main control processor's computing resources. In addition, this flag mechanism also enables data packets to carry "identity tags" when traversing different layers of the protocol stack. The application layer can quickly sort data packets based on these tags, laying the foundation for subsequent differentiated processing.

[0048] S5. If the data packet is a suspicious data packet, the application layer combines the statistical feature parameters with the real-time audio frame parameters of the data packet to decide whether to initiate application layer CRC verification for the data packet.

[0049] In practice, for packets marked as suspicious, the application layer combines statistical feature parameters from the MAC layer with the packet's own real-time audio frame parameters to jointly decide whether to initiate application-layer CRC verification. This is the core of this method's intelligent adaptive mechanism. The technical benefits of this step are multifaceted: First, by fusing noise features from the network layer with acoustic features from the audio content layer, it constructs a more comprehensive evaluation system, significantly improving the accuracy of judging the true damage status of data packets and reducing the probability of false positives and false negatives. Second, this decision-making mechanism based on multi-source information fusion transforms the relatively time-consuming operation of application-layer CRC verification from a fixed or blindly triggered process into a dynamic process that is initiated precisely on demand. This directly avoids the waste of computational resources caused by unnecessary verification under good channel conditions and ensures that deep verification can be initiated promptly to guarantee quality when the channel deteriorates. Most importantly, the entire decision-making process can be executed without waiting for the complete audio data buffer to finish, a characteristic that brings crucial real-time performance. It enables the system to react almost instantly to network fluctuations and noise interference, greatly reducing the time delay from problem detection to initiating countermeasures, which is crucial for maintaining the fluency and naturalness of voice conversations.

[0050] In some preferred embodiments, in step S5, the application layer combines the statistical feature parameters with the real-time audio frame parameters to perform lightweight feature cluster classification, and decides whether to start application layer CRC verification based on the classification results.

[0051] In practical implementation, firstly, "lightweight feature cluster classification" means that the algorithm used at the application layer is an optimized, computationally low-computational-complexity machine learning or pattern recognition method. Its purpose is to quickly classify the mixed feature vector composed of network statistical features and audio frame parameters, rather than performing complex deep analysis. This lightweight characteristic ensures that the decision-making process itself will not become a new performance bottleneck for the system, meeting the stringent requirements of terminal devices for low power consumption and real-time computing. Secondly, and more importantly, is the ability to "execute decisions without waiting for the complete audio data buffer to finish." In traditional audio processing workflows, to extract more representative features or perform batch processing, it is often necessary to cache audio data for a certain period of time, which inevitably introduces inherent processing latency. This embodiment completely eliminates this buffering and waiting step by requiring decisions to be made immediately after the data packet arrives, using real-time audio frame parameters (such as the energy of the current frame, zero-crossing rate, etc.) combined with network parameters. This enables the system to achieve near real-time, packet-by-packet or frame-by-frame dynamic strategy adjustments. For example, when the system detects burst noise in the channel through MAC layer parameters and immediately determines that the frame may be corrupted based on the current audio frame parameters, it can immediately decide to initiate an application-layer CRC check on the data packet. If the check fails, it can quickly initiate a retransmission or discard the packet, without waiting for multiple subsequent audio frames to arrive and form a complete audio segment before unified processing. This instantaneous response capability greatly shortens the time window from the occurrence of an error to the system taking corrective measures, which is crucial for maintaining the coherence and naturalness of voice dialogue. It effectively avoids problems such as voice stuttering, audio segment loss, or slow response caused by processing delays.

[0052] Specifically, step S5 is implemented by including the following steps: First, a composite feature vector is constructed. This vector consists of at least two parts: the first part is the statistical feature parameters from the media access control layer, namely the variance of the maximum consecutive failure length and the density of failed or suspicious clusters above the threshold; the second part is the audio frame parameters extracted in real time from the data packet to be decided, which may include, but are not limited to, short-time energy, zero-crossing rate, or Mel frequency cepstral coefficients.

[0053] Subsequently, the composite feature vector is input into a pre-trained classification model. This classification model is configured to perform lightweight feature cluster classification, and its specific implementation can be one of decision trees, support vector machines, or lightweight neural networks. The classification model is trained based on historical data, and its learning objective is to establish a mapping relationship between the composite feature vector and two decision results: "application-layer CRC check needs to be initiated" or "application-layer CRC check does not need to be initiated".

[0054] Finally, a decision is made based on the output of the classification model. If the model output classifies the data packet as requiring verification, an application-layer CRC check is triggered for the current data packet; if the model output classifies the data packet as not requiring verification, the data packet is marked as trustworthy and handed over to the subsequent audio processing module for normal decoding and playback.

[0055] The entire decision-making process is completed at the granularity of a single data packet or a single audio frame, without waiting for the arrival of subsequent data packets, thus achieving bufferless real-time decision-making and effectively ensuring the low-latency characteristics of voice interaction.

[0056] S6, the application layer performs corresponding processing operations on the data packet based on the decision result of the CRC check.

[0057] In practice, the application layer performs corresponding processing operations on data packets based on the CRC check result, such as normal playback, discarding, or requesting retransmission. This constitutes the execution loop of the method. The technical effect of this step is to transform all the preceding analysis and decisions into concrete actions that directly impact the user experience, ensuring that optimization strategies are effectively implemented and that the most appropriate processing method is adopted according to different data packet quality conditions, thereby achieving refined control over the voice communication process at the system level.

[0058] In some preferred embodiments, the processing operation in step S6 includes at least one of the following: when the decision is not to initiate application layer CRC verification or the application layer CRC verification passes, the data packet is processed normally; when the application layer CRC verification fails, the data packet is discarded; when the application layer CRC verification fails, a retransmission request is sent to the data sender.

[0059] In practice, the processing operations constitute a differentiated response strategy for different data packet quality conditions. When the decision is not to initiate application-layer CRC verification or the application-layer CRC verification passes, the data packet is processed normally. This ensures that when the channel conditions are good or the data packet is deemed safe, the system can transmit and play audio with maximum efficiency, avoiding any unnecessary interference with the normal communication process. When the decision is to initiate application-layer CRC verification and the application-layer CRC verification fails, two options are provided: discarding the data packet and requesting retransmission. This provides fault tolerance for confirmed errors. Directly discarding the data packet is the fastest error correction method. It is suitable for scenarios with extremely high real-time requirements, where a small amount of packet loss is acceptable and can be compensated for by voice recovery techniques such as frame interpolation. Its effect is to quickly cut off the propagation of erroneous data, preventing broken audio data from being played and producing harsh impulse noise or distorted semantics, thereby protecting the auditory experience. Furthermore, by requesting the sender to retransmit the data packet, the system has the opportunity to obtain a correct copy, thereby fundamentally repairing the erroneous segments in the voice stream and ensuring the integrity and comprehensibility of the dialogue content.

[0060] This invention proposes an AI dialogue optimization method based on MAC CRC statistical features, comprising: the MAC layer performing cyclic redundancy check (CRC) on received data packets and generating a MAC CRC check result; the MAC layer calculating at least one statistical feature parameter reflecting the characteristics of audio burst noise based on the MAC CRC check result; the MAC layer synchronizing the MAC CRC check result and the statistical feature parameter to the application layer; the application layer determining whether the data packet is a suspicious data packet based on the MAC CRC check result; if the data packet is a suspicious data packet, the application layer combining the statistical feature parameter and the real-time audio frame parameters of the data packet to decide whether to initiate application-layer CRC check on the data packet; and the application layer performing corresponding processing operations on the data packet according to the decision result of the CRC check. This invention performs CRC check at the MAC layer and generates statistical feature parameters reflecting audio burst noise, which are then synchronized to the application layer. The application layer can quickly filter suspicious data packets and make intelligent decisions by integrating network parameters and real-time audio frame parameters, initiating application-layer checks as needed. This method achieves cross-layer collaborative optimization, significantly improving the real-time performance, accuracy, and system energy efficiency of voice interaction without the need for audio buffering.

[0061] Please see Figure 2 , Figure 2This is a schematic block diagram of a computer device provided in an embodiment of this application. The computer device 500 can be a terminal or a server. The terminal can be an electronic device with communication functions, such as a smartphone, tablet, laptop, desktop computer, personal digital assistant, or wearable device. The server can be a standalone server or a server cluster composed of multiple servers.

[0062] The computer device 500 includes a processor 502, a memory, and a network interface 505 connected via a system bus 501. The memory may include a non-volatile storage medium 503 and internal memory 504.

[0063] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, it causes the processor 502 to execute an AI dialogue optimization method based on MAC CRC statistical characteristics.

[0064] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0065] The internal memory 504 provides an environment for the execution of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute an AI dialogue optimization method based on MAC CRC statistical features.

[0066] The network interface 505 is used for network communication with other devices. Those skilled in the art will understand that the above structure is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device 500 to which the present application is applied. A specific computer device 500 may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements.

[0067] The processor 502 is used to run a computer program 5032 stored in the memory to perform the following steps: S1, the MAC layer performs cyclic redundancy check (CRC) on the received data packets and generates a MAC CRC check result; S2, the MAC layer calculates at least one statistical characteristic parameter reflecting the characteristics of audio burst noise based on the MAC CRC check result; S3, the MAC layer synchronizes the MAC CRC check result and the statistical feature parameters to the application layer; S4, the application layer determines whether the data packet is a suspicious data packet based on the MAC CRC check result; S5, if the data packet is a suspicious data packet, the application layer combines the statistical feature parameters with the real-time audio frame parameters of the data packet to decide whether to initiate application layer CRC verification for the data packet; S6, the application layer performs corresponding processing operations on the data packet based on the decision result of the CRC check.

[0068] In some preferred embodiments, the statistical characteristic parameters in step S2 include at least one of the following: The variance of the maximum consecutive failure length calculated based on MAC CRC failures within the sliding time window; The density of failed or suspicious clusters above the threshold, calculated based on MAC CRC failures within a sliding time window.

[0069] In some preferred embodiments, in step S3, the MAC CRC check result and the statistical characteristic parameters are sent back to the application layer via the Hardware Abstraction Layer (HAL) application programming interface (API) provided by the firmware of the communication module.

[0070] In some preferred embodiments, in step S1, the MAC CRC check result is characterized by setting a suspicious flag bit in the extended field of the MAC layer; In step S4, the application layer determines whether the data packet is a suspicious data packet by reading the status of the suspicious flag bit.

[0071] In some preferred embodiments, in step S5, the application layer combines the statistical feature parameters with the real-time audio frame parameters to perform lightweight feature cluster classification, and decides whether to start application layer CRC verification based on the classification results.

[0072] In some preferred embodiments, the processing operation in step S6 includes at least one of the following: When the decision is not to initiate application-layer CRC check or the application-layer CRC check passes, the data packet is processed normally. The data packet is discarded when the application layer CRC check fails. When the application layer CRC check fails, a retransmission request is sent to the data sender.

[0073] In some preferred embodiments, the communication module is a 4G communication module.

[0074] It should be understood that in the embodiments of this application, the processor 502 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0075] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program may be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0076] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program. When executed by a processor, the computer program causes the processor to perform the following steps: S1, the MAC layer performs cyclic redundancy check (CRC) on the received data packets and generates a MAC CRC check result; S2, the MAC layer calculates at least one statistical characteristic parameter reflecting the characteristics of audio burst noise based on the MAC CRC check result; S3, the MAC layer synchronizes the MAC CRC check result and the statistical feature parameters to the application layer; S4, the application layer determines whether the data packet is a suspicious data packet based on the MAC CRC check result; S5, if the data packet is a suspicious data packet, the application layer combines the statistical feature parameters with the real-time audio frame parameters of the data packet to decide whether to initiate application layer CRC verification for the data packet; S6, the application layer performs corresponding processing operations on the data packet based on the decision result of the CRC check.

[0077] In some preferred embodiments, the statistical characteristic parameters in step S2 include at least one of the following: The variance of the maximum consecutive failure length calculated based on MAC CRC failures within the sliding time window; The density of failed or suspicious clusters above the threshold, calculated based on MAC CRC failures within a sliding time window.

[0078] In some preferred embodiments, in step S3, the MAC CRC check result and the statistical characteristic parameters are sent back to the application layer via the Hardware Abstraction Layer (HAL) application programming interface (API) provided by the firmware of the communication module.

[0079] In some preferred embodiments, in step S1, the MAC CRC check result is characterized by setting a suspicious flag bit in the extended field of the MAC layer; In step S4, the application layer determines whether the data packet is a suspicious data packet by reading the status of the suspicious flag bit.

[0080] In some preferred embodiments, in step S5, the application layer combines the statistical feature parameters with the real-time audio frame parameters to perform lightweight feature cluster classification, and decides whether to start application layer CRC verification based on the classification results.

[0081] In some preferred embodiments, the processing operation in step S6 includes at least one of the following: When the decision is not to initiate application-layer CRC check or the application-layer CRC check passes, the data packet is processed normally. The data packet is discarded when the application layer CRC check fails. When the application layer CRC check fails, a retransmission request is sent to the data sender.

[0082] In some preferred embodiments, the communication module is a 4G communication module.

[0083] The storage medium is a physical, non-transient storage medium, such as a USB flash drive, external hard drive, read-only memory (ROM), magnetic disk, or optical disk, or any other physical storage medium capable of storing program code. The computer-readable storage medium can be non-volatile or volatile.

[0084] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0085] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0086] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0087] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0088] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0089] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Since these modifications and variations fall within the scope of the claims and their equivalents, this invention also intends to include these modifications and variations.

[0090] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An AI conversation optimization method based on MAC CRC statistical characteristics, characterized in that, include: S1, the MAC layer performs cyclic redundancy check (CRC) on the received data packets and generates a MAC CRC check result; S2, the MAC layer calculates at least one statistical characteristic parameter reflecting the characteristics of audio burst noise based on the MAC CRC check result; S3, the MAC layer synchronizes the MAC CRC check result and the statistical feature parameters to the application layer; S4, the application layer determines whether the data packet is a suspicious data packet based on the MAC CRC check result; S5, if the data packet is a suspicious data packet, the application layer combines the statistical feature parameters with the real-time audio frame parameters of the data packet to decide whether to initiate application layer CRC verification for the data packet; S6, the application layer performs corresponding processing operations on the data packet based on the decision result of the CRC check. 2.The AI conversation optimization method based on MAC CRC statistic features according to claim 1, wherein, The statistical characteristic parameters in step S2 include at least one of the following: The variance of the maximum consecutive failure length calculated based on MAC CRC failures within the sliding time window; The density of failed or suspicious clusters above the threshold, calculated based on MAC CRC failures within a sliding time window. 3.The AI conversation optimization method based on MAC CRC statistic features according to claim 1, wherein, In step S3, the MAC CRC check result and the statistical characteristic parameters are sent back to the application layer through the Hardware Abstraction Layer (HAL) application programming interface (API) provided by the firmware of the communication module. 4.The AI conversation optimization method based on MAC CRC statistic features according to claim 1, wherein, In step S1, the MAC CRC check result is characterized by setting a suspicious flag bit in the extended field of the MAC layer; In step S4, the application layer determines whether the data packet is a suspicious data packet by reading the status of the suspicious flag bit.

5. The AI ​​dialogue optimization method based on MAC CRC statistical features according to claim 1, characterized in that, In step S5, the application layer combines the statistical feature parameters with the real-time audio frame parameters to perform lightweight feature cluster classification, and decides whether to start application layer CRC verification based on the classification results.

6. The AI ​​dialogue optimization method based on MAC CRC statistical features according to claim 1, characterized in that, The processing operation in step S6 includes at least one of the following: When the decision is not to initiate application-layer CRC check or the application-layer CRC check passes, the data packet is processed normally. The data packet is discarded when the application layer CRC check fails. When the application layer CRC check fails, a retransmission request is sent to the data sender.

7. The AI ​​dialogue optimization method based on MAC CRC statistical features according to claim 3, characterized in that, The communication module is a 4G communication module.

8. An AI dialogue optimization system based on MAC CRC statistical features, characterized in that, The system for performing the method as described in any one of claims 1-7 includes: Main control base plate; The AI ​​voice board is connected to the main control baseboard via a serial port. The AI ​​voice board includes a communication module and an application processor; The communication module is configured with a MAC layer, which is used to perform cyclic redundancy check (CRC) on the received data packets and generate a MAC CRC check result; based on the MAC CRC check result, at least one statistical feature parameter reflecting the characteristics of audio burst noise is calculated; and the MAC CRC check result and the statistical feature parameter are synchronized to the application layer. The application processor is configured with an application layer, which is used to determine whether the data packet is a suspicious data packet based on the MAC CRC check result; if the data packet is a suspicious data packet, it combines the statistical feature parameters and the real-time audio frame parameters of the data packet to decide whether to start the application layer CRC check on the data packet; and performs corresponding processing operations on the data packet according to the decision result of the CRC check.

9. A computer device, characterized in that, The computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, can implement the method as described in any one of claims 1-7.

Images