Method for increasing decoding speed of web radio and podcasts bitstreams

Abstract

The application provides a method for improving decoding speed of network radio and podcasts code stream, and belongs to the fields of audio decoding technology and network communication and transmission technology.The method comprises the following steps: initially analyzing the code stream of the network radio and podcasts, obtaining code stream basic information, and constructing a code stream characteristic model according to the code stream basic information; implementing an adaptive code stream allocation strategy based on the code stream characteristic model, and generating an adaptive code stream allocation scheme; and simultaneously evaluating the multi-core processor resources, and constructing a parallel decoding resource model.Through the adaptive code stream allocation strategy and the parallel decoding technology, the network bandwidth and the computing resources can be more efficiently utilized, the overall speed of audio decoding can be improved, the delay can be reduced, and the fluency of audio playing can be ensured.

Description

Technical Field

[0001] This invention proposes a method to improve decoding speed based on internet radio and podcast streams, belonging to the fields of audio decoding technology and network communication and transmission technology. Background Technology

[0002] With the rapid development of internet technology, internet radio and podcasts have gained increasing popularity among users due to their rich content resources and convenient access methods. Internet radio can receive broadcast signals from all over the world in real time, while podcasts provide users with a vast library of audio programs covering many fields such as news, music, and education.

[0003] However, in practical use, decoding speed becomes a key factor affecting user experience. Instability in the network environment, such as bandwidth fluctuations and network latency, can obstruct audio stream transmission, thus affecting the timeliness of decoding. At the same time, traditional decoding algorithms suffer from high computational load and low decoding efficiency when processing complex audio formats or high-bitrate streams, easily leading to stuttering and buffering issues.

[0004] Furthermore, different devices have varying hardware performance; some devices have limited processor performance, making it difficult to complete decoding tasks quickly. To improve the user listening experience and meet users' demands for smooth, high-quality audio playback, there is an urgent need for an effective method to improve decoding speed, optimize the playback effect of internet radio and podcasts, and promote the further development of the audio streaming industry. Summary of the Invention

[0005] This invention provides a method for improving decoding speed based on internet radio and podcast streams, in order to solve the problems mentioned in the background section above:

[0006] The present invention proposes a method for improving decoding speed based on internet radio and podcast streams, the method comprising:

[0007] S1: Perform initial parsing of the bitstreams of internet radio and podcasts to obtain basic bitstream information, and construct a bitstream feature model based on the basic bitstream information;

[0008] S2: Based on the bitstream feature model, implement an adaptive bitstream allocation strategy and generate an adaptive bitstream allocation scheme; at the same time, evaluate multi-core processor resources and construct a parallel decoding resource model;

[0009] S3: According to the adaptive bitstream allocation scheme, the bitstream data is divided into multiple sub-bitstream blocks. Using the parallel decoding resource model, each sub-bitstream block is allocated to different cores of the multi-core processor to perform parallel decoding processing and obtain multiple parallel decoding results. The multiple parallel decoding results are synchronously integrated to generate preliminary decoded data.

[0010] S4: Perform quality assessment on the initial decoded data and generate decoding quality feedback data; based on the decoding quality feedback data, fine-tune the adaptive bitstream allocation scheme, and dynamically adjust the resource allocation during the parallel decoding process;

[0011] S5: Based on the optimized adaptive bitstream allocation scheme and parallel decoding resource model, perform final decoding processing to generate high-quality decoded bitstream data; evaluate the smoothness of the high-quality decoded bitstream data, predict the listening smoothness under different network fluctuation conditions, and generate decoding smoothness prediction data;

[0012] S6: Based on decoding smoothness prediction data, implement a risk warning mechanism. When potential stuttering or decoding delay is predicted, adjust the decoding strategy in advance or alert the user to the network status, and generate decoding warning data for internet radio and podcasts.

[0013] The beneficial effects of this invention are as follows: Through adaptive bitstream allocation strategies and parallel decoding technology, network bandwidth and computing resources can be utilized more efficiently, improving the overall speed of audio decoding, reducing latency, and ensuring smooth audio playback. Real-time monitoring and dynamic adjustment of network bandwidth allows for flexible adjustment of bitstream allocation based on different network environment changes, ensuring the stability and continuity of audio decoding even under conditions of significant network fluctuations. Real-time evaluation and feedback mechanisms for decoding quality can promptly identify and repair errors or data loss during the decoding process, optimizing the decoding process and improving the quality of decoded data, thereby enhancing the user experience. The utilization of multi-core processors enables parallel processing of resources during decoding, allowing multiple sub-bitstream blocks to be decoded simultaneously, improving decoding speed and reducing decoding latency. Dynamic adjustment of processor resources and bitstream allocation strategies allows for flexible adjustment of resource allocation based on real-time system load and bandwidth changes, ensuring the efficiency and stability of the decoding process and avoiding resource waste or overload. Decoding smoothness prediction can predict playback smoothness under different network fluctuation conditions and provide early warnings of potential stuttering or decoding latency issues. This mechanism provides users with advance network status alerts, helping them adjust playback strategies or take appropriate measures. Attached Figure Description

[0014] Figure 1 This is a diagram of the method described in this invention. Detailed Implementation

[0015] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0016] One embodiment of the present invention, such as Figure 1 As shown, a method for improving decoding speed based on internet radio and podcast streams includes:

[0017] S1: Perform initial parsing of the bitstreams of internet radio and podcasts to obtain basic bitstream information, including bitrate, encoding format, and initial network bandwidth data; based on the basic bitstream information, construct a bitstream feature model, which is used to describe the transmission characteristics of the bitstream under different network environments;

[0018] S2: Based on the bitstream feature model, an adaptive bitstream allocation strategy is implemented. The adaptive bitstream allocation strategy dynamically adjusts the transmission priority and allocation ratio of each data packet in the bitstream according to the real-time monitored network bandwidth changes, and generates an adaptive bitstream allocation scheme. At the same time, the multi-core processor resources are evaluated to determine the number of cores and processing capabilities available for parallel decoding, and a parallel decoding resource model is constructed.

[0019] S3: According to the adaptive bitstream allocation scheme, the bitstream data is divided into multiple sub-bitstream blocks, each of which contains a portion of the bitstream data and its corresponding decoding instructions; using the parallel decoding resource model, each sub-bitstream block is allocated to different cores of a multi-core processor to perform parallel decoding processing, resulting in multiple parallel decoding results; the multiple parallel decoding results are synchronously integrated to generate preliminary decoded data;

[0020] S4: Perform quality assessment on the initial decoded data, identify potential decoding errors or data loss, and generate decoding quality feedback data; based on the decoding quality feedback data, fine-tune the adaptive bitstream allocation scheme and optimize the bitstream allocation strategy to further improve decoding accuracy; at the same time, dynamically adjust resource allocation during parallel decoding.

[0021] S5: Based on the optimized adaptive bitstream allocation scheme and parallel decoding resource model, perform final decoding processing to generate high-quality decoded bitstream data; evaluate the smoothness of the high-quality decoded bitstream data, predict the listening smoothness under different network fluctuation conditions, and generate decoding smoothness prediction data;

[0022] S6: Based on decoding smoothness prediction data, implement a risk warning mechanism. When potential stuttering or decoding delay is predicted, adjust the decoding strategy in advance or alert the user to the network status, and generate decoding warning data for internet radio and podcasts.

[0023] The working principle and effects of the above technical solution are as follows:

[0024] By constructing a bitstream feature model to adapt to different network transmission characteristics, and combining it with an adaptive bitstream allocation strategy to dynamically adjust the priority of data packets, the interruption of bitstream transmission caused by network bandwidth fluctuations is reduced, the problem that the fixed allocation mode cannot cope with sudden bandwidth changes is avoided, and the adaptability of bitstream transmission to the network environment is improved.

[0025] By evaluating multi-core processor resources to build a parallel decoding model, sub-code stream blocks are reasonably allocated to different cores for synchronous processing, reducing decoding latency caused by single-core overload, avoiding efficiency waste caused by idle core resources, and improving the overall speed of decoding processing.

[0026] By identifying decoding errors and data loss through quality assessment and providing feedback for adjustments, and then dynamically optimizing bitstream allocation and core load, the impact of decoding anomalies on audio quality is reduced, and resource imbalances that continuously slow down processing efficiency are avoided, thereby enhancing the accuracy of decoding results and the balance of resource utilization.

[0027] By predicting listening status under different networks through smoothness assessment and combining risk warnings to adjust strategies or prompt users in advance, the problem of sudden interruptions to listening due to stuttering and delays is reduced, and users are prevented from passively experiencing a decline in sound quality, thereby enhancing the stability of the listening process and the continuity of the user experience.

[0028] In one embodiment of the present invention, S1 includes:

[0029] S11: Obtain the raw bitstream data of internet radio and podcasts, decompose the raw bitstream at the frame level using a bitstream parsing tool, extract the header identifier and data payload of each frame, initially filter out invalid frames (such as empty frames and duplicate frames), and obtain the valid bitstream to be parsed.

[0030] S12: Perform basic information parsing on the valid bitstream, obtain the bitrate (e.g., 128kbps, 320kbps) and encoding format (e.g., MP3, AAC, FLAC) by parsing the frame header identifier, and call the network monitoring module to collect initial network bandwidth data in real time (including bandwidth peak, average and fluctuation frequency), and integrate them to form the original bitstream basic information set.

[0031] S13: Perform data preprocessing on the original bitstream basic information set. The preprocessing includes cleaning abnormal bandwidth data (such as zero values ​​caused by momentary network outages), unifying the encoding format description standard (such as standardizing Advanced Audio Coding to AAC), associating the correspondence between bitrate and bandwidth, and generating a standardized bitstream basic information set.

[0032] S14: Based on the standardized bitstream basic information set, introduce the transmission records of similar bitstreams in the historical transmission database (such as packet loss rate and latency value of the same encoding format under different bandwidths), train the model parameters through machine learning algorithms (such as random forest), and construct a bitstream feature model that can describe the transmission characteristics (packet loss pattern, latency change trend) of bitstreams in different network environments.

[0033] The working principle and effects of the above technical solution are as follows:

[0034] By decomposing the original bitstream at the frame level and removing empty and duplicate frames, the interference of invalid data on subsequent processing is reduced, redundant information is avoided from occupying parsing resources, and the purity of the bitstream data is improved.

[0035] The code analyzes the bitrate and encoding format of the effective bitstream, and simultaneously collects data such as peak and average bandwidth. It integrates these data to form the original information set, reducing the analysis bias caused by missing basic information, avoiding the impact of missing key parameters on subsequent model construction, and enhancing the integrity of the bitstream's basic data.

[0036] Cleaning abnormal bandwidth data, standardizing encoding format descriptions, and associating bitrate with bandwidth reduces errors caused by chaotic data formats or outliers, avoids information misalignment caused by inconsistent standards, and improves the standardization and usability of the basic information set.

[0037] By combining historical transmission records with machine learning to train a bitstream feature model, the limitations of relying solely on current data to build the model are reduced, the problem of the model being unable to adapt to complex network environments is avoided, and the accuracy of predicting bitstream transmission characteristics (such as packet loss and latency) is enhanced.

[0038] In one embodiment of the present invention, S2 includes:

[0039] S21: Deploy a real-time network bandwidth monitoring module, using a combination of periodic sampling (e.g., sampling once every 500ms) and event-triggered sampling (e.g., additional sampling when bandwidth fluctuations exceed 10%) to obtain real-time network bandwidth change data and generate a dynamic bandwidth change curve;

[0040] S22: Input the bandwidth dynamic change curve into the bitstream feature model, and output bitstream transmission optimization suggestions corresponding to different bandwidth ranges based on the bitstream feature model (e.g., prioritize the transmission of high-definition bitstream when the bandwidth is ≥2Mbps). Formulate an adaptive bitstream allocation strategy based on the optimization suggestions, clarify the transmission priority rules of each data packet (e.g., audio key frames have higher priority than redundant frames) and allocation ratio (e.g., high-definition data packets account for 60% and standard-definition data packets account for 40%), and generate a preliminary adaptive bitstream allocation scheme.

[0041] S23: Launch the multi-core processor resource assessment tool to detect the number of physical cores, logical cores, real-time load rate of each core (such as CPU utilization) and single-core decoding processing capability (such as the number of MP3 stream frames that can be decoded per second), record the data interaction latency between cores, and generate a multi-core processor resource assessment report.

[0042] S24: Based on the multi-core processor resource assessment report, high-load cores with a load rate exceeding 80% are removed, and effective cores that can be used for parallel decoding are selected. Combining core processing capabilities and data interaction latency, core collaboration weights are set (e.g., cores with strong processing capabilities and low interaction latency have higher weights), and a parallel decoding resource model including the number of cores, processing capabilities, and collaboration rules is constructed.

[0043] The working principle and effects of the above technical solution are as follows:

[0044] By combining periodic sampling with event-triggered sampling to monitor bandwidth, the situation of missing sudden bandwidth changes in a single sampling mode is reduced, and the subsequent bitstream adjustment is not timely due to sampling lag, thus improving the real-time performance and accuracy of bandwidth data acquisition.

[0045] By inputting the bandwidth dynamic curve into the bitstream feature model to generate optimization suggestions, and then formulating allocation strategies and schemes accordingly, the problem of fixed bitstream allocation being unable to adapt to bandwidth fluctuations is reduced, high-definition bitstreams are prevented from frequently stuttering under low bandwidth, the flexibility of bitstream allocation is improved, and the adaptability to network conditions is enhanced.

[0046] The system detects the number of processor cores, load rate, and single-core decoding capability, and records the interaction latency to generate a report. This reduces the situation of incomplete understanding of multi-core resources, avoids blind allocation during subsequent core allocation, and improves the completeness of processor resource assessment.

[0047] By eliminating high-load cores, selecting effective cores, and setting collaboration weights based on processing capacity and latency, decoding latency caused by over-allocation of tasks to high-load cores is reduced, idle core resources are avoided, the utilization efficiency of multi-core processors is improved, and the stability of parallel decoding is enhanced.

[0048] In one embodiment of the present invention, S3 includes:

[0049] S31: Based on the allocation ratio and priority rules in the preliminary adaptive bitstream allocation scheme, set the sub-bitstream block splitting standard (such as each sub-bitstream block contains 10 frames of continuous data and carries the corresponding decoding priority identifier), and divide the original bitstream into multiple independent sub-bitstream blocks through the bitstream splitting algorithm. Each sub-bitstream block is attached with decoding instructions (such as decoding algorithm type and frame order dependency).

[0050] S32: Match the decoding complexity of sub-stream blocks (e.g., high-rate sub-stream blocks are more complex than low-rate sub-stream blocks) with the core processing capabilities in the parallel decoding resource model, formulate sub-stream block allocation rules (e.g., allocate high-complexity sub-stream blocks to cores with strong processing capabilities), and generate a sub-stream block-core allocation mapping table;

[0051] S33: Based on the allocation mapping table, each sub-stream block is distributed to the corresponding multi-core processor core through an inter-processor communication protocol (such as PCIe). Each core calls a preset decoding algorithm (such as MP3 decoding algorithm, AAC decoding algorithm) to synchronously execute parallel decoding processing and generate an independent decoding result for a single core.

[0052] S34: Build a decoding result synchronization and integration module, receive the independent decoding results of each core, adjust the integration order according to the frame order dependency of the sub-code stream blocks, correct the frame order deviation caused by parallel decoding, and fill the small data gaps caused by the data interaction delay between cores to generate continuous and complete preliminary decoding data.

[0053] The working principle and effects of the above technical solution are as follows:

[0054] By setting sub-stream block splitting criteria according to allocation ratio and priority, and adding decoding instructions to split the bitstream, the arbitrariness of bitstream splitting is reduced, and subsequent decoding chaos caused by uneven sub-stream block sizes or missing instructions is avoided, thereby improving the standardization and operability of bitstream splitting.

[0055] By matching the complexity of sub-stream blocks with the core processing capabilities and formulating allocation rules to generate a mapping table, decoding stuttering caused by assigning high-complexity tasks to weak cores is reduced, core load imbalance is avoided, the rationality of sub-stream block allocation is improved, and the adaptability of multi-core resources is enhanced.

[0056] Based on the mapping table, sub-bit stream blocks are distributed through a high-speed protocol. Each core decodes synchronously and in parallel, reducing the time spent by a single core processing the entire bit stream, avoiding the problem of low decoding efficiency, improving the overall decoding speed, and enhancing the parallelism of processing.

[0057] By building a synchronous integration module to adjust the frame order and fill data gaps, the frame order disorder or data breakage caused by parallel decoding is reduced, the situation of discontinuous data in the initial decoding is avoided, and the integrity and smoothness of the decoding results are improved.

[0058] In one embodiment of the present invention, step S31 includes:

[0059] Extract the allocation ratio parameters (e.g., 60% HD bitstream and 40% SD bitstream) and priority rules (e.g., set the priority value of audio keyframes to 1 and redundant frames to 5) from the preliminary adaptive bitstream allocation scheme, quantize the rules (use numerical levels from 1 to 5 to represent priority), and generate a standardized bitstream allocation rule set.

[0060] Based on the standardized bitstream allocation rule set, and combined with the frame length and frame type correlation of the bitstream (e.g., three consecutive frames form a complete audio segment), sub-bitstream block splitting parameters are set: the number of consecutive frames contained in a single block (e.g., 10±2 frames for dynamic adaptation), the embedding position of the priority identifier (the second byte in the sub-bitstream block header), and the data boundary check character (e.g., FFEE as the block end identifier), thus forming the sub-bitstream block splitting standard.

[0061] The dynamic frame segmentation algorithm is invoked to scan and segment the original bitstream frame by frame according to the segmentation criteria, ensuring that the number of frames in each sub-bitstream block is within the set range and carries a priority identifier. Defective blocks generated by the segmentation (such as fragments with less than 5 frames) are filtered out, and an initial set of sub-bitstream blocks is generated.

[0062] For each sub-stream block in the initial sub-stream block set, its encoding format (such as MP3, AAC) and frame sequence dependency relationship (such as the 8th frame needing to be decoded based on the 7th frame) are parsed, and corresponding decoding instructions (including the adapted decoding algorithm, frame dependency chain, and maximum decoding duration) are generated. The instructions are written into the extended field of the sub-stream block to form an independent sub-stream block with decoding instructions.

[0063] The working principle and effects of the above technical solution are as follows:

[0064] By extracting and quantifying the allocation ratio parameters and priority rules, the problems of ambiguous priority determination and inconsistent ratio expression are reduced, and deviations caused by rule confusion during subsequent splitting are avoided, thereby improving the standardization and executability of the bitstream allocation rule set.

[0065] Based on standardized rules, and combined with frame length and correlation, the splitting parameters (number of frames, identifier position, check character) are set, which reduces the uneven size and blurred boundaries of sub-stream blocks, avoids the problem that the split data blocks cannot be adapted to subsequent decoding, and enhances the rationality and pertinence of the sub-stream block splitting standard.

[0066] By calling the dynamic frame segmentation algorithm and filtering out defective blocks, the situation where fragmented blocks with insufficient frames occupy decoding resources is reduced, decoding errors or interruptions caused by defective blocks are avoided, and the purity and availability of the initial sub-stream block set are improved.

[0067] The encoding format and dependencies of each sub-stream block are parsed and decoding instructions are attached, which reduces the problems of choosing the wrong algorithm and ignoring frame dependencies during decoding, avoids decoding failure or disorder due to missing instructions, and enhances the accuracy and smoothness of independent decoding of sub-stream blocks.

[0068] In one embodiment of the present invention, S32 includes:

[0069] Extract the decoding instructions (such as encoding format and frame dependency relationship) and bit rate information of each sub-bit stream block, and quantify the decoding complexity into 1-5 levels (level 1 is the lowest and level 5 is the highest) through a complexity evaluation algorithm (combining the single frame decoding computation and frame dependency chain length), and generate a sub-bit stream block complexity level table;

[0070] Extract the effective core processing capability parameters (such as the number of frames decoded per second per core and the current idle computing power ratio) from the parallel decoding resource model, sort them from high to low processing capability and label them as AE category (A category is the strongest and E category is the weakest) to form a core capability classification table;

[0071] Based on the complexity level table and the core capability classification table, the matching rules are formulated as follows: Level 5 complexity sub-stream blocks are matched with Class A cores, Level 4 are matched with Class A / B cores, Level 3 are matched with Class B / C cores, Level 2 are matched with Class C / D cores, and Level 1 is matched with Class D / E cores. At the same time, the maximum number of sub-stream blocks that can be allocated to a single core is limited (e.g., no more than 8 Class A cores) to avoid overload.

[0072] Based on the matching rules, each sub-stream block is assigned to a core of the corresponding type according to its complexity level. The load balancing algorithm is used for fine-tuning (e.g., transferring the 8th level 5 block of a certain type A core to a type A core with a lower load). Finally, a sub-stream block-core allocation mapping table containing sub-stream block ID, corresponding core number, and allocation priority is generated.

[0073] The working principle and effects of the above technical solution are as follows:

[0074] By extracting sub-stream block information and quantizing the decoding complexity into levels 1-5, the ambiguity of relying on experience to judge complexity in the past is reduced, the problem of core matching misalignment caused by unclear complexity definition is avoided, and the accuracy of decoding complexity assessment is improved.

[0075] Extracting core parameters from the parallel decoding resource model and classifying them by capability reduces the incomplete understanding of each core's processing capabilities, avoids blind allocation of strong cores to be idle and weak cores to be overloaded, and improves the clarity and availability of core capability classification.

[0076] By formulating matching rules based on complexity and core capabilities and limiting the maximum number of tasks per core, the decoding stutter caused by allocating high-complexity sub-stream blocks to weak cores is reduced, and the overload protection of cores due to too many tasks is avoided, thereby enhancing the rationality of task allocation and the stability of core operation.

[0077] By fine-tuning the allocation and generating a mapping table through a load balancing algorithm, the situation of excessive load differences between cores of the same type is reduced, avoiding resource waste of some cores being idle and others being fully loaded. This improves the overall resource utilization efficiency of the multi-core processor and lays the foundation for the smoothness of subsequent parallel decoding.

[0078] In one embodiment of the present invention, S33 includes:

[0079] Extract key information from the sub-stream block-core allocation mapping table, including the ID of each sub-stream block, target core number, data size and transmission priority. At the same time, call the multi-core processor status monitoring module to verify the current load rate of the target core (≤75%) and the availability of the communication link in real time, remove core allocation records with temporary overload or link abnormality, and generate an executable sub-stream block distribution list.

[0080] Based on the distribution list, a high-speed inter-processor communication protocol (such as PCIe 4.0) is selected, and the sub-stream block data is transmitted to the target core in descending order of transmission priority. After each sub-stream block is transmitted, the receiving core sends a feedback data reception confirmation signal and generates a sub-stream block transmission status log (including transmission duration and whether retransmission is required).

[0081] After each target core receives the sub-bit stream block, it automatically parses the decoding instructions in its extended fields (such as adapting to the AAC decoding algorithm and following the frame dependency chain order), calls the corresponding local preset decoding algorithm module, starts the synchronous decoding process, records the frame decoding time and data verification value in real time during the decoding process, and generates the single-core decoding intermediate result.

[0082] After each core completes decoding, the intermediate decoding results are compared with the preset verification value using a data integrity verification algorithm (such as MD5 checksum). Once it is confirmed that there are no decoding errors (such as frame loss or data distortion), the core number and decoding completion timestamp are appended to the result, generating a single core independent decoding result with traceability information.

[0083] The working principle and effects of the above technical solution are as follows:

[0084] By extracting mapping table information and verifying core load and link availability, abnormal allocation records are removed to generate a distribution list, which reduces invalid operations of distributing tasks to overloaded or link-failed cores, avoids transmission interruption or task backlog, and improves the executability of the distribution list.

[0085] High-speed communication protocols are used for priority transmission, coupled with a receive confirmation mechanism and log generation, which reduces transmission delay and data packet loss, avoids decoding lag caused by transmission problems, and improves the efficiency and reliability of sub-stream block transmission.

[0086] The core automatically parses and decodes the instructions and calls the corresponding algorithm to decode synchronously, recording the time consumption and verification value in real time. This reduces the problem of selecting the wrong decoding algorithm or the decoding process being chaotic, avoids single-core decoding operation errors, and enhances the smoothness of parallel decoding.

[0087] After decoding, the results are verified by MD5 and appended with a core number and timestamp, which reduces residual errors such as frame loss and data distortion, avoids the trouble of tracing problematic results, and improves the accuracy and traceability of independent decoding results.

[0088] In one embodiment of the present invention, step S4 includes:

[0089] S41: Construct a decoding quality evaluation index system, which includes bit error rate (the difference rate between the decoded audio data and the original data), data integrity (the proportion of frames without loss to the total number of frames), and audio distortion (detected by comparing audio waveforms), and define the qualified thresholds for each index (such as bit error rate ≤ 0.5% and data integrity ≥ 99%).

[0090] S42: Input the preliminary decoded data into the quality assessment module, check each item according to the assessment index system, record the location of abnormal data that does not reach the qualified threshold (such as the bit error rate exceeding the standard in frames 120-150) and the type of abnormality (such as data loss caused by packet loss, bit error caused by decoding algorithm deviation), and generate a decoding quality assessment report.

[0091] S43: Extract decoding quality feedback data (such as the bitstream type corresponding to abnormal data and the bandwidth value during transmission) based on the decoding quality assessment report, input the feedback data into the preliminary adaptive bitstream allocation scheme, fine-tune the transmission priority (such as increasing the priority of bitstream types with high abnormality rates by 10%) and allocation ratio (such as increasing the proportion of standard definition data packets to 50%), and generate an optimized adaptive bitstream allocation scheme.

[0092] S44: Real-time monitoring of load changes in each core during parallel decoding (e.g., the load rate of a certain core rises to 85%), combined with decoding quality feedback data (e.g., the abnormal rate of the sub-stream blocks processed by this core is high), and using resource scheduling algorithms to transfer some sub-stream blocks of the high-load core to the low-load core, dynamically adjusting core resource allocation and updating the parallel decoding resource model.

[0093] The working principle and effects of the above technical solution are as follows:

[0094] By constructing an evaluation system that includes indicators such as bit error rate and integrity, and by clarifying the qualified threshold, the problems of vague indicators and inconsistent standards in quality evaluation are reduced, the bias of judging decoding quality based on experience is avoided, and the pertinence and objectivity of the evaluation are improved.

[0095] By detecting and recording the location and type of anomalies in the initial decoded data item by item and generating reports, the number of cases where anomalies are hidden and not discovered is reduced, and the blindness of subsequent optimization due to a lack of accurate basis is avoided, thereby enhancing the efficiency and accuracy of problem investigation.

[0096] Based on the feedback data extracted from the evaluation report, the priority and allocation ratio of the bitstream transmission were fine-tuned, which reduced decoding anomalies caused by insufficient adaptability of the original allocation scheme, avoided the continuous transmission of high anomaly rate bitstreams affecting quality, and improved the rationality and adaptability of the bitstream allocation scheme.

[0097] By monitoring core load in real time and dynamically transferring tasks and updating resource models based on the anomaly rate, decoding errors caused by overload on high-load cores are reduced, the waste of core resources due to both idleness and overload is avoided, the balance of multi-core resource utilization is improved, and the stability of parallel decoding is enhanced.

[0098] In one embodiment of the present invention, S44 includes:

[0099] The core load real-time monitoring module collects key load indicators (including CPU utilization, memory usage, and sub-stream block decoding progress) of each core of the multi-core processor at a sampling interval of 100ms. At the same time, it records the sub-stream block ID currently being processed by each core and generates a core real-time load data table with timestamps.

[0100] Extract the anomaly rate of each sub-stream block from the decoding quality feedback data (e.g., the anomaly rate of a sub-stream block processed by a certain core reaches 3%), associate it with the core real-time load data table by sub-stream block ID, filter out high-load and high-anomaly cores with load rate ≥ 80% and anomaly rate ≥ 2%, and clarify the number of the core to be adjusted and the number of sub-stream blocks currently being processed.

[0101] Invoke dynamic resource scheduling algorithms (such as minimum load priority algorithm), analyze the idle computing power of low-load cores (load rate ≤ 50%), determine the target core for each type of sub-stream block to be transferred (such as high complexity and low complexity), formulate a sub-stream block transfer plan (including transfer order and data transmission link), execute the transfer operation, and generate a sub-stream block transfer completion log;

[0102] Based on the sub-stream block transfer completion log, update the load metrics and processed sub-stream block information of each core, and at the same time correct the parameters in the parallel decoding resource model, including real-time processing capability and load threshold (e.g., lower the load threshold of the original high-load core to 75%), and generate the updated parallel decoding resource model.

[0103] The working principle and effects of the above technical solution are as follows:

[0104] By collecting load metrics such as core CPU utilization and memory usage at 100ms intervals, and recording sub-stream block IDs to generate a timestamped load table, the problem of load monitoring lag or incomplete data is reduced, avoiding missing the initial adjustment opportunity of core overload, and improving the real-time performance and data integrity of core load monitoring.

[0105] By associating the sub-stream block anomaly rate with the load table, cores with a load rate ≥80% and an anomaly rate ≥2% are selected. This reduces the biased judgment that only considers the load and not the decoding quality, avoids mistakenly including normal cores with high load but low anomaly in the adjustment range, and improves the accuracy of core selection to be adjusted.

[0106] By using a dynamic scheduling algorithm to analyze the idle computing power of low-load cores, formulate a transfer plan and generate logs, the random allocation of sub-stream blocks during transfer is reduced, and decoding interruptions caused by insufficient computing power or disordered transfer order of low-load cores are avoided. This enhances the rationality of resource scheduling and improves the efficiency of transfer operations.

[0107] By updating core information and correcting parameters such as model load thresholds based on the transfer log, the problem of outdated parameters in the parallel decoding resource model is reduced. This avoids the core from being overloaded or malfunctioning again due to subsequent allocations still using the old parameters, thus improving the timeliness of the resource model and enhancing the stability of subsequent parallel decoding.

[0108] In one embodiment of the present invention, step S5 includes:

[0109] S51: Based on the optimized adaptive bitstream allocation scheme, the original bitstream is re-splittered into sub-bitstream blocks to ensure that the split sub-bitstream blocks meet the adjusted priority and allocation ratio. At the same time, the sub-bitstream block-core allocation mapping table is re-formulated according to the updated parallel decoding resource model.

[0110] S52: Distribute the sub-stream blocks to the multi-core processor cores according to the new allocation mapping table. Each core uses an optimized decoding algorithm (such as a low-latency decoding algorithm for AAC encoding) to perform the final parallel decoding process and output the final decoding result of each core.

[0111] S53: Performs secondary synchronization and integration on the final decoding results of each core, verifies data integrity through audio data verification algorithms (such as CRC check), corrects possible minor deviations, and generates high-quality decoded bitstream data that is free of anomalies and flows smoothly.

[0112] S54: Set decoding smoothness evaluation indicators, including buffer time (waiting time from decoding to playback), stuttering frequency (number of stutters per minute), playback latency (time difference between decoding completion and audio output), and build a network fluctuation simulation library (including scenarios such as weak network, network outage recovery, and sudden bandwidth drop).

[0113] S55: Input high-quality decoded bitstream data into the smoothness test module, conduct playback tests in different scenarios of the network fluctuation simulation library, record the smoothness index values ​​in each scenario, analyze the index change trend through time series prediction algorithms (such as ARIMA model), and generate decoding smoothness prediction data (such as the stuttering frequency will rise to 2 times / minute when the bandwidth drops by 30%).

[0114] The working principle and effects of the above technical solution are as follows:

[0115] Based on the optimized scheme, the bitstream was split and a mapping table was created, which reduced the problem of splitting not conforming to priority or allocation ratio, avoided the hidden danger of imbalance in core allocation, and improved the accuracy of sub-bitstream block splitting and core matching.

[0116] Distributing data according to the new mapping table and using an optimized decoding algorithm for parallel decoding reduces decoding latency caused by algorithm inefficiency, avoids uneven core processing efficiency, improves the final decoding speed, and enhances the stability of parallel processing.

[0117] During the secondary synchronization integration, data verification is performed and minor deviations are corrected, reducing the frame gaps or data distortions left over from parallel decoding. This avoids noise and disconnections during stream playback, and improves the integrity and auditory smoothness of the high-quality decoded stream.

[0118] By setting smoothness indicators such as buffer time and stuttering frequency, and building simulation libraries for weak networks and sudden bandwidth drops, the one-sidedness of smoothness evaluation with a single dimension or lack of support from real scenarios is reduced, and the comprehensiveness and pertinence of the evaluation are enhanced.

[0119] After testing, time series algorithms were used to analyze trends and generate predictive data, which reduced the occurrence of sudden stuttering caused by network fluctuations, prevented users from passively experiencing listening interruptions, improved the ability to predict smoothness risks, and laid the foundation for subsequent early warnings.

[0120] In one embodiment of the present invention, S53 includes:

[0121] The multi-core result collection module is started, and the core number in the sub-stream block-core allocation mapping table is used to receive the final decoding results of each core in batches. At the same time, the core processing time and data volume corresponding to each result are collected to generate an association list containing core number-decoding result-processing time.

[0122] Extract the frame sequence identifier (such as frame sequence number) of each decoding result in the associated list, use a fast sorting algorithm to sort all results in ascending order of frame sequence number, remove duplicate frames (such as duplicate data mistransmitted by multiple cores), and generate an ordered sequence of decoding results;

[0123] Parallel data verification is initiated for the ordered decoding result sequence: the sequence is divided into multiple segments and allocated to idle processing threads (utilizing idle computing power to balance the overall resource load). Each thread synchronously performs CRC verification to check data integrity; at the same time, the audio parameters (such as sampling rate and number of channels) of adjacent frames are compared, abnormal frames with inconsistent parameters are marked, and a verification report (including the proportion of normal frames and the position of abnormal frames) is generated.

[0124] For abnormal frames in the verification report, if the error is due to parameter deviation (such as single-frame sampling rate offset), a frame parameter correction algorithm (such as interpolation compensation based on adjacent frames) is called to make adjustments; if the error is due to missing data (such as a frame not being fully decoded), a complete frame is generated based on the data of the frames before and after the frame (to ensure audio continuity), and a corrected complete decoding sequence is generated.

[0125] The corrected complete decoding sequence is subjected to inter-frame smoothing processing. Minor stutters at frame boundaries (such as inter-frame volume difference correction) are eliminated through audio transition algorithms. Then, it is encapsulated into a continuous data stream according to standard audio formats (such as MP3 and AAC) to generate high-quality decoded bitstream data with no anomalies and low latency.

[0126] The working principle and effects of the above technical solution are as follows:

[0127] The collection module receives the decoding results and records the core number and processing time to generate an associated list, which reduces the trouble of being unable to trace the source of the results later, avoids the problem of core processing efficiency data loss, and improves the convenience of tracing the source of decoding results.

[0128] Extracting frame sequence identifiers and using quick sort to arrange and remove duplicate frames reduces frame order disorder caused by parallel decoding, avoids frame skipping caused by duplicate data occupying playback resources, and improves the efficiency and orderliness of decoding result integration.

[0129] The ordered sequence is segmented into sub-threads for parallel verification. By utilizing idle computing power to balance the load, the overall latency caused by the verification process occupying main resources is reduced, the problem of excessive time consumption for single-thread verification is avoided, and the speed of data integrity detection is improved.

[0130] The system corrects for abnormal frames (parameter deviation interpolation compensation, data missing data generation to complete the frame), reduces audio noise or disconnection caused by abnormal frames, avoids directly discarding frames and affecting the integrity of the bitstream, and enhances the continuity of the decoding sequence.

[0131] By performing inter-frame smoothing on the complete sequence and encapsulating it in a standard format, minor stutters and volume abrupt changes at frame boundaries are reduced, avoiding auditory discomfort and improving the playback smoothness of high-quality decoded streams and the user listening experience.

[0132] In one embodiment of the present invention, step S6 includes:

[0133] S61: Based on the listening experience requirements of internet radio and podcasts, set decoding risk warning thresholds (such as buffer time > 2 seconds, stuttering frequency > 1 time / minute, playback delay > 500ms), and clarify the triggering conditions corresponding to different warning levels (such as yellow warning: single indicator exceeds the standard; red warning: two or more indicators exceed the standard);

[0134] S62: Compare the decoding smoothness prediction data with the warning threshold in real time to determine whether there is a decoding risk in the current and future period (e.g., if the predicted stuttering frequency reaches 3 times / minute in a certain scenario, triggering a red warning), and generate a risk judgment result;

[0135] S63: If a yellow alert is triggered, the decoding strategy fine-tuning mechanism is activated, such as caching 30 seconds of high-quality decoded bitstream data in advance, reducing bitstream resolution to reduce bandwidth usage, and then re-predicting smoothness to verify the effect of the adjustment.

[0136] S64: If a red alert or yellow alert is triggered and the adjustment is ineffective, the user prompt mechanism will be activated to inform the user of the current network status (such as low network bandwidth, which may affect the smoothness of listening) through device pop-ups, audio broadcasts, etc., and at the same time provide optional operation suggestions (such as whether to switch to standard definition audio quality).

[0137] S65: Record the trigger time, warning level, handling measures and handling results of risk warnings (such as reducing the stuttering frequency to 0.5 times / minute after adjustment), and integrate them to form network radio and podcast decoding warning data.

[0138] The working principle and effects of the above technical solution are as follows:

[0139] By setting warning thresholds and level triggering conditions based on listening needs, the problems of vague warning standards and unclear level classification in the past have been reduced, avoiding misjudgment or omission of risks, and improving the pertinence and accuracy of risk warnings.

[0140] By comparing fluency prediction data with thresholds in real time to assess risks, we reduce the passive situation of being caught off guard when risks suddenly occur, avoid problems such as stuttering and delays that seriously affect listening, and enhance the ability to predict decoding risks.

[0141] When a yellow alert is issued, the strategy is fine-tuned and its effectiveness is verified, which reduces the probability of small risks escalating into big problems, avoids the tedious process of triggering user prompts for minor fluctuations, and improves the efficiency of risk handling.

[0142] When a red alert or fine-tuning proves ineffective, the system notifies the user and provides operational suggestions, reducing the user's confusion about unexplained listening interruptions, preventing the user from passively experiencing a decline in sound quality, and enhancing the user's proactive experience.

[0143] Recording data throughout the entire early warning process reduces the lack of reference for handling similar risks in the future, provides practical basis for optimizing early warning thresholds and strategies, and improves the ability of the early warning mechanism to iterate and improve.

[0144] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for improving decoding speed based on internet radio and podcast streams, characterized in that, The method includes: S1: Perform initial parsing of the bitstreams of internet radio and podcasts to obtain basic bitstream information, and construct a bitstream feature model based on the basic bitstream information; S2: Based on the bitstream feature model, implement an adaptive bitstream allocation strategy and generate an adaptive bitstream allocation scheme; at the same time, evaluate multi-core processor resources and construct a parallel decoding resource model; S3: According to the adaptive bitstream allocation scheme, the bitstream data is divided into multiple sub-bitstream blocks. Using the parallel decoding resource model, each sub-bitstream block is allocated to different cores of the multi-core processor to perform parallel decoding processing and obtain multiple parallel decoding results. The multiple parallel decoding results are synchronously integrated to generate preliminary decoded data. S4: Perform quality assessment on the initial decoded data and generate decoding quality feedback data; based on the decoding quality feedback data, fine-tune the adaptive bitstream allocation scheme, and dynamically adjust the resource allocation during the parallel decoding process; S5: Based on the optimized adaptive bitstream allocation scheme and parallel decoding resource model, perform final decoding processing to generate high-quality decoded bitstream data; evaluate the smoothness of the high-quality decoded bitstream data, predict the listening smoothness under different network fluctuation conditions, and generate decoding smoothness prediction data; S6: Based on decoding smoothness prediction data, implement a risk warning mechanism. When potential stuttering or decoding delay is predicted, adjust the decoding strategy in advance or alert the user to the network status, and generate decoding warning data for internet radio and podcasts.

2. The method for improving decoding speed based on internet radio and podcast streams according to claim 1, characterized in that, S1 includes: S11: Obtain the raw bitstream data of internet radio and podcasts, decompose the raw bitstream at the frame level using a bitstream parsing tool, extract the header identifier and data payload of each frame, initially filter out invalid frames, and obtain the valid bitstream to be parsed; S12: Perform basic information parsing on the valid bitstream, obtain the bit rate and encoding format by parsing the frame header identifier, and call the network monitoring module to collect the initial network bandwidth data in real time, and integrate them to form the original bitstream basic information set; S13: Perform data preprocessing on the original bitstream basic information set to generate a standardized bitstream basic information set; S14: Based on the standardized bitstream basic information set, introduce the transmission records of similar bitstreams from the historical transmission database, train the model parameters through machine learning algorithms, and construct a bitstream feature model that can describe the transmission characteristics of bitstreams in different network environments.

3. The method for improving decoding speed based on internet radio and podcast streams according to claim 1, characterized in that, The S2 includes: S21: Deploy a real-time network bandwidth monitoring module, using a combination of periodic sampling and event-triggered sampling to obtain real-time network bandwidth change data and generate a dynamic bandwidth change curve; S22: Input the bandwidth dynamic change curve into the bitstream feature model, output bitstream transmission optimization suggestions corresponding to different bandwidth intervals based on the bitstream feature model, formulate an adaptive bitstream allocation strategy based on the optimization suggestions, clarify the transmission priority rules and allocation ratio of each data packet, and generate a preliminary adaptive bitstream allocation scheme. S23: Launch the multi-core processor resource assessment tool to detect the number of physical cores, logical cores, real-time load rate of each core and single-core decoding processing capability of the processor, record the data interaction latency between cores, and generate a multi-core processor resource assessment report. S24: Based on the multi-core processor resource assessment report, high-load cores with a load rate exceeding 80% are eliminated, and effective cores that can be used for parallel decoding are selected. Combining core processing capabilities and data interaction latency, core collaboration weights are set to construct a parallel decoding resource model.

4. The method for improving decoding speed based on internet radio and podcast streams according to claim 1, characterized in that, The S3 includes: S31: Based on the allocation ratio and priority rules in the preliminary adaptive bitstream allocation scheme, set the sub-bitstream block splitting standard, and divide the original bitstream into multiple independent sub-bitstream blocks through the bitstream splitting algorithm, with each sub-bitstream block having additional decoding instructions attached; S32: Match the decoding complexity of the sub-stream block with the core processing capability in the parallel decoding resource model, formulate sub-stream block allocation rules, and generate a sub-stream block-core allocation mapping table; S33: Based on the allocation mapping table, each sub-code stream block is distributed to the corresponding multi-core processor core through the inter-processor communication protocol. Each core calls the preset decoding algorithm to synchronously execute parallel decoding processing and generate an independent decoding result for a single core. S34: Build a decoding result synchronization and integration module, receive the independent decoding results of each core, adjust the integration order according to the frame order dependency of the sub-code stream blocks, and generate continuous and complete preliminary decoding data.

5. The method for improving decoding speed based on internet radio and podcast streams according to claim 4, characterized in that, S32 includes: Extract the decoding instructions and bitrate information of each sub-bitstream block, and quantify the decoding complexity into 1-5 levels using a complexity evaluation algorithm; where level 1 is the lowest and level 5 is the highest, and generate a sub-bitstream block complexity level table; The effective core processing capability parameters are extracted from the parallel decoding resource model, sorted from high to low processing capability and labeled as A, E, and E classes; among them, class A is the strongest and class E is the weakest, forming a core capability classification table. Based on the complexity level table and the core capability classification table, matching rules are formulated: Level 5 complexity sub-stream blocks are matched with Class A cores, Level 4 is matched with Class A / B cores, Level 3 is matched with Class B / C cores, Level 2 is matched with Class C / D cores, and Level 1 is matched with Class D / E cores. At the same time, the maximum number of sub-stream blocks allocated to a single core is limited to avoid overload. Based on the matching rules, each sub-stream block is assigned to a core of the corresponding type according to its complexity level. The assignment is then fine-tuned using a load balancing algorithm to finally generate a sub-stream block-core allocation mapping table containing the sub-stream block ID, the corresponding core number, and the allocation priority.

6. The method for improving decoding speed based on internet radio and podcast streams according to claim 1, characterized in that, The S4 includes: S41: Construct a decoding quality evaluation index system; S42: Input the preliminary decoded data into the quality assessment module, check each item according to the assessment index system, record the location and type of abnormal data that did not reach the qualified threshold, and generate a decoding quality assessment report; S43: Extract decoding quality feedback data based on the decoding quality assessment report, input the feedback data into the preliminary adaptive bitstream allocation scheme, fine-tune the transmission priority and allocation ratio, and generate an optimized adaptive bitstream allocation scheme; S44: Monitors the load changes of each core in real time during parallel decoding, and combines decoding quality feedback data to transfer some sub-code stream blocks of high-load cores to low-load cores through resource scheduling algorithms, dynamically adjusts core resource allocation, and updates the parallel decoding resource model.

7. The method for improving decoding speed based on internet radio and podcast streams according to claim 6, characterized in that, S44 includes: The core load real-time monitoring module collects key load indicators of each core of the multi-core processor at a sampling interval of 100ms, and records the sub-stream block ID currently being processed by each core, generating a core real-time load data table with timestamps. Extract the anomaly rate of each sub-stream block from the decoding quality feedback data, associate it with the core real-time load data table by sub-stream block ID, filter out high-load, high-anomaly cores with a load rate ≥ 80% and anomaly rate ≥ 2%, and clarify the number of the core to be adjusted and the number of sub-stream blocks currently being processed. The system invokes a dynamic resource scheduling algorithm to analyze the idle computing power of low-load cores, determines the target core for each type of sub-stream block to be transferred, formulates a sub-stream block transfer plan, executes the transfer operation, and generates a sub-stream block transfer completion log. Based on the sub-stream block transfer completion log, update the load metrics and processed sub-stream block information of each core, and at the same time correct the parameters in the parallel decoding resource model to generate an updated parallel decoding resource model.

8. The method for improving decoding speed based on internet radio and podcast streams according to claim 1, characterized in that, The S5 includes: S51: Based on the optimized adaptive bitstream allocation scheme, the original bitstream is re-split into sub-bitstream blocks, and the sub-bitstream block-core allocation mapping table is re-formulated according to the updated parallel decoding resource model. S52: Distribute the sub-stream blocks to the multi-core processor cores according to the new allocation mapping table. Each core uses the optimized decoding algorithm to perform the final parallel decoding process and outputs the final decoding result of each core. S53: Performs secondary synchronization and integration on the final decoding results of each core, verifies data integrity through audio data verification algorithm, and generates decoded bitstream data; S54: Set decoding smoothness evaluation metrics and build a network fluctuation simulation library; S55: Input high-quality decoded bitstream data into the smoothness test module, conduct playback tests in different scenarios of the network fluctuation simulation library, record the smoothness index values ​​in each scenario, analyze the index change trend through time series prediction algorithm, and generate decoding smoothness prediction data.

9. The method for improving decoding speed based on internet radio and podcast streams according to claim 8, characterized in that, S53 includes: The multi-core result collection module is started, and the core number in the sub-stream block-core allocation mapping table is used to receive the final decoding results of each core in batches. At the same time, the core processing time and data volume corresponding to each result are collected to generate an association list. Extract the frame sequence identifier of each decoding result in the associated list, use a fast sorting algorithm to sort all results in ascending order of frame sequence number, remove duplicate frames, and generate an ordered sequence of decoding results; Parallel data verification is initiated on the ordered decoding result sequence: the sequence is divided into multiple segments and allocated to idle processing threads. Each thread performs CRC verification synchronously to check data integrity; at the same time, the audio parameters of adjacent frames are compared, abnormal frames with inconsistent parameters are marked, and a verification report is generated. For abnormal frames in the verification report, if the error is due to parameter deviation, the frame parameter correction algorithm is called to make adjustments; if the error is due to missing data, a complete frame is generated based on the data of the frames before and after the frame, and a corrected complete decoding sequence is generated. The corrected complete decoded sequence is subjected to inter-frame smoothing processing, and minor stutters at frame boundaries are eliminated through an audio transition algorithm. Then, it is encapsulated into a continuous data stream according to the standard audio format to generate high-quality decoded bitstream data.

10. The method for improving decoding speed based on internet radio and podcast streams according to claim 1, characterized in that, The S6 includes: S61: Based on the listening experience requirements of internet radio and podcasts, set decoding risk warning thresholds and clarify the triggering conditions corresponding to different warning levels; S62: Compare the decoding fluency prediction data with the warning threshold in real time to determine whether there is a decoding risk in the current and future period and generate a risk judgment result; S63: If a yellow alert is triggered, the decoding strategy fine-tuning mechanism will be activated; S64: If a red alert or yellow alert is triggered and the adjustment is ineffective, activate the user prompt mechanism to inform the user of the current network status and provide optional operation suggestions; S65: Record the trigger time, warning level, handling measures and handling results of risk warnings, and integrate them to form network radio and podcast decoding warning data.

Images