Cloud-based vocal music remote interactive teaching system

Abstract

The present application relates to the technical field of remote interactive transmission, in particular to a vocal music remote interactive teaching system based on cloud, which comprises a network channel monitoring module, an audio uploading control module, a comment signal identification module, a playing buffer adjustment module and a synchronous control module. In the present application, the real-time audio stream and bandwidth capacity information in the uploading channel are collected to dynamically identify the transmission accumulation section, and coincidence screening and rhythm adjustment are performed in combination with the uploading time, so as to effectively alleviate the accumulation imbalance problem in the audio transmission process. The response time and the intensive degree of the teacher's comment signal are used to mark the key teaching section, and the playing speed and the buffer period of the teaching frame are adjusted as needed, so as to realize the high matching of the comment content and the playing rhythm. The synchronous structure of the teaching rhythm is constructed through the unified regulation of the teaching uploading and playing process, the coordination between the audio uploading, comment response and playing is enhanced, and the stability and interactive smoothness of the vocal music teaching audio transmission are improved in the variable network environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote interactive transmission, in particular to a vocal music remote interactive teaching system based on a cloud. BACKGROUND

[0002] The technical field of remote interactive transmission relates to the real-time or non-real-time bidirectional transmission and response of information between multiple terminals separated by geographical locations, mainly including the transmission mechanism of multi-modal data such as audio, video, image, and text, covering communication protocol design, data synchronization mechanism, network transmission optimization, delay control, audio and video coding and decoding processing, and terminal interaction control, etc. technical content, dedicated to ensuring efficient and stable data exchange between remote terminals in a dynamic network environment, widely used in remote medical treatment, online education, virtual meeting, remote command, etc. multiple scenes. Among them, the traditional vocal music remote interactive teaching system refers to a system that realizes real-time audio transmission and interactive guidance between teachers and students for vocal music courses through a communication network, mainly aiming at the lack of face-to-face guidance caused by spatial distance in the teaching process, usually relying on telephone, video call or voice chat room based on social software for vocal music teaching and communication, specifically through the use of a camera for face capture, a microphone for sound signal acquisition, and real-time transmission of audio signals after basic compression and coding through a communication network and played by the opposite end device to realize interaction, but in the vocal music teaching scene, the unstable network, severe audio compression, and high delay often lead to unsmooth teaching interaction and voice distortion, thereby affecting the teaching experience and effect.

[0003] In the traditional vocal music remote interactive teaching process, basic video call or voice chat tools are mainly relied on for communication. When the network environment is unstable, audio transmission is easily subject to bandwidth limitations and data accumulation, and the transmission bottleneck existing in the channel cannot be effectively identified or responded to, resulting in audio delay or frame loss problems in real-time transmission. In addition, there is a lack of rhythm matching and buffer adjustment mechanism between the teacher's comment information and the student's receiving feedback in the teaching process, making it difficult to ensure the clear transmission of comment content and the coherence of listening. The playback process lacks targeted rhythm adjustment strategies, and the playback rhythm cannot be flexibly adjusted according to the intensity of teaching interaction, further exacerbating the problems of audio distortion and interaction delay, limiting the teaching experience and reducing the real-time interaction effect. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art, and a vocal music remote interactive teaching system based on a cloud is proposed.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme, the vocal music remote interactive teaching system based on a cloud comprises:

[0006] The network channel monitoring module collects the real-time audio stream size, the number of audio data packets per unit time, and the maximum bandwidth capacity of the channel in the teacher's upload channel. It calculates the bandwidth utilization ratio sequence of the vocal upload task, evaluates the density of vocal data accumulation within the interval, marks the accumulation section, and generates an audio channel data accumulation section identifier set.

[0007] The audio upload control module filters the time points in the current teaching task where the audio frame upload time coincides with the stacking segment based on the audio channel data stacking segment identifier set, determines whether the upload rhythm falls into the stacking segment, sets the upload frame group delay frame mark according to the stacking segment length, and generates a vocal task upload frame adjustment tag list.

[0008] The comment signal recognition module listens to the teacher comment signal frames received in the student terminal, extracts the text content and audio content response timestamps of each comment, filters out segments with a time interval of less than a set interval threshold, extracts the time index and marks them as dense comment intervals, and generates a time index table of dense vocal comment segments.

[0009] Based on the vocal review dense segment time index table, the playback buffer adjustment module compares the overlap between the current teaching audio segment playback rate and the review dense interval, marks the playback buffer trigger point, updates the playback rhythm adjustment ratio and expands the buffer period range, and generates a vocal playback rhythm adjustment buffer setting table.

[0010] The synchronization control module establishes a set of instructions corresponding to the teaching segment based on the list of adjustment tags for the uploaded frames of the vocal task and the setting table of the vocal playback rhythm adjustment buffer. It then synchronously binds the uploaded adjustment tags with the playback buffer settings to generate a cloud-based remote vocal teaching rhythm control synchronization record.

[0011] As a further aspect of the present invention, the accumulation segment specifically refers to the time period in the bandwidth utilization ratio sequence where the data accumulation frequency is greater than the standard frequency threshold.

[0012] As a further embodiment of the present invention, the audio channel data accumulation segment identifier set includes accumulation time period identifier, accumulation frequency label, and accumulation density level; the vocal task upload frame adjustment label list includes frame group delay label, upload time adjustment label, and scheduling table update record; the vocal commentary dense segment time index table includes commentary segment index, commentary response density, and continuous commentary interval information; the vocal playback rhythm adjustment buffer setting table includes playback rate adjustment factor, buffer period parameter, and buffer trigger time point; and the cloud-based remote vocal teaching rhythm control synchronization record includes rhythm synchronization label, playback upload binding command, and teaching segment control identifier.

[0013] As a further aspect of the present invention, the network channel monitoring module includes:

[0014] The data stream receiving submodule collects real-time audio stream data from the teacher's upload channel, records the total number of bytes transmitted per second and the corresponding timestamp information, and simultaneously calculates the number of audio data packets and the maximum bandwidth capacity of the channel per unit time. Combining the data stream timestamp and transmission length, it calculates the data accumulation value within a unit time period and generates a sequence of audio data accumulation per unit time.

[0015] The bandwidth utilization calculation submodule reads the maximum bandwidth capacity value of the corresponding channel based on the accumulation amount value of each time period in the audio data accumulation amount sequence per unit time, calculates the bandwidth utilization rate value in each time period, and the bandwidth utilization rate is the ratio of the data accumulation amount per unit time to the bandwidth capacity, generating a bandwidth utilization rate ratio sequence.

[0016] Based on the ratio changes in the bandwidth utilization ratio sequence, the stacking segment identification submodule sets a frequency threshold as the stacking frequency judgment threshold, counts the number of times the ratio exceeds the stacking frequency judgment threshold in each time period, and divides the stacking density level according to the ratio size during the time period when the ratio frequency continuously exceeds the stacking frequency judgment threshold, establishes the marking information of the corresponding level in the interval, and generates an audio channel data stacking segment identification set.

[0017] As a further aspect of the present invention, the audio upload control module includes:

[0018] The upload time filtering submodule obtains the audio channel data accumulation segment identifier set, extracts the upload audio frame timestamp sequence of each vocal teaching content in the current teaching task, extracts each segment time and matches each time point in the timestamp sequence item by item, uses time intersection judgment logic to determine whether each audio frame time point is within any segment time, marks the time point falling into the accumulation segment, generates an equal-length Boolean vector as the overlap identifier sequence according to the marked state, and establishes an audio frame accumulation overlap state vector.

[0019] The frame group rhythm adjustment submodule extracts the segment length and number of overlapping frames of the corresponding frame group based on the overlapping frame points marked in the audio frame stacking state vector, counts the timestamp set of all marked frames in the frame group, calculates the average timestamp, uses the standard deviation as the frame time fluctuation index, jointly models the stacking segment length and frame time dispersion, calculates the frame group delay time adjustment value, inserts the delay adjustment amount according to the original scheduling order of each frame group, reconstructs the frame scheduling interval structure, and generates the upload frame rhythm delay amount matrix.

[0020] The upload tag generation submodule constructs upload frame adjustment tags based on the delay adjustment value corresponding to each frame group in the upload frame rhythm delay matrix, generates triplet structure tags according to the frame group index, and appends all frame group tags to the upload scheduling control table in a time series manner to establish a vocal task upload frame adjustment tag list.

[0021] As a further aspect of the present invention, the formula for calculating the frame group delay time adjustment value is as follows:

[0022] ;

[0023] in, This represents the upload latency adjustment value for the k-th frame group. Indicates the duration of the accumulation section. This represents the original upload timestamp of the i-th overlapping frame. This represents the average timestamps of all overlapping frames within the stacked segment. This indicates the number of overlapping frames within the stacked segment.

[0024] As a further aspect of the present invention, the comment signal recognition module includes:

[0025] The signal frame listening submodule listens to all teacher comment signal frames received in the student terminal, extracts the response timestamps of the text and audio content in each comment frame, and assembles the timestamps into a comment timeline sequence according to the frame order, constructs a comment time data table for the corresponding teaching segment, and generates a comment signal time sequence dataset.

[0026] The dense review filtering submodule, based on the review signal time series dataset, groups the review frame sequence according to the teaching segment, calculates the time interval value sequence between consecutive review frames in each segment, filters consecutive time interval segments that are less than the interval threshold, outputs the segment start and end time index sequence, and generates a set of consecutive review frame time intervals.

[0027] The teaching segment annotation submodule extracts the corresponding teaching segment number for each time index segment in the continuous comment frame time interval set, performs dense comment marking in the teaching content structure, records the correspondence between the segment index and the dense label, and establishes a vocal comment dense segment time index table.

[0028] As a further aspect of the present invention, the playback buffer adjustment module includes:

[0029] The playback rate comparison submodule extracts the playback start timestamp and playback rate parameter value of the current frame segment of the teaching audio based on the vocal review dense segment time index table, calculates the time overlap length between the playback frame time period and the review dense interval, and determines whether the playback start point is earlier than the average response interval time of the review response frames in the dense interval. If the judgment condition is met, the current frame segment is marked as the playback buffer trigger frame segment, and a playback buffer trigger frame index set is generated.

[0030] The rhythm adjustment parameter calculation submodule extracts the playback rate value of the corresponding frame and the average response interval time of the comment frame based on the frame segment index information in the playback buffer trigger frame index set, calculates the rate difference, sets a baseline buffer period factor, incrementally adjusts the original playback rhythm ratio, calculates the upper limit of the adjusted buffer period based on the ratio of playback segment rate to response interval, outputs the rhythm adjustment value and buffer period interval parameters corresponding to each frame segment, and generates a playback rhythm adjustment parameter table.

[0031] The buffer setting table generation submodule establishes a triplet structure table of frame segment number, rhythm adjustment ratio, and buffer period range based on the rhythm ratio and period upper limit value of each frame segment in the playback rhythm adjustment parameter table, and integrates it into the frame segment playback configuration index structure in chronological order to establish the vocal playback rhythm adjustment buffer setting table.

[0032] As a further aspect of the present invention, the synchronization control module includes:

[0033] The task tag integration submodule obtains the list of adjustment tags for the uploaded frames of the vocal task and the buffer setting table for the adjustment of the vocal playback rhythm. It classifies and compares the data according to the frame timestamp and the teaching segment identifier, extracts the records with the same teaching segment index from the two types of data, reads all tag types, tag positions and tag parameters in the corresponding segments in the tag list, extracts the playback rhythm adjustment ratio and buffer period upper limit parameter of the corresponding segment, constructs paired data pairs according to the segment order, and summarizes the pairing results in all segments to generate a tag rhythm pairing cluster structure.

[0034] The paragraph instruction construction submodule reconstructs the instruction structure of the pairing groups in each teaching paragraph according to the tag rhythm pairing cluster structure, reads each tag item in the tag content set, adds the corresponding rhythm setting parameters in sequence, constructs a three-element instruction structure of tag type, rhythm adjustment factor, and buffer cycle time, binds the three-element with the paragraph unique index, constructs a serialized array of all paragraph instruction items, and obtains the segmented rhythm synchronization instruction set.

[0035] The synchronization control structure generation submodule, based on all the record items in the segmented rhythm synchronization instruction set, arranges them in ascending order according to the segment number, synchronizes the timestamp and control status flag, and writes them uniformly into the cloud storage to establish a cloud-based remote vocal music teaching rhythm control synchronization record.

[0036] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0037] In this invention, by dynamically identifying transmission congestion sections by collecting real-time audio stream and bandwidth capacity information in the upload channel, and combining this with overlap filtering and rhythm adjustment based on upload time, the problem of congestion and imbalance during audio transmission is effectively alleviated. Key teaching sections are marked based on the response time and density of teacher feedback signals, and the playback rate and buffering period of teaching frames are adjusted as needed to achieve a high degree of matching between feedback content and playback rhythm. By uniformly controlling the teaching upload and playback process, a synchronous structure for the teaching rhythm is constructed, significantly enhancing the coordination between audio upload, feedback response, and playback. This improves the stability and smoothness of vocal music teaching audio transmission in a dynamic network environment. Attached Figure Description

[0038] Figure 1 This is a system flowchart of the present invention;

[0039] Figure 2 This is a flowchart of the network channel monitoring module of the present invention;

[0040] Figure 3 This is a flowchart of the audio upload control module of the present invention;

[0041] Figure 4 This is a flowchart of the signal recognition module of the present invention;

[0042] Figure 5 This is a flowchart of the playback buffer adjustment module of the present invention;

[0043] Figure 6 This is a flowchart of the synchronization control module of the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0045] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0046] Please see Figure 1 The cloud-based remote interactive vocal teaching system includes:

[0047] The network channel monitoring module collects the real-time audio stream size, the number of audio data packets per unit time, and the maximum bandwidth capacity of the channel in the teacher's upload channel. Based on the data accumulation value per second divided by the bandwidth capacity, it calculates the bandwidth utilization ratio sequence of the vocal upload task, marks the time period in the sequence where the data accumulation frequency is greater than the standard frequency threshold, and performs segmented evaluation and marking of the density of vocal data accumulation within the interval, generating an audio channel data accumulation segment identifier set.

[0048] The audio upload control module filters the time points in the current teaching task whose audio frame upload times coincide with the stacking section based on the audio channel data stacking section identifier set, extracts the timestamp corresponding to each vocal teaching content, determines whether the upload rhythm falls into the stacking section, sets the transmission interval extension item if they coincide, sets the upload frame group delay frame flag according to the stacking section length, updates the upload scheduling table, and generates a list of vocal task upload frame adjustment tags.

[0049] The comment signal recognition module listens to the teacher comment signal frames received in the student terminal, extracts the response timestamps of the text content and audio content of each comment, classifies and counts the density of comment frames according to teaching segments, filters segments with a time interval of less than a set interval threshold, extracts the time index of the corresponding teaching segment and marks it as the comment dense interval, and generates a vocal comment dense segment time index table.

[0050] The playback buffer adjustment module is based on the vocal review dense segment time index table. It compares the current teaching audio segment playback rate with the overlap of the review dense interval. If the teaching frame playback start time is earlier than the average interval time of the review response frame, the playback buffer trigger point is marked, the playback rhythm adjustment ratio is updated and the buffer period range is expanded, and a vocal playback rhythm adjustment buffer setting table is generated.

[0051] The synchronization control module adjusts the tag list and the vocal playback rhythm adjustment buffer setting table based on the uploaded frames of the vocal task. It establishes the instruction set corresponding to the teaching segment according to the order of the teaching segments, and synchronously binds the uploaded adjustment tags and playback buffer settings under the same teaching segment. It constructs a unified control structure consisting of tag content, rhythm instructions, and segment identifiers, and generates a cloud-based remote vocal teaching rhythm control synchronization record.

[0052] The audio channel data accumulation segment identifier set includes accumulation time period identifier, accumulation frequency label, and accumulation density level; the vocal task upload frame adjustment label list includes frame group delay label, upload time adjustment label, and scheduling table update record; the vocal review dense segment time index table includes review segment index, review response density, and continuous review interval information; the vocal playback rhythm adjustment buffer setting table includes playback rate adjustment factor, buffer period parameter, and buffer trigger time point; and the cloud-based remote vocal teaching rhythm control synchronization record includes rhythm synchronization label, playback upload binding command, and teaching segment control identifier.

[0053] Please see Figure 2 The network channel monitoring module includes:

[0054] The data stream receiving submodule collects real-time audio stream data from the teacher's upload channel, records the total number of bytes transmitted per second and the corresponding timestamp information, and simultaneously calculates the number of audio data packets and the maximum bandwidth capacity of the channel per unit time. Combining the data stream timestamp and transmission length, it calculates the data accumulation value within a unit time period and generates a sequence of audio data accumulation per unit time.

[0055] When collecting real-time audio stream data from the teacher's upload channel, the system first configures a data monitoring interface at the access layer to capture data within a specific time range. The timestamp and byte count of each batch of audio data packets are recorded by a UDP stream monitoring program at 1-second sampling intervals, representing the total amount of data within that time period. And synchronously record the number of data packets. If 20 audio data packets are detected within a certain time period, and the average size of each data packet is 25kB, then the total number of bytes transmitted in that second is . At this time, the system simultaneously collects the maximum bandwidth capacity of the channel. And record the timestamp of that second. All monitored values ​​are stored in a cache to form a time series array. Then, the data accumulation value per unit time period is calculated based on the difference in data volume between adjacent time points. ,Right now ,like , ,but This process is repeated to create a stacked set of values ​​from 60 consecutive seconds of data. The system identifies abnormal fluctuations; if the difference between adjacent stack values ​​exceeds a baseline threshold... If so, it is marked as a stacking fluctuation event. Threshold The settings were determined based on the stability experiment results of the teacher-side network interface. In the experiment, during 10 audio upload tests, the backlog value fluctuated between 80 and 120 kB under normal conditions, and exceeded 150 kB under abnormal conditions. Therefore, a threshold was set. .

[0056] Table 1. Sample data of audio stream acquisition:

[0057]

[0058] As shown in Table 1, the system can clearly record the stacking change value per second, which is convenient for subsequent bandwidth calculation. If the average stacking value is 95kB within a 60s sampling period, this period is recorded as a stable segment. After continuous sampling and stacking judgment, the system generates a sequence of audio data stacking amount per unit time.

[0059] The bandwidth utilization calculation submodule reads the maximum bandwidth capacity value of the corresponding channel based on the accumulation amount value of each time period in the audio data accumulation amount sequence per unit time, calculates the bandwidth utilization rate value in each time period, and the bandwidth utilization rate is the ratio of the data accumulation amount per unit time to the bandwidth capacity, generating a bandwidth utilization rate ratio sequence.

[0060] Based on the accumulated data volume values ​​for each time period in the unit time audio data accumulation sequence, the maximum bandwidth capacity of the system call channel is... Calculate bandwidth utilization per second For example, if , ,but That is, the utilization rate is 5%, and the utilization rate per second is stored in a ratio sequence array. During continuous calculations, the system calculates the smoothing value of the current window based on the historical 5-second average utilization rate to identify sudden spikes. For example, with a 5-second sliding window, if the utilization rates for the five time periods are 0.05, 0.06, 0.05, 0.08, and 0.10, the average window utilization rate is 0.068. The judgment is based on a utilization rate change threshold. If the utilization rate increases by more than 2% within two consecutive seconds, it is marked as a bandwidth surge event. Threshold The threshold was determined through experimental statistics. In 20 network tests, when the increase exceeded 2%, the probability of audio buffer queue backlog increased to more than 20%, thus this threshold was determined to be within an acceptable range.

[0061] Calculation formula: ;

[0062] in, Indicates the first Stack size per second (MB). This represents bandwidth capacity (MB / s). Experimental results show that when... The channel state is stable when varying between 0.05 and 0.25. Mild congestion occurs when the value continuously exceeds 0.3. If the average value over three consecutive seconds... If the value exceeds the upper limit of the baseline interval by 0.25, it is marked as a moderately congested interval. The above calculations yield a complete bandwidth utilization ratio sequence, which is used for subsequent accumulation segment identification analysis. The final bandwidth utilization ratio sequence is then generated.

[0063] The stacking segment identification submodule sets a frequency threshold as the stacking frequency judgment threshold based on the ratio change in the bandwidth utilization ratio sequence, counts the number of times the ratio exceeds the stacking frequency judgment threshold in each time period, and divides the stacking density level according to the ratio size during the time period when the ratio frequency continuously exceeds the stacking frequency judgment threshold, establishes the marking information of the corresponding level in the interval, and generates the audio channel data stacking segment identification set.

[0064] Based on the changes in the ratios in the bandwidth utilization ratio sequence, the system sets a threshold for judging the accumulation frequency. And count the number of times the ratio exceeded this threshold in each time period. When The timer represents the frequency of a single accumulation event. If 15 accumulation events occur within a 60-second monitoring period, the accumulation frequency is 15 events per minute. Threshold The settings were determined based on experimental data, see Table 2.

[0065] Table 2. Bandwidth utilization threshold and backlog event statistics:

[0066]

[0067] Referring to Table 2, when the threshold is raised to 0.3, the system can effectively distinguish between lightly accumulated and stable areas. Subsequently, the aforementioned accumulation frequency data is retrieved, and time intervals where the ratio continuously exceeds the threshold are merged. If the interval between adjacent intervals is less than 2 seconds, they are merged into the same interval. Then, the accumulation density is classified according to the average ratio: 0.3–0.4 is Level 1 density, 0.4–0.5 is Level 2 density, and greater than 0.5 is Level 3 density. Taking a certain monitoring period as an example, the detection interval... The average ratio is 0.46, classifying it into a secondary dense zone. The system ultimately generates a mapping table based on the start and end times and level codes of each zone, in the following format: For example, (20, 40, 2). All segment data are indexed and organized into a unified set, generating an audio channel data accumulation segment identifier set.

[0068] Please see Figure 3 The audio upload control module includes:

[0069] The upload time filtering submodule obtains the audio channel data accumulation segment identifier set, extracts the upload audio frame timestamp sequence of each vocal teaching content in the current teaching task, extracts each segment time and each time point in the timestamp sequence for item-by-item matching, uses time intersection judgment logic to determine whether each audio frame time point is within any segment time, marks the time point falling into the accumulation segment, generates an equal-length Boolean vector as the overlap identifier sequence according to the marking status, and establishes an audio frame accumulation overlap status vector;

[0070] To obtain the audio channel data backlog segment identifier set, first extract all backlog segment times from this set, which are structured as a set of time period triples. Then, the timestamp information of the audio frames corresponding to each teaching content in the vocal teaching task is read, and an array of upload timestamp sequences is constructed. The process involves calling each stacked segment in the segment set, sequentially checking the timestamps of all frames for overlap according to the timeline, and using Boolean comparison logic for each check. satisfy If the time point is not specified, it is marked as an overlapping frame, and a Boolean state vector with the same length as the original timestamp array is constructed. ,in This indicates that the j-th frame coincides with a certain accumulation section. If there are 5 uploaded frames with timestamps of [2.3, 4.6, 7.8, 10.1, 13.4]s in the current task, and the stacking segments are [4.0, 5.0] and [10.0, 12.0], then the 2nd and 4th frames are determined to be overlapping frames, with corresponding Boolean vectors of 0, 1, 0, 1, 0. The system generates an overlap state array accordingly. After comparing all frame upload points in sequence and recording the overlap results, an audio frame stacking overlap state vector is established.

[0071] The frame group rhythm adjustment submodule extracts the segment length and number of overlapping frames for the corresponding frame group based on the overlapping frame points marked in the audio frame stacking state vector. It then calculates the timestamp set of all marked frames within the frame group, calculates the average timestamp, and uses the standard deviation as the frame time fluctuation index. The stacking segment length and frame time dispersion are jointly modeled using the following formula:

[0072] ;

[0073] The calculation obtains the frame group delay time adjustment value. Based on the original scheduling order of each frame group, the delay adjustment amount is inserted to reconstruct the frame scheduling interval structure and generate the upload frame rhythm delay matrix; where... This represents the upload latency adjustment value for the k-th frame group. This indicates the duration of the accumulation zone. This represents the original upload timestamp of the i-th overlapping frame. This represents the average timestamps of all overlapping frames within this segment. Indicates the number of overlapping frames within this segment;

[0074] Based on the frame points marked as overlapping in the audio frame stacking overlap state vector, extract the frame group number to which the frame point belongs and retrieve the corresponding stacking segment length. Number of overlapping frames within this frame group Get the upload timestamp of each overlapping point in each frame. Calculate the average timestamp of all overlapping frames in this group. Then, for each overlapping frame, calculate its deviation from the mean, square it, sum it, and divide by . The standard deviation of the timestamp is obtained by taking the square root, and then a weighted average is calculated by combining the stacking section length and the standard deviation, and then divided by the adjustment factor. To obtain the rhythm adjustment time The frame group delay value matrix is ​​obtained by performing the same operation on each frame group using the formula. Let the length of a certain stacking segment be... s, whose overlapping frame timestamps are respectively s, calculate the mean s, to obtain the square of each deviation. The sum is 2.25, divided by the number of frames 3, we get 0.75, and taking the square root gives the standard deviation. Substitute into the formula:

[0075] s;

[0076] This means the recommended upload latency for this frame group is 1.1165 seconds. This value is then used to construct the latency control field for the corresponding frame group in the latency matrix.

[0077] Table 3: Upload Pace Delay Calculation Table

[0078]

[0079] As shown in Table 3, the rhythm delay value can be adjusted by combining the segment duration and time fluctuation to obtain the uploaded frame rhythm delay matrix.

[0080] formula The computational logic is manifested as a composite model of the upload rhythm adjustment latency, where the numerator contains two core components with a time dimension: the length of the stacking segment. With frame time distribution standard deviation .in, The first term represents the duration of the current congestion on the timeline, reflecting the extent of overall network congestion and forming the structural basis for upload latency. The second term is the square root of the variance of the frame timestamps, i.e., the standard deviation, used to measure the uniformity of the distribution of uploaded frames within this segment. If the frame timestamps are relatively concentrated, this term approaches 0; if the frame distribution is dispersed, this term increases, indicating an increased probability of frame collisions. Therefore, this term participates in latency calculation as a dynamic adjustment factor. The square root operation maintains dimensional consistency while mitigating the impact of extreme deviations. The sum of these two terms constitutes the incremental base value of latency, reflecting the combined impact of congestion duration and frame density on upload rhythm adjustment, while the denominator term... To control the smoothness of the normalization term, its setting ensures that the latency per unit frame decreases as the number of frames increases. At the same time, the "+1" design avoids the division by zero problem when the number of frames is 0. Therefore, the formula as a whole logically reflects the principle that the upload latency is dynamically adjusted according to the accumulation time and the distribution of frame conflicts. Furthermore, the square root construction of the standard deviation term preserves the inter-frame fluctuation information instead of non-linear superposition, reflecting a control mechanism that integrates segment structure and frame behavior characteristics.

[0081] The frame group delay time adjustment value refers to the specific delay time value allocated to each frame group during audio upload scheduling. This is done to proactively adjust the upload timing of audio frames within a data accumulation segment in the network channel. The value is used to shift the original upload time backward by a corresponding amount of time, thereby increasing the transmission interval between frames and reducing the probability of conflicts caused by concentrated frame uploads. The value is determined by the duration of the accumulation segment and the dispersion of the time distribution of audio frames within the frame group. When the accumulation segment is long or the frame upload time is concentrated, the delay value increases accordingly to fully stretch the upload rhythm. Conversely, when the frame uploads are more dispersed or the accumulation time is short, the value is smaller, making only a slight adjustment. Therefore, the frame group delay time adjustment value reflects the ability of the upload scheduling system to dynamically control frame group scheduling behavior based on local network congestion and frame upload density. It is a key numerical parameter connecting network status awareness and upload task scheduling.

[0082] The upload tag generation submodule constructs upload frame adjustment tags based on the delay adjustment value corresponding to each frame group in the upload frame rhythm delay matrix, and generates triplet structure tags according to the frame group index. ,in The classification rule is: if Marked as D2, if Marked as D1, if Marked as D0, all frame group tags are appended to the upload scheduling control table in a time-series manner to create a list of adjustment tags for the upload frames of the vocal task.

[0083] Based on the latency values ​​of all frame groups in the uploaded frame rhythm latency matrix For each value, a numerical range is divided, and a triplet label is generated based on its numerical range. The frame group number is read from the original scheduling table, the corresponding delay value comes from the matrix item, and the label type is classified according to the delay magnitude, with the following rules set: If Then it is marked as D2, indicating a significant delay; if Then it is marked as D1; ​​if Then it is marked as D0. Taking frame group number G1 as an example, its corresponding delay value is 1.1165s, which meets the D1 interval, so the label [G1, 1.1165, D1] is generated. All labels are sorted in ascending order by frame group number and written into the upload scheduling control table in the order of generation. After completing the frame control instruction update operation in the scheduling table, a list of adjustment labels for the vocal task upload frame is established.

[0084] Please see Figure 4 The review signal recognition module includes:

[0085] The signal frame listening submodule listens to all teacher comment signal frames received in the student terminal, extracts the response timestamps of the text and audio content in each comment frame, and assembles the timestamps into a comment timeline sequence according to the frame order, constructs a comment time data table for the corresponding teaching segment, and generates a comment signal time sequence dataset.

[0086] To monitor the teacher's comment signal frames received on the student's terminal, the frame content capture interface needs to be activated. This involves extracting the comment signal data stream from the receive buffer frame by frame, determining if the frame type field is a "comment" identifier, and if a match is found, proceeding to the content parsing stage. This involves extracting the text information segments and audio data blocks encapsulated in fields within the frame, and then retrieving the corresponding timestamps by calling the audio timecode and the text reception trigger time, respectively. The timestamps of each comment frame are arranged chronologically to construct a comment timeline sequence. Then, based on the teaching task structure, the comment frames in the timeline are classified into paragraphs, establishing a correspondence between paragraph IDs and comment timestamps. If 12 comment frames are monitored in a lesson, their timestamps would be [32.4, 33.6, 35.2, 70.1, 70.4, 70.9, 110.3, 111.2, 135.4, 135.6, 136.0, 138.2]. The course structure is divided into segments, with segments 0–60 seconds for segment 1, 60–120 seconds for segment 2, and 120–180 seconds for segment 3. The first three frames belong to segment 1, frames 4 to 8 belong to segment 2, and the last four frames belong to segment 3. Based on the above frame segmentation results and time series sorting relationship, the system establishes a unified data table structure. The structure fields include frame number, timestamp, segment number, and frame type identifier. This constructs the basic data resources for subsequent intensive judgment of comments and generates a comment signal time series dataset.

[0087] The dense review filtering submodule is based on the review signal time series dataset. It groups the review frame sequence according to the teaching paragraph, calculates the time interval value sequence between consecutive review frames in each paragraph, filters consecutive time interval segments that are less than the interval threshold, outputs the segment start and end time index sequence, and generates a set of consecutive review frame time intervals.

[0088] Based on the time series dataset of comment signals, the frame data is grouped according to the teaching segment number to which it belongs. The timestamp sequence of comment frames in each group is obtained after being sorted by time. The time difference between adjacent comment frames is calculated and an interval sequence array is generated. Set a threshold for the review frame interval. The system filters all frame segments with consecutive intervals less than a threshold. Frames with adjacent time intervals less than 1 second are considered as a continuous dense comment sequence. The start and end times of these frames are recorded as a time interval. The filtering results for each segment are merged and output. If the comment time sequence in segment 2 is [70.1, 70.4, 70.9, 73.2], with an interval of [0.3, 0.5, 2.3], then the first 3 frames form a dense segment with intervals of 70.1 and 70.9. The last frame is removed because the interval is greater than 1 second. The system uses the time index segment of this segment as the dense comment period within that segment and continues to process all comment frame groups within all segments, extracting all interval information that meets the interval criteria to generate a continuous comment frame time interval set.

[0089] The teaching segment annotation submodule extracts the corresponding teaching segment number for each time index segment in the continuous comment frame time interval set, performs dense comment marking in the teaching content structure, records the correspondence between the segment index and the dense label, and establishes a vocal comment dense segment time index table.

[0090] Based on all time intervals within the continuous comment frames, the positions of the time index values ​​are cross-referenced with the teaching segment intervals in the course structure. The teaching segment number to which each time interval belongs is extracted and a mapping table is generated. When constructing the mapping table, the time interval is used as the primary key index, and the teaching segment number is used as the content value. A comment density tag field is added, with the tag field fixed as "DENSE," used to identify that the segment contains continuous comment density behavior. All tag information is then organized and summarized according to the segment number order. The processing results are written into the teaching task segment structure table as additional segment structure annotation items. The final output structure is as follows: The four-tuple information, if there is a dense commentary section at 70.1 and 70.9 seconds in paragraph 2, then generate entries 2, 70.1, 70.9, DENSE. All tag information is arranged and archived in paragraph order, and a time index table of dense vocal commentary paragraphs is established.

[0091] Please see Figure 5 The playback buffer adjustment module includes:

[0092] The playback rate comparison submodule extracts the playback start timestamp and playback rate parameter value of the current frame segment of the teaching audio based on the vocal review dense segment time index table, calculates the time overlap length between the playback frame time period and the review dense interval, and determines whether the playback start point is earlier than the average response interval time of the review response frames in the dense interval. If the judgment condition is met, the current frame segment is marked as the playback buffer trigger frame segment, and a playback buffer trigger frame index set is generated.

[0093] Based on the time index table of dense segments in vocal music commentary, the playback start timestamp of the currently playing teaching audio frame is first extracted. Playback rate of frames Then extract the start and end time intervals of all densely reviewed sections from the index table. and the corresponding average response interval of the comment frames Based on the segment order, determine whether the current frame segment timestamp falls within any densely populated comment segment, and then further compare... Whether it is earlier than the average response interval within the review segment, i.e., whether it satisfies the requirement. If the current frame is considered to have a risk of buffer overlap, the current frame index needs to be recorded as the trigger point. If the start time of the current frame is 69.8 seconds, the corresponding comment segment is [70.1, 71.0], and the average response interval of the comment frame is 0.6 seconds, then because... If the trigger judgment is valid, the system records the frame number of that segment and continues to process all subsequent frame segments and perform judgment and comparison. After completing the comparison operation in all teaching segments, the trigger frame segment numbers are integrated and sorted according to the playback order to generate a playback buffer trigger frame index set.

[0094] The rhythm adjustment parameter calculation submodule extracts the playback rate value of the corresponding frame and the average response interval of the comment frame based on the frame segment index information in the playback buffer trigger frame index set, calculates the rate difference, sets a baseline buffer period factor, incrementally adjusts the original playback rhythm ratio, calculates the upper limit of the adjusted buffer period based on the ratio of playback segment rate to response interval, outputs the rhythm adjustment value and buffer period interval parameters corresponding to each frame segment, and generates a playback rhythm adjustment parameter table.

[0095] Extract the original playback rate for each frame segment number listed in the playback buffer trigger frame index set. Average response interval of review frames in the same dense review segment Calculate the rate difference The system checks if the playback rate is greater than the average review response frequency. If the difference is positive, there is a risk of playback being too fast, and playback rhythm adjustment is required. The system sets a buffer rhythm adjustment benchmark factor. The recommended rhythm adjustment ratio is calculated by multiplying the difference by this coefficient. And based on the original playback cycle of that frame segment Multiply Obtain the upper limit of the buffer period ,like Frames per second seconds, then Adjustment ratio The upper limit of the buffer period is Every second, the system processes all triggered frames sequentially according to the above rules, generating a playback rhythm adjustment parameter table.

[0096] The buffer setting table generation submodule establishes a triplet structure table of frame segment number, rhythm adjustment ratio, and buffer period range based on the rhythm ratio and period upper limit value of each frame segment in the playback rhythm adjustment parameter table, and integrates it into the frame segment playback configuration index structure in time sequence to establish the vocal playback rhythm adjustment buffer setting table.

[0097] Extract the corresponding rhythm adjustment ratio based on each frame number in the playback rhythm adjustment parameter table. With the upper limit of the buffer period The data is assembled into a triplet structure record item [frame segment number, tempo ratio, buffer period upper limit], and sorted according to the frame segment timestamp order. It is then written into a unified playback configuration table structure. This configuration table is divided into a segment number column, an adjustment ratio column, and a buffer interval column. It maintains all playback tempo adjustment requirements in a structured manner, forming a parameter index set suitable for playback scheduler calls. For example, if the 4th segment frame number is G4, the corresponding ratio is 0.1075, and the buffer upper limit is 1.1075 seconds, then the record items G4, 0.1075, and 1.1075 are generated. After all records are summarized, a vocal playback tempo adjustment buffer setting table is established.

[0098] Please see Figure 6 The synchronous control module includes:

[0099] The task tag integration submodule obtains the vocal task upload frame adjustment tag list and vocal playback rhythm adjustment buffer setting table. It classifies and compares the data according to the frame timestamp and teaching segment identifier, extracts the records with the same teaching segment index from the two types of data, reads all tag types, tag positions and tag parameters in the corresponding segment from the tag list, extracts the playback rhythm adjustment ratio and buffer period upper limit parameter of the corresponding segment, constructs paired data pairs according to the segment order, and summarizes the pairing results in all segments to generate a tag rhythm pairing cluster structure.

[0100] To obtain the list of uploaded frame adjustment tags for vocal tasks and the vocal playback rhythm adjustment buffer setting table, firstly, extract the paragraph index, tag type, tag position, and tag parameter value for each tag from the frame adjustment tag list. Simultaneously, parse the rhythm adjustment ratio and buffer cycle upper limit fields for all paragraphs from the buffer setting table. Then, compare the frame segment timestamp with the teaching segment identifier value one by one. The comparison condition is that the difference in frame segment timestamps is less than or equal to 0.2 seconds and the paragraph numbers are completely identical. For records that meet this condition, pair the tag content from the frame adjustment tag list with the rhythm parameters from the buffer setting table. For successfully matched paragraph items, combine all tag items and buffer rhythm parameters for that paragraph into a tag content set and a rhythm setting set. If a paragraph has a certain number of tags... If the quantity is 3, then the tag content set consists of three items. The rhythm setting set contains the unified rhythm adjustment ratio and cycle upper limit value for that paragraph. In this process, a mapping table for the paragraph dimension needs to be established to bind the tag set and rhythm setting set corresponding to each paragraph into data pairs. For example, the tag set corresponding to paragraph 3 is [Intensity=0.7, Position=1.3s, Type=A], [Intensity=0.5, Position=2.8s, Type=B], [Intensity=0.9, Position=4.1s, Type=A], while the corresponding rhythm setting set is [Rhythm Adjustment Ratio=1.12, Buffer Cycle=1.5s]. This forms a paired structure pair. All paragraph paired structures are merged after being uniformly formatted to form a summary structure, named the Tag Rhythm Pairing Cluster Structure, as shown in Table 4.

[0101] Table 4: Tag Rhythm Pairing Cluster Structure Table

[0102]

[0103] As shown in Table 4, each row in the structure represents the correspondence between the tag set and rhythm parameters under a teaching paragraph. The tag content is described in the format of "type@position_intensity", and finally a tag rhythm pairing cluster structure is generated.

[0104] The paragraph instruction construction submodule reconstructs the instruction structure of the pairing groups in each teaching paragraph based on the tag rhythm pairing cluster structure. It reads each tag item in the tag content set, adds the corresponding rhythm setting parameters in sequence, constructs a three-element instruction structure of tag type, rhythm adjustment factor, and buffer cycle time, binds the three-element to the paragraph unique index, constructs a serialized array of all paragraph instruction items, and obtains the segmented rhythm synchronization instruction set.

[0105] Based on the tag rhythm pairing cluster structure, the pairing groups in each teaching segment are processed. First, the tag set content and corresponding rhythm adjustment ratio and buffer cycle parameters in the pairing group are parsed. Then, the tag set is iteratively processed, reading the tag type field, time position field, and parameter value field of each tag. According to its position in the set and the two parameters in the rhythm setting set, it is combined to form three elements: tag type, rhythm adjustment ratio, and buffer cycle time, constructing a ternary structure. Then, this ternary structure is concatenated with the segment's unique number to form a quaternary record item. During operation, if there are two label items [B@0.9s_0.6, C@3.3s_0.8] in paragraph 4, the rhythm adjustment ratio is 0.98, and the upper limit of the buffer period is 1.2 seconds, then the constructed quadruple is [4, B, 0.98, 1.2], [4, C, 0.98, 1.2]. All quadruple record items are combined to form a segmented rhythm synchronization instruction set, which is represented as a structured array. Each item in the array is the rhythm control information within a teaching paragraph. This array will be used for the downstream control structure generation submodule operation to finally obtain the segmented rhythm synchronization instruction set.

[0106] The synchronization control structure generation submodule is based on all the record items in the segmented rhythm synchronization instruction set. It is sorted in ascending order according to the segment number, and the synchronization timestamp and control status flag are written to the cloud storage to establish a cloud-based remote vocal music teaching rhythm control synchronization record.

[0107] Based on the records of the segmented rhythm synchronization instruction set, sort them in ascending order, with the sorting rule being that the segment numbers are from 1 to n. If there are duplicate segment numbers, maintain the original tag order. Then, insert each record into the control structure template. Fill in the remote control identifier field in the structure template, using a Boolean value of 1 to indicate that the remote control is valid. Fill in the synchronization timestamp field, aligning it with the tag time position field and retaining it to the millisecond level. Finally, fill in the control status flag field, assigning a value according to the tag type. If the tag type is A, the status value is 01; if the type is B, the status value is 10. If the value is C, the status value is 11. After completing the field completion, all structure items are written to the cloud control table storage module. The record structure format is [paragraph number, tag type, rhythm adjustment ratio, buffer period, control flag, synchronization timestamp, control identifier], such as [3, A, 1.12, 1.5, 01, 1300ms, 1], [3, B, 1.12, 1.5, 10, 2800ms, 1], [4, C, 0.98, 1.2, 11, 3300ms, 1], etc., which ultimately form a unified cloud-based remote vocal music teaching rhythm control synchronization record.

[0108] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A cloud-based vocal music remote interactive teaching system, characterized in that, The network channel monitoring module collects the real-time audio stream size in the teacher terminal upload channel, the number of audio data packets per unit time, and the maximum bandwidth capacity of the channel, calculates the bandwidth utilization ratio sequence of the vocal music upload task, evaluates and marks the data accumulation density in the interval, and generates an audio channel data accumulation section identification set; The audio upload control module filters the time points where the audio frame upload time of the current teaching task coincides with the accumulation section according to the audio channel data accumulation section identification set, judges whether the upload rhythm falls into the accumulation section, sets the upload frame group delay frame label according to the length of the accumulation section, and generates a vocal music task upload frame adjustment label list; The comment signal recognition module listens to the teacher comment signal frame received in the student terminal, extracts each comment text content and audio content response timestamp, filters the segment with a continuous comment frame time interval below the set interval threshold, extracts the time index and marks it as a comment dense interval, and generates a vocal music comment dense paragraph time index table; The playback buffer adjustment module compares the current teaching audio segment playback rate with the comment dense interval overlap degree based on the vocal music comment dense paragraph time index table, marks the playback buffer trigger point, updates the playback rhythm adjustment ratio and expands the buffer period range, and generates a vocal music playback rhythm adjustment buffer setting table; The synchronization control module establishes a teaching segment corresponding instruction set according to the vocal music task upload frame adjustment label list and the vocal music playback rhythm adjustment buffer setting table, synchronously binds the upload adjustment label and the playback buffer setting content, and generates a cloud remote vocal music teaching rhythm control synchronization record. The accumulation section specifically refers to the time period in the bandwidth utilization ratio sequence where the data accumulation frequency is greater than the standard frequency threshold.

2. The cloud-based remote vocal interactive teaching system according to claim 1, wherein, The audio channel data accumulation section identification set includes accumulation period identification, accumulation frequency label, and accumulation density level. The vocal music task upload frame adjustment label list includes frame group delay label, upload time adjustment label, and scheduling table update record. The vocal music comment dense paragraph time index table includes comment paragraph index, comment response density, and continuous comment interval information. The vocal music playback rhythm adjustment buffer setting table includes playback rate adjustment factor, buffer period parameter, and buffer trigger point. The cloud remote vocal music teaching rhythm control synchronization record includes rhythm synchronization label, playback upload binding instruction, and teaching segment control identification.

3. The cloud-based remote vocal interactive teaching system according to claim 1, wherein, The network channel monitoring module includes:

4. The cloud-based remote vocal interactive teaching system according to claim 1, wherein, The data stream receiving submodule collects real-time audio stream data content in the teacher terminal upload channel, records the total number of data bytes transmitted per second and the corresponding timestamp information, and simultaneously counts the number of audio data packets per unit time and the maximum bandwidth capacity value of the channel, calculates the data accumulation value per unit time period based on the data stream timestamp and transmission length, and generates a unit time audio data accumulation sequence; The bandwidth utilization calculation submodule calculates the bandwidth utilization rate value in each time period according to the accumulation value of each time period in the unit time audio data accumulation sequence, reads the maximum bandwidth capacity value of the corresponding channel, and generates a bandwidth utilization ratio sequence. ​ The accumulation section identification sub-module sets a frequency threshold as an accumulation frequency judgment threshold based on the change of the ratio in the bandwidth utilization ratio sequence, counts the number of occurrences of the ratio exceeding the accumulation frequency judgment threshold in each time period, and divides the accumulation density level according to the ratio size in the time period in which the ratio frequency continuously exceeds the accumulation frequency judgment threshold, establishes the marking information of the interval corresponding to the level, and generates an audio channel data accumulation section identification set.

5. The cloud-based remote vocal interactive teaching system according to claim 1, wherein, The audio uploading control module comprises: The uploading time screening sub-module acquires the audio channel data accumulation section identification set, extracts the uploading audio frame timestamp sequence of each vocal music teaching content in the current teaching task, and extracts each time point in the section time and the timestamp sequence for item-by-item matching. The time intersection judgment logic is used to determine whether each audio frame time point is in any section time. The time points falling into the accumulation section are marked. The equal-length Boolean vector is generated according to the marking state as the overlap identification sequence, and the audio frame accumulation overlap state vector is established; The frame group rhythm adjustment sub-module extracts the section length and the number of overlapping frames of the corresponding frame group according to the frame points marked as overlapping in the audio frame accumulation overlap state vector, counts the timestamp set of all marked frames in the frame group, calculates the average value of the timestamp, uses the standard deviation as the frame time fluctuation index, jointly models the accumulation section length and the frame time dispersion, calculates the frame group delay time adjustment value, inserts the delay adjustment amount according to the original scheduling order of each frame group, reconstructs the frame scheduling interval structure, and generates an uploading frame rhythm delay matrix; The uploading label generation sub-module constructs the uploading frame adjustment label based on the delay adjustment value corresponding to each frame group in the uploading frame rhythm delay matrix, generates a triple structure label according to the frame group index, and appends all frame group labels to the uploading scheduling control table in the form of a time sequence to establish a vocal music task uploading frame adjustment label list.

6. The cloud-based remote vocal interactive teaching system according to claim 5, wherein, The calculation formula of the frame group delay time adjustment value is: ； wherein, denotes the upload delay time adjustment value of the kth frame group, denotes the duration of the accumulation section, denotes the original upload timestamp of the ith coincident frame, denotes the average of all coincident frame timestamps within the accumulation section, denotes the number of coincident frames within the accumulation section.

7. The cloud-based remote vocal interactive teaching system according to claim 1, wherein, The comment signal recognition module comprises: The signal frame monitoring sub-module monitors all teacher comment signal frames received in the student terminal, extracts the response timestamp of the text content and the audio content in each comment frame, and groups the timestamps in the order of frames to form a comment time axis sequence, constructs a comment time data table corresponding to the teaching paragraph, and generates a comment signal time sequence data set; The comment dense screening sub-module groups the comment frame sequence according to the teaching paragraph based on the comment signal time sequence data set, calculates the time interval value sequence between the continuous comment frames in each paragraph, screens the continuous time interval segments less than the interval threshold, outputs the segment start and end time index sequence, and generates a continuous comment frame time interval set; The teaching paragraph labeling sub-module extracts the teaching paragraph number corresponding to each time index section in the continuous comment frame time interval set, labels the comment dense in the teaching content structure, records the correspondence between the paragraph index and the dense label, and establishes a vocal music comment dense paragraph time index table.

8. The cloud-based remote vocal interactive teaching system according to claim 1, wherein, The playback buffer adjustment module comprises: The playing rate comparison submodule extracts the playing start time stamp and the playing rate parameter value of the current frame segment of the teaching audio based on the vocal comment dense paragraph time index table, calculates the time overlap length of the playing frame time segment and the comment dense interval, and judges whether the playing start point is earlier than the average response interval time of the comment response frame in the dense interval. If the judgment condition is met, the current frame segment is marked as a playing buffer trigger frame segment, and a playing buffer trigger frame index set is generated; The rhythm adjustment parameter calculation submodule extracts the corresponding frame playing rate value and comment frame average response interval time according to the frame segment index information in the playing buffer trigger frame index set, calculates the rate difference value, sets a reference buffer period factor, incrementally adjusts the original playing rhythm ratio, calculates the upper limit value of the adjusted buffer period according to the proportional relationship between the playing segment rate and the response interval, outputs the corresponding rhythm adjustment value and buffer period interval parameter of each frame segment, and generates a playing rhythm adjustment parameter table; The buffer setting table generation submodule establishes a frame segment number, rhythm adjustment ratio, and buffer period range triple structure table according to the rhythm ratio and period upper limit value of each frame segment in the playing rhythm adjustment parameter table, and integrates and writes it into the frame segment playing configuration index structure in time sequence order to establish a vocal playing rhythm adjustment buffer setting table.

9. The cloud-based remote vocal interactive teaching system according to claim 1, wherein, The synchronization control module comprises: The task label integration submodule obtains the vocal task upload frame adjustment label list and the vocal playing rhythm adjustment buffer setting table, classifies and compares according to the frame timestamp and teaching paragraph identifier, extracts the record items with the same teaching paragraph index in the two types of data, respectively reads all label types, label positions and label parameters in the corresponding paragraph in the label list, extracts the playing rhythm adjustment ratio and buffer period upper limit parameter of the corresponding paragraph, constructs a pair of data according to the paragraph order, uniformly summarizes all the pairing results in the paragraphs, and generates a label rhythm pairing cluster structure; The paragraph instruction construction submodule reconstructs the instruction structure of each teaching paragraph in the label rhythm pairing cluster structure, reads each label item in the label content set, sequentially adds the corresponding rhythm setting parameter, constructs a triple instruction structure of label type, rhythm adjustment factor, and buffer period time, binds the triple and the unique paragraph index, constructs a serialized array of all paragraph instruction items, and obtains a segmented rhythm synchronization instruction set; The synchronization control structure generation submodule arranges all the record items in the segmented rhythm synchronization instruction set in ascending order according to the paragraph number, synchronizes the time stamp and the control state flag bit, and uniformly writes them into the cloud storage to establish a cloud remote vocal teaching rhythm control synchronization record.

Images