Method and device for determining bit rate threshold, storage medium and electronic device

Abstract

The present invention discloses a method and device for determining a bit rate threshold, a storage medium and an electronic device. The method comprises: obtaining a first group of media information corresponding to a group of bit rates of a target media resource; inputting the media information in the first group of media information into a target stutter rate prediction model in sequence to obtain a predicted stutter rate corresponding to the input media information; and determining a first bit rate threshold of the target media resource under a first definition and a first network bandwidth according to the predicted stutter rate corresponding to the input media information. The present invention solves the technical problem of a high video stutter rate caused by increasing the video bit rate.

Description

Technical Field

[0001] The present invention relates to the field of computers, and in particular to a method and device for determining a bit rate threshold, a storage medium, and an electronic device. Background Art

[0002] In the related art, the higher the video bit rate, the higher the video quality. The quality of the video can be changed by changing the video bit rate. However, the higher the bit rate, the larger the video file, and the higher the network conditions required for watching the video. If the network conditions do not meet the bit rate requirements, the video will freeze.

[0003] In the prior art, the number of freezes when users watch videos is usually counted, the current average freeze rate of users is counted, and then the freeze rate increase is estimated in the same proportion according to the bit rate increase ratio. However, some users with good network conditions and high bandwidth may still be able to watch high-bitrate videos smoothly, and the freeze rate of these users directly based on the estimated ratio is not appropriate; some users with poor network conditions may have an exponential increase in freeze rate when watching high-bitrate videos, rather than a linear increase. In this case, increasing the bit rate is likely to lead to a higher video freeze rate, affecting the user experience.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present invention provide a method and device for determining a bit rate threshold, a storage medium, and an electronic device, so as to at least solve the technical problem of a high video freeze rate caused by increasing the video bit rate.

[0006] According to one aspect of an embodiment of the present invention, a method for determining a bitrate threshold is provided, comprising: obtaining a first group of media information corresponding to a group of bitrates of a target media resource, wherein each media information in the first group of media information comprises a resource identification vector of the target media resource, a first definition of the target media resource, a corresponding bitrate in the group of bitrates, and a first network bandwidth of the target media resource; inputting the media information in the first group of media information into a target stuttering rate prediction model in sequence to obtain a predicted stuttering rate corresponding to the input media information, wherein each predicted stuttering rate is used to represent a stuttering rate when the target media resource is played under the first definition, a corresponding bitrate, and the first network bandwidth, and the target stuttering rate prediction model is used to determine the predicted stuttering rate according to the resource identification vector, the first definition, a corresponding bitrate, and the first network bandwidth; determining a first bitrate threshold of the target media resource under the first definition and the first network bandwidth according to the predicted stuttering rate corresponding to the input media information, wherein the first bitrate threshold is used to represent a threshold allowed to be adjusted when the bitrate of the target media resource is adjusted.

[0007] Optionally, the step of sequentially inputting the media information in the first group of media information into the target stuttering rate prediction model to obtain the predicted stuttering rate corresponding to the input media information comprises: sequentially inputting the media information in the first group of media information into the target stuttering rate prediction model in the order of the bit rates in the group of bit rates from small to large to obtain the predicted stuttering rate corresponding to the input media information; the step of determining the first bit rate threshold of the target media resource under the first clarity and the first network bandwidth according to the predicted stuttering rate corresponding to the input media information comprises: obtaining the The target stuttering rate prediction model sequentially obtains the predicted stuttering rate; when it is detected that the predicted stuttering rate currently obtained is greater than a preset stuttering rate threshold, and the predicted stuttering rate previously obtained before the current time is less than or equal to the stuttering rate threshold, the target bitrate corresponding to the predicted stuttering rate previously obtained is determined as the first bitrate threshold of the target media resource, wherein the predicted stuttering rate previously obtained is the predicted stuttering rate determined by the target stuttering rate prediction model according to the resource identification vector, the first clarity, the target bitrate and the first network bandwidth.

[0008] Optionally, the step of sequentially inputting the media information in the first group of media information into the target stuttering rate prediction model to obtain the predicted stuttering rate corresponding to the input media information comprises: sequentially inputting the media information in the first group of media information into the target stuttering rate prediction model in the order of the bit rates in the group of bit rates from large to small to obtain the predicted stuttering rate corresponding to the input media information; the step of determining the first bit rate threshold of the target media resource under the first clarity and the first network bandwidth according to the predicted stuttering rate corresponding to the input media information comprises: obtaining the The target stuttering rate prediction model sequentially obtains the predicted stuttering rate; when it is detected that the predicted stuttering rate currently obtained is less than or equal to a preset stuttering rate threshold, and the predicted stuttering rate previously obtained before the current time is greater than the stuttering rate threshold, the target bit rate corresponding to the predicted stuttering rate currently obtained is determined as the first bit rate threshold of the target media resource, wherein the predicted stuttering rate currently obtained is the predicted stuttering rate determined by the target stuttering rate prediction model according to the resource identification vector, the first clarity, the target bit rate and the first network bandwidth.

[0009] Optionally, before inputting the media information in the first group of media information into the target stuttering rate prediction model in sequence, the method also includes: obtaining sample media information of a sample media resource and an actual sample stuttering rate of the sample media resource, wherein the sample media information includes a resource identification vector of the sample media resource, the clarity of the sample media resource, the bit rate of the sample media resource and the network bandwidth of the sample media resource; training an original stuttering rate prediction model through the sample media information and the actual sample stuttering rate, and during the training process, if a loss value between a predicted sample stuttering rate determined by the original stuttering rate prediction model according to the sample media resource and the actual sample stuttering rate does not meet a preset convergence condition, adjusting parameters in the original stuttering rate prediction model, and if the loss value meets the preset convergence condition, ending the training, and determining the original stuttering rate prediction model at the end of the training as the target stuttering rate prediction model.

[0010] Optionally, the training of the original jamming rate prediction model through the sample media information and the actual sample jamming rate includes: performing multiple rounds of iterative training on the original jamming rate prediction model through the sample media information and the actual sample jamming rate, wherein i is a positive integer greater than or equal to 1, and the original jamming rate prediction model determined in the 0th round of training is the untrained original jamming rate prediction model, and the i-th round of training in the multiple rounds of iterative training includes: inputting the sample media information into the original jamming rate prediction model determined in the i-1th round of training to obtain the predicted sample jamming rate determined in the i-th round of training, wherein the original jamming rate prediction model includes multiple hidden layers, and each of the hidden layers includes a preset number of neural networks. Elements, each of the hidden layers includes a Dense layer, a LeakyReLU activation function layer, a Batchnormalization layer and a Dropout layer; if the loss value between the predicted sample jamming rate determined by the i-th round of training and the actual sample jamming rate does not meet the preset convergence condition, adjust the parameters in the original jamming rate prediction model obtained by the i-1-th round of training to obtain the original jamming rate prediction model determined by the i-th round of training; if the loss value between the predicted sample jamming rate determined by the i-th round of training and the actual sample jamming rate meets the preset convergence condition, determine the original jamming rate prediction model obtained by the i-1-th round of training as the target jamming rate prediction model.

[0011] Optionally, after determining the first bitrate threshold of the target media resource under the first clarity and the first network bandwidth, the method further includes: obtaining a second group of media information corresponding to the set of bitrates of the target media resource, wherein each piece of media information in the second group of media information includes a resource identification vector of the target media resource, a second clarity of the target media resource, a bitrate corresponding to the set of bitrates, and the first network bandwidth of the target media resource; inputting the media information in the second group of media information into the target stuttering rate prediction model in sequence to obtain a predicted stuttering rate corresponding to the media information in the second group of media information, wherein each piece of media information corresponding to the set of bitrates is a second clarity of the target media resource, a bitrate corresponding to the set of bitrates, and the first network bandwidth of the target media resource. The predicted stuttering rate corresponding to the media information in the second group of media information is used to indicate the stuttering rate when the target media resource is played at the second clarity, a corresponding bit rate and the first network bandwidth, and the target stuttering rate prediction model is used to determine the predicted stuttering rate based on the resource identification vector, the second clarity, a corresponding bit rate and the first network bandwidth; based on the predicted stuttering rate corresponding to the media information in the second group of media information, determine the second bit rate threshold of the target media resource at the second clarity and the first network bandwidth, wherein the second bit rate threshold is used to indicate the threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0012] Optionally, the method further includes: obtaining a third group of media information corresponding to the set of bit rates of the target media resource, wherein each piece of media information in the third group of media information includes a resource identification vector of the target media resource, the first clarity of the target media resource, a bit rate corresponding to the set of bit rates, and a second network bandwidth of the target media resource; inputting the media information in the third group of media information into a target stuttering rate prediction model in sequence to obtain a predicted stuttering rate corresponding to the media information in the third group of media information, wherein each piece of predicted stuttering rate corresponding to the media information in the third group of media information is The stuttering rate is used to indicate the stuttering rate when the target media resource is played at the first clarity, a corresponding bit rate and the second network bandwidth. The target stuttering rate prediction model is used to determine the predicted stuttering rate based on the resource identification vector, the first clarity, a corresponding bit rate and the second network bandwidth. According to the predicted stuttering rate corresponding to the media information in the third group of media information, a third bit rate threshold of the target media resource at the first clarity and the second network bandwidth is determined, wherein the third bit rate threshold is used to indicate the threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0013] Optionally, the method further includes: obtaining a fourth group of media information corresponding to the set of bit rates of the target media resource, wherein each media information in the fourth group of media information includes a resource identification vector of the target media resource, a third clarity of the target media resource, a bit rate corresponding to the set of bit rates, and a third network bandwidth of the target media resource; inputting the media information in the fourth group of media information into a target stuttering rate prediction model in sequence to obtain a predicted stuttering rate corresponding to the media information in the fourth group of media information, wherein each predicted stuttering rate corresponding to the media information in the fourth group of media information The rate is used to represent the stuttering rate when the target media resource is played under the third clarity, a corresponding bit rate and the third network bandwidth, and the target stuttering rate prediction model is used to determine the predicted stuttering rate according to the resource identification vector, the third clarity, a corresponding bit rate and the third network bandwidth; according to the predicted stuttering rate corresponding to the media information in the fourth group of media information, determine a fourth bit rate threshold of the target media resource under the third clarity and the third network bandwidth, wherein the fourth bit rate threshold is used to represent the threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0014] Optionally, after determining the first bit rate threshold of the target media resource under the first definition and the first network bandwidth, the method further includes: when the current bit rate of the target media resource is the original bit rate, the current definition of the target media resource is the first definition, and the current network bandwidth of the target media resource is the first network bandwidth, adjusting the bit rate of the target media resource from the original bit rate to the target bit rate, wherein the original bit rate is less than the first bit rate threshold, the target bit rate is less than or equal to the first bit rate threshold, and is greater than the original bit rate; or when the current bit rate of the target media resource is the original bit rate, the current definition of the target media resource is the first definition, and the difference between the current network bandwidth of the target media resource and the first network bandwidth meets a preset threshold, adjusting the bit rate of the target media resource from the original bit rate to the target bit rate, wherein the original bit rate is less than the first bit rate threshold, the target bit rate is less than or equal to the first bit rate threshold, and is greater than the original bit rate.

[0015] Optionally, after determining the first bitrate threshold of the target media resource under the first definition and the first network bandwidth, the method further includes: when the current bitrate of the target media resource is the original bitrate, the current definition of the target media resource is a target definition different from the first definition, and the current network bandwidth of the target media resource is a target network bandwidth different from the first network bandwidth, obtaining a target bitrate threshold of the target media resource under the target definition and the target network bandwidth, wherein the target bitrate threshold is a bitrate threshold determined according to a target predicted stuttering rate, and the target predicted stuttering rate is a predicted stuttering rate determined by the target stuttering rate prediction model according to the resource identification vector, the target definition, a corresponding bitrate in the set of bitrates, and the target network bandwidth; and increasing the bitrate of the target media resource from the original bitrate to the target bitrate, wherein the original bitrate is less than the target bitrate threshold, and the target bitrate is less than or equal to the target bitrate threshold and greater than the original bitrate.

[0016] According to another aspect of an embodiment of the present invention, a device for determining a bitrate threshold is provided, comprising: an acquisition module, configured to acquire a first group of media information corresponding to a group of bitrates of a target media resource, wherein each piece of media information in the first group of media information comprises a resource identification vector of the target media resource, a first definition of the target media resource, a corresponding bitrate in the group of bitrates, and a first network bandwidth of the target media resource; an input module, configured to input the media information in the first group of media information into a target stuttering rate prediction model in sequence, to obtain a predicted stuttering rate corresponding to the input media information, wherein each predicted stuttering rate is used to represent a stuttering rate when the target media resource is played at the first definition, a corresponding bitrate, and the first network bandwidth, and the target stuttering rate prediction model is used to determine the predicted stuttering rate according to the resource identification vector, the first definition, a corresponding bitrate, and the first network bandwidth; a determination module, configured to determine a first bitrate threshold of the target media resource at the first definition and the first network bandwidth according to the predicted stuttering rate corresponding to the input media information, wherein the first bitrate threshold is used to represent a threshold allowed to be adjusted when the bitrate of the target media resource is adjusted.

[0017] According to another aspect of the embodiments of the present invention, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for determining the bit rate threshold when running.

[0018] According to another aspect of the embodiments of the present application, a computer program product or a computer program is provided, the computer program product or the computer program includes a computer program, the computer program is stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device performs the above method for determining a bit rate threshold.

[0019] According to another aspect of an embodiment of the present invention, there is further provided an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the method for determining a bit rate threshold through the computer program.

[0020] In an embodiment of the present invention, a neural network model is used to learn the relationship between video bit rate, clarity, bandwidth, and resource identification vectors of media resources, and predict the change in the freeze rate after the bit rate is increased. Under the same clarity and bandwidth, the video freeze rate of the target media resource at different bit rates is obtained through the neural network model. By limiting the bit rate threshold, it is prevented that the bit rate of the target media resource is adjusted too much, resulting in a high video freeze rate and affecting the user experience. This solves the technical problem of a high video freeze rate caused by increasing the video bit rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0022] Figure 1 is a schematic diagram of an application environment of an optional method for determining a bit rate threshold according to an embodiment of the present invention;

[0023] Figure 2 is a flowchart of an optional method for determining a bit rate threshold according to an embodiment of the present invention;

[0024] Figure 3 is an optional flow chart according to an embodiment of the present invention;

[0025] Figure 4 is another optional flow chart according to an embodiment of the present invention;

[0026] Figure 5 is an optional system framework diagram according to an embodiment of the present invention;

[0027] Figure 6 is a schematic diagram of an optional CBOW model according to an embodiment of the present invention;

[0028] Figure 7 is a schematic diagram of an optional Skipgram model according to an embodiment of the present invention;

[0029] Figure 8 is an optional schematic diagram of similar words according to an embodiment of the present invention;

[0030] Fig. 9 is a schematic diagram of an optional deep neural network model structure according to an embodiment of the present invention;

[0031] Fig.10 is an optional network structure comparison diagram according to an embodiment of the present invention;

[0032] Fig.11 is a schematic diagram of an optional loss curve according to an embodiment of the present invention;

[0033] Fig.12 is a schematic structural diagram of an optional device for determining a bit rate threshold according to an embodiment of the present invention;

[0034] Fig.13 is a schematic structural diagram of an optional electronic device according to an embodiment of the present invention;

[0035] Fig.14 It is a block diagram of a computer system structure of an optional electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0037] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0038] According to one aspect of an embodiment of the present invention, a method for determining a bit rate threshold is provided. Optionally, as an optional implementation, the method for determining a bit rate threshold can be but is not limited to being applied to: Figure 1 The application environment shown in the figure includes a user device 102 , a network 110 and a server 112 .

[0039] Optionally, in this embodiment, the user device may be a terminal device configured with a target client, which may include but is not limited to at least one of the following: a mobile phone (such as an Android phone, an iOS phone, etc.), a laptop, a tablet computer, a PDA, a MID (Mobile Internet Devices), a PAD, a desktop computer, a smart TV, a vehicle-mounted terminal, etc. The target client may be a video client, an instant messaging client, a browser client, a game client, etc. The user device includes: a memory 104, a processor 106 and a display 108, wherein the memory 104 is used to store data including but not limited to storing target media resources, and the processor 106 is used to process data, including but not limited to parsing the resource identifier of the target media resource, parsing the bit rate and definition of the target media resource, and determining the bandwidth of the user device, etc. The display may be used to display the target media resource, and the target media resource includes but is not limited to video.

[0040] The above-mentioned network 110 may include but is not limited to: a wired network and a wireless network, wherein the wired network includes: a local area network, a metropolitan area network and a wide area network, and the wireless network includes: Bluetooth, WIFI and other networks that realize wireless communication.

[0041] The server 112 may be a single server, a server cluster consisting of multiple servers, or a cloud server. The server includes a database 114 and a processing engine 116, wherein the database is used to store data, including but not limited to the first set of media information. The processing engine may be used to perform the following steps:

[0042] Step S102, inputting the media information in the first group of media information into a target jam rate prediction model in sequence, and obtaining a predicted jam rate corresponding to the input media information;

[0043] Step S104: determining a first bitrate threshold of the target media resource under the first definition and the first network bandwidth according to the predicted freeze rate corresponding to the input media information.

[0044] The above is only an example and is not limited in this embodiment.

[0045] The key terms in this application are explained below:

[0046] Video bit rate: the number of data bits transmitted per unit time during data transmission, measured in kbps. The higher the bit rate, the higher the video quality, the larger the video file, and the higher the viewing network requirements.

[0047] Number of freezes: The number of times the video playback is intermittent, choppy, or even stuck while the user is watching a video in its entirety.

[0048] Stuttering rate: Stuttering rate usually refers to the total number of times a user stutters when watching a video within a period of time / the total number of video playbacks * 100%, where the total number of times a user stutters when watching a video is the number of stutters that occur in all videos watched during the period, assuming that multiple stutters when a user watches a video are counted as one stutter.

[0049] Kafka is a high-throughput distributed publish-subscribe messaging system that can process website log stream data for multiple consumers to pull consumption data. It has the characteristics of fault tolerance, persistence, and multi-partitioning.

[0050] Flink is an open source stream processing framework under the Apache Software Foundation. Its core is a distributed data stream engine that can process large-scale streaming data with high throughput, low latency and high performance.

[0051] A neural network is an artificial neural network, consisting of an input layer, a hidden layer, and an output layer. The hidden layer consists of multiple nodes. The hidden layer can have multiple layers, and each node represents a specific output function. The nodes of each layer are fully connected, and the connection between each two nodes is calculated through weights and passed to the subsequent nodes, and finally the calculation results are output in the output layer. A deep neural network is a neural network with many hidden layers.

[0052] One-hot encoding refers to the method of converting categorical variables into numerical variables. It sets the value of each type of feature variable to 1 and the others to 0. For example, assuming that videos have four definition levels: standard definition, high definition, ultra-high definition, and Blu-ray. If the user is watching standard definition, then his network definition one-hot encoding is (1,0,0,0). If he is watching high definition, then the network definition one-hot encoding is (0,1,0,0), and so on.

[0053] Word2Vec is an algorithm model that converts text language into numerical data. Since various text languages ​​cannot be input into the model for training, and one-hot encoding of text will result in too large a dimension, word2Vec uses a three-layer neural network to map the word vector in one-hot encoding form to a distributed word vector, which has a smaller dimension than one-hot encoding and is a numerical vector.

[0054] Optionally, as an optional implementation, as Figure 2 As shown, the above method for determining a bit rate threshold may be executed by a user device or a server, or may be executed by the user device and the server together. In the embodiment of the present application, the method is executed by the server as an example for description. The above method includes the following steps:

[0055] Step S202, obtaining a first set of media information corresponding to a set of bit rates of a target media resource, wherein each piece of media information in the first set of media information includes a resource identification vector of the target media resource, a first definition of the target media resource, a bit rate corresponding to the set of bit rates, and a first network bandwidth of the target media resource;

[0056] Among them, the above-mentioned target media resource can be a video. Video bit rate is the number of data bits transmitted per unit time during data transmission, and the unit is kbps. The above-mentioned set of bit rates can be selected according to actual conditions, for example, from 100 to 600, the bit rate is selected every 50, that is, the above-mentioned set of bit rates can be 100, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600. The above-mentioned resource identifier is the video name, and the numerical vector obtained by converting the text in the video name through the Word2Vec model. The clarity and network bandwidth between different media information in the above-mentioned first group of media information are the same, and the difference is the bit rate.

[0057] Step S204: inputting the media information in the first group of media information into a target jamming rate prediction model in sequence to obtain a predicted jamming rate corresponding to the input media information, wherein each predicted jamming rate is used to represent a jamming rate when the target media resource is played at the first definition, a corresponding bit rate, and the first network bandwidth, and the target jamming rate prediction model is used to determine the predicted jamming rate according to the resource identification vector, the first definition, a corresponding bit rate, and the first network bandwidth;

[0058] The target jam rate model is a neural network model obtained through machine learning. The neural network model is trained using training sample data, which includes a resource identification vector, definition, bit rate, and actual jam rate of a sample media resource.

[0059] Step S206: Determine a first bitrate threshold of the target media resource under the first clarity and the first network bandwidth based on the predicted stuttering rate corresponding to the input media information, wherein the first bitrate threshold is used to indicate a threshold allowed to be adjusted when adjusting the bitrate of the target media resource.

[0060] Among them, the target stuttering rate prediction model can be used to obtain the predicted stuttering rate corresponding to a set of bit rates of the target media resource under the first definition and the first network bandwidth. By predicting the impact of the stuttering rate on the user's video viewing, the first bit rate threshold of the target media resource can be set to prevent the video from being too high due to excessive bit rate adjustment when the bit rate of the target media resource is adjusted, thereby affecting the user experience.

[0061] The following is described by a specific embodiment:

[0062] Assuming that the above-mentioned set of bit rates is 100, 150, and 200, the above-mentioned first definition is 1080p, and the above-mentioned first network bandwidth is 1000 megabytes, then the above-mentioned first set of media information includes media information 1 (bit rate 100, definition is 1080p, network bandwidth is 1000 megabytes), media information 2 (bit rate 150, definition is 1080p, network bandwidth is 1000 megabytes), and media information 3 (bit rate 200, definition is 1080p, network bandwidth is 1000 megabytes).

[0063] The media information 1, media information 2, and media information 3 in the first group of media are input into the target stutter rate prediction model in sequence, and the stutter rate of the media information is estimated by the target stutter rate model to obtain the predicted stutter rate corresponding to the media information. The first bitrate threshold of the target media resource under the conditions of 1080p definition and 1000 Mbps network bandwidth can be obtained through the stutter rate estimated by the target stutter rate prediction model and the user's viewing experience at different stutter rates.

[0064] Optionally, the step of sequentially inputting the media information in the first group of media information into a target stuttering rate prediction model to obtain a predicted stuttering rate corresponding to the input media information comprises: sequentially inputting the media information in the first group of media information into a target stuttering rate prediction model in an ascending order of bit rates in the group of bit rates to obtain a predicted stuttering rate corresponding to the input media information;

[0065] As an optional implementation, take the media information 1, media information 2, and media information 3 included in the first group of media in the above embodiment as an example. According to the above group of bit rates 100, 150, and 200, the corresponding media information is sequentially input into the above target jam rate prediction model in order from small to large, and the corresponding predicted jam rate is obtained until the predicted jam rate corresponding to the first bit rate threshold of the target media resource is determined. Assuming that the preset jam rate threshold is 0.25, Figure 3As shown, media information 1 is input into the above-mentioned target jam rate prediction model. Assuming that the predicted jam rate estimated for the above-mentioned media information 1 is 0.1, and 0.1 is less than the jam rate threshold, then media information 2 is continuously input into the above-mentioned target jam rate prediction model. Assuming that the predicted jam rate estimated for the above-mentioned media information 2 is 0.3, and 0.3 is greater than the jam rate threshold, then the first bit rate threshold of the target media resource is determined to be 100, and the predicted jam rate corresponding to the first bit rate threshold of the target media resource is 0.1. Since the bit rate and the jam rate are proportional, a group of media information corresponding to bit rates 100, 150, and 200 are input into the target jam rate prediction model in order from small to large. When the media information 2 corresponding to the bit rate 150 is input into the target jam rate prediction model, the first bit rate threshold has been determined. At this time, the input of media information into the target jam rate prediction model can be stopped, that is, there is no need to continue to input media information 3 into the target jam rate prediction model. In this way, efficiency can be improved and resources can be saved. Therefore, the bit rate 100 in the media information 1 can be determined as the first bit rate threshold of the target media resource under the conditions of a definition of 1080p and a network bandwidth of 1000 Mbps.

[0066] The method of determining a first bitrate threshold of the target media resource at the first clarity and the first network bandwidth based on the predicted stuttering rate corresponding to the input media information includes: obtaining predicted stuttering rates obtained in sequence by the target stuttering rate prediction model; and when it is detected that the predicted stuttering rate currently obtained is greater than a preset stuttering rate threshold, and the predicted stuttering rate previously obtained before the current time is less than or equal to the stuttering rate threshold, determining the target bitrate corresponding to the predicted stuttering rate previously obtained as the first bitrate threshold of the target media resource, wherein the predicted stuttering rate previously obtained is a predicted stuttering rate determined by the target stuttering rate prediction model based on the resource identification vector, the first clarity, the target bitrate, and the first network bandwidth.

[0067] As an optional implementation, assuming that the above-mentioned jam rate threshold is 0.25, the above-mentioned predicted jam rate currently obtained is Figure 3 The target jam rate model shown in the figure predicts the predicted jam rate of media information 2 to be 0.3, which is greater than the preset jam rate threshold of 0.25. The predicted jam rate obtained last time is Figure 3 The target jamming rate model shown in is used to predict the predicted jamming rate of media information 1 to obtain a predicted jamming rate of 0.1, and the bit rate 100 in the media information 1 is determined as the first bit rate threshold of the target media resource.

[0068] In the above embodiment, by sequentially inputting the media information in the first group of media information into the target jam rate prediction model in the order of the bit rates in the above group of bit rates from small to large, the first bit rate threshold can be determined by the predicted jam rate outputted by the target jam rate prediction model each time. In this way, the efficiency of determining the first bit rate threshold can be improved.

[0069] Optionally, the step of sequentially inputting the media information in the first group of media information into a target jamming rate prediction model to obtain a predicted jamming rate corresponding to the input media information comprises: sequentially inputting the media information in the first group of media information into a target jamming rate prediction model in descending order of bit rates in the group of bit rates to obtain a predicted jamming rate corresponding to the input media information;

[0070] As an optional implementation, take the media information 1, media information 2, and media information 3 included in the first group of media in the above embodiment as an example. Assuming that the above freeze rate threshold is 0.4, the corresponding media information is sequentially input into the above target freeze rate prediction model in the order of the above group of bit rates 200, 150, and 100 from large to small to obtain the corresponding predicted freeze rate. Figure 4 As shown, media information 3 is input into the above-mentioned target jam rate prediction model. Assuming that the predicted jam rate estimated for the above-mentioned media information 3 is 0.5, and 0.5 is greater than the jam rate threshold value 0.4, then media information 2 is continuously input into the above-mentioned target jam rate prediction model. Assuming that the predicted jam rate estimated for the above-mentioned media information 2 is 0.3, and 0.3 is less than the jam rate threshold value 0.4, then the first bit rate threshold value of the target media resource is determined to be 150, and the predicted jam rate corresponding to the first bit rate threshold value of the target media resource is 0.3. Since the bit rate and the jam rate are proportional, a group of media information corresponding to bit rates 200, 150, and 100 are input into the target jam rate prediction model in order from large to small. When the media information 2 corresponding to the bit rate 150 is input into the target jam rate prediction model, the first bit rate threshold value has been determined. At this time, the input of media information into the target jam rate prediction model can be stopped, that is, there is no need to continue to input media information 1 into the target jam rate prediction model. In this way, efficiency can be improved and resources can be saved. Therefore, it can be determined that when the definition of the target media resource is 1080p and the network bandwidth is 1000 Mbps, the first bit rate threshold of the target media resource is 150.

[0071] The determining, based on the predicted stuttering rate corresponding to the input media information, the first bitrate threshold of the target media resource under the first clarity and the first network bandwidth includes: obtaining the predicted stuttering rates obtained in sequence by the target stuttering rate prediction model; when it is detected that the predicted stuttering rate obtained currently is less than or equal to a preset stuttering rate threshold, and the predicted stuttering rate obtained last time before the current time is greater than the stuttering rate threshold, determining the target bitrate corresponding to the predicted stuttering rate obtained currently as the first bitrate threshold of the target media resource, wherein the predicted stuttering rate obtained currently is the predicted stuttering rate determined by the target stuttering rate prediction model according to the resource identification vector, the first clarity, the target bitrate and the first network bandwidth.

[0072] As an optional implementation, assuming that the above-mentioned jam rate threshold is 0.4, the above-mentioned predicted jam rate currently obtained is Figure 4 The target jam rate model shown in the figure predicts the predicted jam rate of media information 2 to be 0.3, which is less than the preset jam rate threshold of 0.4. The predicted jam rate obtained last time is Figure 4 The target stuttering rate model shown in is used to predict the predicted stuttering rate of 0.5 for media information 3, which is greater than the preset stuttering rate threshold of 0.4. The predicted stuttering rate currently obtained is the predicted stuttering rate of 0.3 obtained by predicting media information 2, and its corresponding target bit rate is 0.3, then the target bit rate 0.3 is determined as the first bit rate threshold of the target media resource, that is, the bit rate 150 corresponding to media information 2 is determined as the first bit rate threshold of the target media resource.

[0073] In the above embodiment, by sequentially inputting the media information in the first group of media information into the target jamming rate prediction model in the order of the bit rates in the above group of bit rates from large to small, the first bit rate threshold can be determined by the predicted jamming rate outputted by the target jamming rate prediction model each time. In this way, the efficiency of determining the first bit rate threshold can be improved.

[0074] Optionally, before inputting the media information in the first group of media information into the target stuttering rate prediction model in sequence, the method also includes: obtaining sample media information of a sample media resource and an actual sample stuttering rate of the sample media resource, wherein the sample media information includes a resource identification vector of the sample media resource, the clarity of the sample media resource, the bit rate of the sample media resource and the network bandwidth of the sample media resource; training an original stuttering rate prediction model through the sample media information and the actual sample stuttering rate, and during the training process, if a loss value between a predicted sample stuttering rate determined by the original stuttering rate prediction model according to the sample media resource and the actual sample stuttering rate does not meet a preset convergence condition, adjusting parameters in the original stuttering rate prediction model, and if the loss value meets the preset convergence condition, ending the training, and determining the original stuttering rate prediction model at the end of the training as the target stuttering rate prediction model.

[0075] Optionally, the training of the original jamming rate prediction model through the sample media information and the actual sample jamming rate includes: performing multiple rounds of iterative training on the original jamming rate prediction model through the sample media information and the actual sample jamming rate, wherein i is a positive integer greater than or equal to 1, and the original jamming rate prediction model determined in the 0th round of training is the untrained original jamming rate prediction model, and the i-th round of training in the multiple rounds of iterative training includes: inputting the sample media information into the original jamming rate prediction model determined in the i-1th round of training to obtain the predicted sample jamming rate determined in the i-th round of training, wherein the original jamming rate prediction model includes multiple hidden layers, and each of the hidden layers includes a preset number of neural networks. Elements, each of the hidden layers includes a Dense layer, a LeakyReLU activation function layer, a Batchnormalization layer and a Dropout layer; if the loss value between the predicted sample jamming rate determined by the i-th round of training and the actual sample jamming rate does not meet the preset convergence condition, adjust the parameters in the original jamming rate prediction model obtained by the i-1-th round of training to obtain the original jamming rate prediction model determined by the i-th round of training; if the loss value between the predicted sample jamming rate determined by the i-th round of training and the actual sample jamming rate meets the preset convergence condition, determine the original jamming rate prediction model obtained by the i-1-th round of training as the target jamming rate prediction model.

[0076] As an optional implementation, Figure 5 The figure shows the system framework diagram. When training the deep neural network model, the input is the processed user network bandwidth and video information data (including: video bit rate, video clarity, etc.). After the neural network model learns the data information, the model is used to predict the user freeze rate data corresponding to the new improved bit rate.

[0077] In the system framework diagram, user device data is reported. Every time a user watches a video, event data is reported during the viewing process. Starting from entering the details page, relevant data will be reported for events such as loading advertisements, first video loading, first loading definition, first video frame rendering, loading subtitles, secondary buffering, dragging, switching definition, switching audio tracks, and playback completion. The data here is reported to Kafka in the background, and the data is saved for a preset length of time (for example, 8 hours). The flink big data framework engine is used to process the Kafka streaming data reported by the user device, and the data is stored in the database by filtering out empty data and data with abnormal values.

[0078] The resource identification vector of the sample media resource is a data vector obtained by converting the text of the video name. The database stores the names of various videos. Specifically, the video names obtained in the database can be converted into word vectors through the Word2Vec model.

[0079] Word2Vec is a natural language processing technology that uses a neural network model to learn word associations from a large amount of text data. After training, different words can be represented by vectors in numerical form. The similarity between vectors is compared by calculating cosine similarity. The formula for calculating the cosine similarity of two vectors is:

[0080]

[0081] Among them, A and B are two n-dimensional vectors, and Ai and Bi are the i-th values ​​in the vectors.

[0082] Word2Vec usually has two model architectures: CBOW and Skipgram. Figure 6 The CBOW model shown uses the one-hot encoded vector of the word as input and a V*N weight matrix W V*N Multiply them to get some 1*N vectors, and then concatenate these vectors (take the average) to get the N-dimensional vector h i , the hidden layer to the output layer is multiplied by a coefficient matrix W′ N*V , we get the vector y j , the training tries to make the corresponding elements of this vector close to the conditional probability of the predicted word appearing under the corpus context. The word vector is essentially an intermediate product of model training and is input into the hidden layer weight matrix W V*N , all input one-hot encoding vectors multiplied by it are equivalent to the corresponding word vectors.

[0083] like Figure 7As shown in the figure, the Skipgram model takes a word w as input and predicts 2c word vectors as output. It is expected that the prediction probability of these words is greater than that of other words. The other steps are similar to CBOW.

[0084] Since the number of video titles watched by users is limited, in order to improve the accuracy of the model, a large number of video titles are needed. Words related to the video title can be searched in the database to obtain similar words. Figure 8 As shown in the figure, the top ten words closest to "Soldier Assault" are found in the database. The first column shown in the figure is the word, and the second column is the correlation between the word and "Soldier Assault". The video name can be converted into a 200-dimensional vector through conversion. When encountering a word w1 that is not in the pre-trained word vector, Jieba word segmentation is used, and the word vectors found after word segmentation are averaged as the word vector of w1. This can increase the amount of training sample data and improve the accuracy of the model.

[0085] As an optional implementation, the original jam rate prediction model is a deep neural network model. Before training the original jam rate prediction model, a deep neural network model is built. The structure of the deep neural network model is as follows: Fig. 9 As shown in the figure, there are 7 layers in total, including 1 input layer, 6 hidden layers and 1 output layer, using the LeakyReLU activation function, Batch-Normalization method and dropout method. The main function of the LeakyReLU activation function is to solve the gradient disappearance and accelerate convergence, the main function of the batchnormalization layer is to make the neural network converge faster and more stable, and the function of the dropout layer is to prevent overfitting.

[0086] LeakyReLU activation function: multiply all values ​​less than 0 by a non-zero slope, where, as in formula 1, a i is a fixed parameter greater than 1, input data of the LeakyReLU layer, a i The value can be set arbitrarily, for example, it can be 0.01.

[0087]

[0088] Batchnormalization: Normalize the data in each layer in the smallest batch, so that the data distribution becomes normal distribution, to prevent the data distribution from becoming more and more biased, which will lead to the disappearance of gradient and the failure of training to converge. As shown in Formula 2, B is the size of the smallest batch of data, m is the size of the entire training set, μ B is the mean of the smallest batch, is the variance. For a d-dimensional input data (x1, x2, .. xd), as shown in Formula 3, the data of each input dimension is normalized. ε is an arbitrarily small constant. It is added to the denominator to prevent the denominator from being 0. The parameter γ in Formula 4 is (k) and β (k) Learned during training, y i It is the batchnormalization layer output.

[0089]

[0090]

[0091]

[0092] like Fig.10 The figure shows a comparison of the standard neural network structure and the neural network structure after Dropout. During the deep neural network training process, the Dropout layer temporarily discards neurons with a certain probability. The probability here is set to 0.5, so each minimum batch trains a different network. Multiple different networks can learn more information. Dropout forces the neural network to learn more powerful features. These features are very useful when combined with many different random subsets of other neurons, making the network more robust and preventing overfitting.

[0093] As an optional implementation, you can use python to retrieve data from a database that stores video data watched by users. The videos watched by users are used as sample media resources, and the video data is used as sample media information. The sample media information includes but is not limited to the user terminal type, video clarity, the freeze rate of the user's video, video bit rate data, and the user's video download speed. The above video clarity, video bit rate data, and user video download speed can be the average data of the user watching videos over a period of time. The average bandwidth of the user ≈ the average download speed / 8 / 1024 / 1024, so as to obtain the average bit rate of the user's video viewing, the average bandwidth of the user, and the video clarity. The above average bit rate, the average bandwidth of the user, the video clarity, and the freeze rate of the user's video viewing are used as training data for the model.

[0094] As an optional implementation, during the neural network training process, X of the training set is the network definition (including: standard definition, high definition, ultra-high definition, full high definition), user bit rate, and user bandwidth of the sample media resources.

[0095] Among them, the user bit rate and bandwidth can be normalized to the minimum and maximum values ​​by the following formula:

[0096]

[0097] The user bit rate and bandwidth can make the data dimension consistent, the model training convergence speed can be accelerated, and the network clarity can be converted into a numerical vector using one-hot encoding.

[0098] The Y of the training set is the user freeze rate, ranging from 0 to 1. Fig. 9 In the model architecture of the neural network shown, data is input for training. Here, the dimension of the training set X is 200-dimensional video album word vector + 3-dimensional video clarity, video bit rate, user bandwidth = 203-dimensional training set data, and the training set Y is the user's freeze rate.

[0099] After the model is input, the first hidden layer is a fully connected hidden layer of 1024 neurons, using leakyReLU activation function + batchNormalization layer + dropout layer, where the dropout probability is set to 0.4; then the second hidden layer is a fully connected layer of 512 neurons, where the activation function and batch normalization are the same as above, and the dropout probability is set to 0.3; the third hidden layer is a fully connected layer of 512 neurons, where the activation function and batch normalization are the same as above, and the dropout probability is set to 0.3; the fourth hidden layer is a fully connected layer of 256 neurons, where the activation function and batch normalization are the same as above, and the dropout probability is set to 0.2; the fifth hidden layer is a fully connected layer of 256 neurons, where the activation function is leakyReLU, and the dropout probability is set to 0.01; the sixth hidden layer is a fully connected layer of 128 neurons, where the activation function is leakyReLU, and the dropout probability is set to 0.05; finally, the output layer is a neuron, using a linear activation function, and outputs the predicted jamming rate. Fig.11 It is the training loss curve. When the training reaches 9000 times, the model has converged. The model parameters at this time are saved to obtain the target jam rate prediction model. By loading the trained standard jam rate prediction model, inputting the bit rate and definition of the newly transcoded video, and taking the average bandwidth of the current network users in the past 10 days as the user bandwidth (because it is generally believed that the user bandwidth will not change dramatically in the short term), the jam rate predicted by the model is output.

[0100] Optionally, after determining the first bitrate threshold of the target media resource under the first clarity and the first network bandwidth, the method further includes: obtaining a second group of media information corresponding to the set of bitrates of the target media resource, wherein each piece of media information in the second group of media information includes a resource identification vector of the target media resource, a second clarity of the target media resource, a bitrate corresponding to the set of bitrates, and the first network bandwidth of the target media resource; inputting the media information in the second group of media information into the target stuttering rate prediction model in sequence to obtain a predicted stuttering rate corresponding to the media information in the second group of media information, wherein each piece of media information corresponding to the set of bitrates is a second clarity of the target media resource, a bitrate corresponding to the set of bitrates, and the first network bandwidth of the target media resource. The predicted stuttering rate corresponding to the media information in the second group of media information is used to indicate the stuttering rate when the target media resource is played at the second clarity, a corresponding bit rate and the first network bandwidth, and the target stuttering rate prediction model is used to determine the predicted stuttering rate based on the resource identification vector, the second clarity, a corresponding bit rate and the first network bandwidth; based on the predicted stuttering rate corresponding to the media information in the second group of media information, determine the second bit rate threshold of the target media resource at the second clarity and the first network bandwidth, wherein the second bit rate threshold is used to indicate the threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0101] As an optional implementation, the second definition is different from the first definition, and the second definition may be greater than the first definition or less than the first definition.

[0102] The following is described by a specific embodiment:

[0103] Assuming that the above-mentioned set of bit rates is 100, 150, and 200, the above-mentioned second definition is 2160p, and the above-mentioned first network bandwidth is 1000 megabits, then the above-mentioned second set of media information includes media information 4 (bit rate 100, definition 2160p, network bandwidth 1000 megabits), media information 5 (bit rate 150, definition 2160p, network bandwidth 1000 megabits), and media information 6 (bit rate 200, definition 2160, network bandwidth 1000 megabits).

[0104] The media information 3, media information 4, and media information 5 in the second group of media are input into the target stutter rate prediction model in sequence, and the stutter rate of the media information is estimated by the target stutter rate model to obtain the predicted stutter rate corresponding to the media information. The bitrate threshold of the target media resource under the conditions of 2160p definition and 1000 Mbps network bandwidth can be obtained through the stutter rate estimated by the neural network model and the user's viewing experience at different stutter rates.

[0105] Optionally, the method further includes: obtaining a third group of media information corresponding to the set of bit rates of the target media resource, wherein each piece of media information in the third group of media information includes a resource identification vector of the target media resource, the first clarity of the target media resource, a bit rate corresponding to the set of bit rates, and a second network bandwidth of the target media resource; inputting the media information in the third group of media information into a target stuttering rate prediction model in sequence to obtain a predicted stuttering rate corresponding to the media information in the third group of media information, wherein each piece of predicted stuttering rate corresponding to the media information in the third group of media information is The stuttering rate is used to indicate the stuttering rate when the target media resource is played at the first clarity, a corresponding bit rate and the second network bandwidth. The target stuttering rate prediction model is used to determine the predicted stuttering rate based on the resource identification vector, the first clarity, a corresponding bit rate and the second network bandwidth. According to the predicted stuttering rate corresponding to the media information in the third group of media information, a third bit rate threshold of the target media resource at the first clarity and the second network bandwidth is determined, wherein the third bit rate threshold is used to indicate the threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0106] As an optional implementation manner, the second network bandwidth is different from the first network bandwidth, and the second network bandwidth may be greater than or less than the first network bandwidth.

[0107] As an optional implementation, assuming that the above-mentioned set of bit rates is 100, 150, and 200, the above-mentioned first definition is 1080p, and the above-mentioned second network bandwidth is 2000 megabytes, then the above-mentioned third set of media information includes media information 7 (bit rate 100, definition 1080p, network bandwidth 2000 megabytes), media information 8 (bit rate 150, definition 1080p, network bandwidth 2000 megabytes), and media information 9 (bit rate 200, definition 1080, network bandwidth 2000 megabytes).

[0108] The media information 7, media information 8, and media information 9 in the second group of media are input into the target stutter rate prediction model in sequence, and the stutter rate of the media information is estimated by the target stutter rate model to obtain the predicted stutter rate corresponding to the media information. The bitrate threshold of the target media resource under the conditions of 1080p definition and 2000 Mbps network bandwidth can be obtained through the stutter rate estimated by the neural network model and the user's viewing experience at different stutter rates.

[0109] Optionally, the method further includes: obtaining a fourth group of media information corresponding to the set of bit rates of the target media resource, wherein each media information in the fourth group of media information includes a resource identification vector of the target media resource, a third clarity of the target media resource, a bit rate corresponding to the set of bit rates, and a third network bandwidth of the target media resource; inputting the media information in the fourth group of media information into a target stuttering rate prediction model in sequence to obtain a predicted stuttering rate corresponding to the media information in the fourth group of media information, wherein each predicted stuttering rate corresponding to the media information in the fourth group of media information The rate is used to represent the stuttering rate when the target media resource is played under the third clarity, a corresponding bit rate and the third network bandwidth, and the target stuttering rate prediction model is used to determine the predicted stuttering rate according to the resource identification vector, the third clarity, a corresponding bit rate and the third network bandwidth; according to the predicted stuttering rate corresponding to the media information in the fourth group of media information, determine a fourth bit rate threshold of the target media resource under the third clarity and the third network bandwidth, wherein the fourth bit rate threshold is used to represent the threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0110] As an optional implementation, the third network bandwidth is different from the first network bandwidth, and the third network bandwidth can be greater than or less than the first network bandwidth. The third definition is different from the first definition, and the third definition can be greater than or less than the first definition.

[0111] As an optional implementation, assuming that the above-mentioned set of bit rates is 100, 150, and 200, the above-mentioned third definition is 2160p, and the above-mentioned second network bandwidth is 2000 megabytes, then the above-mentioned second set of media information includes media information 10 (bit rate 100, definition 2160p, network bandwidth 2000 megabytes), media information 11 (bit rate 150, definition 2160p, network bandwidth 2000 megabytes), and media information 12 (bit rate 200, definition 2160, network bandwidth 2000 megabytes).

[0112] The media information 10, media information 11, and media information 12 in the second group of media are input into the target stutter rate prediction model in sequence, and the stutter rate of the media information is estimated by the target stutter rate model to obtain the predicted stutter rate corresponding to the media information. The bitrate threshold of the target media resource under the conditions of 2160p definition and 2000 Mbps network bandwidth can be obtained through the stutter rate estimated by the neural network model and the user's viewing experience at different stutter rates.

[0113] Optionally, after determining the first bitrate threshold of the target media resource under the first definition and the first network bandwidth, the method further comprises: when the current bitrate of the target media resource is the original bitrate, the current definition of the target media resource is the first definition, and the current network bandwidth of the target media resource is the first network bandwidth, increasing the bitrate of the target media resource from the original bitrate to a target bitrate, wherein the original bitrate is less than the first bitrate threshold, and the target bitrate is less than or equal to the first bitrate threshold and greater than the original bitrate;

[0114] As an optional implementation, when the bitrate threshold of the target media resource at the first definition and the first network bandwidth is obtained, the bitrate of the target media resource can be adjusted according to the bitrate threshold. Assume that the current definition of the target media resource is 1080p, the current network bandwidth is 1000 megabits, and the original bitrate is 30. If the bitrate threshold of the target media resource obtained by the target stuttering rate model in the above embodiment is 100 when the definition is 1080p and the network bandwidth is 1000 megabits, the bitrate of the target media resource can be increased to any value between 30 and 100, that is, the target bitrate is any value between 30 and 100.

[0115] Alternatively, when the current bit rate of the target media resource is the original bit rate, the current definition of the target media resource is the first definition, and the difference between the current network bandwidth of the target media resource and the first network bandwidth meets a preset threshold, the bit rate of the target media resource is increased from the original bit rate to the target bit rate, wherein the original bit rate is less than the first bit rate threshold, and the target bit rate is less than or equal to the first bit rate threshold and greater than the original bit rate.

[0116] As an optional implementation, the above preset threshold can be set according to actual conditions, such as 100, 200, etc. Assume that the current definition of the target media resource is 1080p, the current network bandwidth is 1100 megabits, and the original bit rate is 30. If the bit rate threshold of the target media resource obtained by the target stuttering rate model in the above embodiment is 100 when the definition is 1080p and the network bandwidth is 1000 megabits. The difference between the current network bandwidth 1100 and 1000 of the target media resource is 100. Assuming that the above preset threshold is 200, the difference 100 is less than the preset threshold 200. Then the bit rate of the target media resource can be increased to any value between 30 and 100, that is, the target bit rate is any value between 30 and 100.

[0117] Optionally, after determining the first bitrate threshold of the target media resource under the first definition and the first network bandwidth, the method further includes: when the current bitrate of the target media resource is the original bitrate, the current definition of the target media resource is a target definition different from the first definition, and the current network bandwidth of the target media resource is a target network bandwidth different from the first network bandwidth, obtaining a target bitrate threshold of the target media resource under the target definition and the target network bandwidth, wherein the target bitrate threshold is a bitrate threshold determined according to a target predicted stuttering rate, and the target predicted stuttering rate is a predicted stuttering rate determined by the target stuttering rate prediction model according to the resource identification vector, the target definition, a corresponding bitrate in the set of bitrates, and the target network bandwidth; and increasing the bitrate of the target media resource from the original bitrate to the target bitrate, wherein the original bitrate is less than the target bitrate threshold, and the target bitrate is less than or equal to the target bitrate threshold and greater than the original bitrate.

[0118] As an optional implementation, assume that the current bit rate of the target media resource is 30, the current definition is 2160p, and the current network bandwidth is 2000 Mbps. The first definition is 1080, and the first network bandwidth is 1000 Mbps. The current definition of the target media resource is different from the first definition, and the current network bandwidth of the target media resource is different from the first network bandwidth.

[0119] Assuming that the bitrate threshold is 150 when the definition is 2160p and the network bandwidth is 2000 Mbps, the current bitrate of the target media resource can be increased to any value between 30 and 150.

[0120] Through the above embodiments, a freeze rate prediction model is constructed through a neural network model, and the freeze rate of the video watched by the user can be predicted according to the video clarity, bit rate, bandwidth, and video name. Video renovation can be applied to estimate the impact of the bit rate increase on the user's playback experience when converting from a low-bit rate blurry, less colorful, and dark video to a high-bit rate high-definition, richly colored, and better visual video, thereby defining an increased bit rate threshold to prevent excessive impact on the user experience.

[0121] It is understandable that in the specific implementation of this application, related data such as user information is involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0122] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0123] According to another aspect of the embodiments of the present invention, a device for determining a bit rate threshold for implementing the method for determining a bit rate threshold is also provided. Fig.12 As shown, the device includes: an acquisition module 1202, which is used to acquire a first group of media information corresponding to a group of bit rates of a target media resource, wherein each media information in the first group of media information includes a resource identification vector of the target media resource, a first definition of the target media resource, a corresponding bit rate in the group of bit rates, and a first network bandwidth of the target media resource; an input module 1204, which is used to input the media information in the first group of media information into a target stuttering rate prediction model in sequence, and obtain a predicted stuttering rate corresponding to the input media information, wherein each predicted stuttering rate is used to represent the first network bandwidth of the target media resource; A first definition, a corresponding bit rate and a stuttering rate when the target media resource is played under the first network bandwidth, the target stuttering rate prediction model is used to determine the predicted stuttering rate according to the resource identification vector, the first definition, a corresponding bit rate and the first network bandwidth; a determination module 1206 is used to determine a first bit rate threshold of the target media resource under the first definition and the first network bandwidth according to the predicted stuttering rate corresponding to the input media information, wherein the first bit rate threshold is used to indicate a threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0124] Optionally, the above-mentioned device is also used to input the media information in the first group of media information into the target stuttering rate prediction model in sequence according to the order of the bit rates in the group of bit rates from small to large, and obtain the predicted stuttering rate corresponding to the input media information.

[0125] Optionally, the above-mentioned device is also used to obtain the predicted stuttering rates obtained in sequence by the target stuttering rate prediction model; when it is detected that the predicted stuttering rate currently obtained is greater than a preset stuttering rate threshold, and the predicted stuttering rate previously obtained before the current time is less than or equal to the stuttering rate threshold, the target bit rate corresponding to the predicted stuttering rate previously obtained is determined as the first bit rate threshold of the target media resource, wherein the predicted stuttering rate previously obtained is the predicted stuttering rate determined by the target stuttering rate prediction model based on the resource identification vector, the first clarity, the target bit rate and the first network bandwidth.

[0126] Optionally, the above-mentioned device is also used to input the media information in the first group of media information into the target stuttering rate prediction model in sequence according to the order of the bit rates in the group of bit rates from large to small, and obtain the predicted stuttering rate corresponding to the input media information.

[0127] Optionally, the above-mentioned device is also used to obtain the predicted stuttering rates obtained in sequence by the target stuttering rate prediction model; when it is detected that the predicted stuttering rate currently obtained is less than or equal to a preset stuttering rate threshold, and the predicted stuttering rate previously obtained before the current time is greater than the stuttering rate threshold, the target bit rate corresponding to the predicted stuttering rate currently obtained is determined as the first bit rate threshold of the target media resource, wherein the predicted stuttering rate currently obtained is the predicted stuttering rate determined by the target stuttering rate prediction model based on the resource identification vector, the first clarity, the target bit rate and the first network bandwidth.

[0128] Optionally, the above-mentioned device is also used to obtain sample media information of a sample media resource and an actual sample stuttering rate of the sample media resource before inputting the media information in the first group of media information into the target stuttering rate prediction model in sequence, wherein the sample media information includes a resource identification vector of the sample media resource, the clarity of the sample media resource, the bit rate of the sample media resource and the network bandwidth of the sample media resource; train the original stuttering rate prediction model through the sample media information and the actual sample stuttering rate, and during the training process, if the loss value between the predicted sample stuttering rate determined by the original stuttering rate prediction model according to the sample media resource and the actual sample stuttering rate does not meet the preset convergence condition, adjust the parameters in the original stuttering rate prediction model, and if the loss value meets the preset convergence condition, end the training, and determine the original stuttering rate prediction model at the end of the training as the target stuttering rate prediction model.

[0129] Optionally, the above-mentioned device is also used to perform multiple rounds of iterative training on the original jam rate prediction model through the sample media information and the actual sample jam rate, wherein i is a positive integer greater than or equal to 1, and the original jam rate prediction model determined in the 0th round of training is the untrained original jam rate prediction model, and the i-th round of training in the multiple rounds of iterative training includes: inputting the sample media information into the original jam rate prediction model determined in the i-1th round of training to obtain the predicted sample jam rate determined in the i-th round of training, wherein the original jam rate prediction model includes multiple hidden layers, each of the hidden layers includes a preset number of neurons, and each of the hidden layers includes Dense layer, LeakyReLU activation function layer, Batchnormalization layer and Dropout layer; if the loss value between the predicted sample stuttering rate determined by the i-th round of training and the actual sample stuttering rate does not meet the preset convergence condition, adjust the parameters in the original stuttering rate prediction model obtained by the i-1-th round of training to obtain the original stuttering rate prediction model determined by the i-th round of training; if the loss value between the predicted sample stuttering rate determined by the i-th round of training and the actual sample stuttering rate meets the preset convergence condition, determine the original stuttering rate prediction model obtained by the i-1-th round of training as the target stuttering rate prediction model.

[0130] Optionally, the above-mentioned device is also used to obtain a second group of media information corresponding to the set of bit rates of the target media resource after determining the first bit rate threshold of the target media resource under the first clarity and the first network bandwidth, wherein each media information in the second group of media information includes a resource identification vector of the target media resource, the second clarity of the target media resource, a bit rate corresponding to the set of bit rates, and the first network bandwidth of the target media resource; the media information in the second group of media information is sequentially input into the target stuttering rate prediction model to obtain the predicted stuttering rate corresponding to the media information in the second group of media information, wherein each media information corresponding to the set of bit rates is a second bit rate threshold of the target media resource under the first clarity and the first network bandwidth; The predicted stuttering rate corresponding to the media information in the second group of media information is used to indicate the stuttering rate when the target media resource is played at the second clarity, a corresponding bit rate and the first network bandwidth, and the target stuttering rate prediction model is used to determine the predicted stuttering rate based on the resource identification vector, the second clarity, a corresponding bit rate and the first network bandwidth; based on the predicted stuttering rate corresponding to the media information in the second group of media information, determine the second bit rate threshold of the target media resource at the second clarity and the first network bandwidth, wherein the second bit rate threshold is used to indicate the threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0131] Optionally, the above-mentioned device is also used to obtain a third group of media information corresponding to the set of bit rates of the target media resource, wherein each media information in the third group of media information includes a resource identification vector of the target media resource, the first clarity of the target media resource, a bit rate corresponding to the set of bit rates, and a second network bandwidth of the target media resource; the media information in the third group of media information is sequentially input into the target stuttering rate prediction model to obtain a predicted stuttering rate corresponding to the media information in the third group of media information, wherein each predicted stuttering rate corresponding to the media information in the third group of media information The rate is used to represent the stuttering rate when the target media resource is played at the first clarity, a corresponding bit rate and the second network bandwidth, and the target stuttering rate prediction model is used to determine the predicted stuttering rate according to the resource identification vector, the first clarity, a corresponding bit rate and the second network bandwidth; according to the predicted stuttering rate corresponding to the media information in the third group of media information, determine the third bit rate threshold of the target media resource at the first clarity and the second network bandwidth, wherein the third bit rate threshold is used to represent the threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0132] Optionally, the above-mentioned device is also used to obtain a fourth group of media information corresponding to the set of bit rates of the target media resource, wherein each media information in the fourth group of media information includes a resource identification vector of the target media resource, a third clarity of the target media resource, a bit rate corresponding to the set of bit rates, and a third network bandwidth of the target media resource; the media information in the fourth group of media information is sequentially input into the target stuttering rate prediction model to obtain the predicted stuttering rate corresponding to the media information in the fourth group of media information, wherein each predicted stuttering rate corresponding to the media information in the fourth group of media information Used to represent the stuttering rate when the target media resource is played at the third clarity, a corresponding bit rate and the third network bandwidth, the target stuttering rate prediction model is used to determine the predicted stuttering rate according to the resource identification vector, the third clarity, a corresponding bit rate and the third network bandwidth; according to the predicted stuttering rate corresponding to the media information in the fourth group of media information, determine a fourth bit rate threshold of the target media resource at the third clarity and the third network bandwidth, wherein the fourth bit rate threshold is used to represent the threshold allowed to be adjusted when adjusting the bit rate of the target media resource.

[0133] Optionally, the above-mentioned device is also used for, after determining the first bit rate threshold of the target media resource under the first definition and the first network bandwidth, when the current bit rate of the target media resource is the original bit rate, the current definition of the target media resource is the first definition, and the current network bandwidth of the target media resource is the first network bandwidth, adjusting the bit rate of the target media resource from the original bit rate to the target bit rate, wherein the original bit rate is less than the first bit rate threshold, the target bit rate is less than or equal to the first bit rate threshold, and is greater than the original bit rate; when the current bit rate of the target media resource is the original bit rate, the current definition of the target media resource is the first definition, and the difference between the current network bandwidth of the target media resource and the first network bandwidth meets a preset threshold, adjusting the bit rate of the target media resource from the original bit rate to the target bit rate, wherein the original bit rate is less than the first bit rate threshold, the target bit rate is less than or equal to the first bit rate threshold, and is greater than the original bit rate.

[0134] Optionally, the above-mentioned device is also used to, after determining the first bit rate threshold of the target media resource under the first definition and the first network bandwidth, when the current bit rate of the target media resource is the original bit rate, the current definition of the target media resource is the target definition different from the first definition, and the current network bandwidth of the target media resource is the target network bandwidth different from the first network bandwidth, obtain the target bit rate threshold of the target media resource under the target definition and the target network bandwidth, wherein the target bit rate threshold is a bit rate threshold determined according to a target predicted stuttering rate, and the target predicted stuttering rate is a predicted stuttering rate determined by the target stuttering rate prediction model according to the resource identification vector, the target definition, the corresponding bit rate in the set of bit rates, and the target network bandwidth; and increase the bit rate of the target media resource from the original bit rate to the target bit rate, wherein the original bit rate is less than the target bit rate threshold, and the target bit rate is less than or equal to the target bit rate threshold and greater than the original bit rate.

[0135] According to another aspect of the embodiments of the present invention, an electronic device for implementing the above-mentioned method for determining the bit rate threshold is also provided. The electronic device may be Figure 1 The user device or server shown in the figure. This embodiment is described by taking the electronic device as a server as an example. Fig.13 As shown, the electronic device includes a memory 1302 and a processor 1304. The memory 1302 stores a computer program, and the processor 1304 is configured to execute the steps in any of the above method embodiments through the computer program.

[0136] Optionally, in this embodiment, the electronic device may be located in at least one network device among a plurality of network devices of a computer network.

[0137] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0138] S1, obtaining a first set of media information corresponding to a set of bit rates of a target media resource, wherein each piece of media information in the first set of media information includes a resource identification vector of the target media resource, a first definition of the target media resource, a bit rate corresponding to the set of bit rates, and a first network bandwidth of the target media resource;

[0139] S2, inputting the media information in the first group of media information into the target jamming rate prediction model in sequence, and obtaining the predicted jamming rate corresponding to the input media information, wherein each predicted jamming rate is used to represent the jamming rate when the target media resource is played under the first definition, a corresponding bit rate and the first network bandwidth, and the target jamming rate prediction model is used to determine the predicted jamming rate according to the resource identification vector, the first definition, a corresponding bit rate and the first network bandwidth;

[0140] S3. Determine a first bitrate threshold of the target media resource under the first clarity and the first network bandwidth based on the predicted stuttering rate corresponding to the input media information, wherein the first bitrate threshold is used to indicate a threshold allowed to be adjusted when adjusting the bitrate of the target media resource.

[0141] Alternatively, a person skilled in the art may understand that: Fig.13 The structure shown is for illustration only. Fig.13 The electronic device and the electronic equipment described above are not limited in structure. Fig.13 More or fewer components (such as network interfaces, etc.) as shown in, or with Fig.13 Different configurations are shown.

[0142] Among them, the memory 1302 can be used to store software programs and modules, such as the program instructions / modules corresponding to the method and device for determining the bit rate threshold in the embodiment of the present invention. The processor 1304 executes various functional applications and data processing by running the software programs and modules stored in the memory 1302, that is, the above-mentioned method for determining the bit rate threshold is realized. The memory 1302 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1302 may further include a memory remotely arranged relative to the processor 1304, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. Among them, the memory 1302 may be specifically, but not limited to, used to store information such as target media resources. As an example, such as Fig.13 As shown, the memory 1302 may include, but is not limited to, the acquisition module 1202, input module 1204, and determination module 1206 in the bit rate threshold determination device. In addition, other module units in the bit rate threshold determination device may also be included but are not limited to, which will not be repeated in this example.

[0143] Optionally, the transmission device 1306 is used to receive or send data via a network. Specific examples of the network may include a wired network and a wireless network. In one example, the transmission device 1306 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices and routers via a network cable so as to communicate with the Internet or a local area network. In one example, the transmission device 1306 is a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0144] In addition, the electronic device mentioned above further includes: a connection bus 1308, which is used to connect various module components in the electronic device mentioned above.

[0145] In other embodiments, the user device or server may be a node in a distributed system, wherein the distributed system may be a blockchain system, and the blockchain system may be a distributed system formed by connecting the multiple nodes in the form of network communication. Among them, the nodes may form a peer-to-peer (P2P) network, and any form of computing device, such as a server, terminal or other electronic device, may become a node in the blockchain system by joining the peer-to-peer network.

[0146] According to one aspect of the present application, a computer program product is provided, the computer program product comprising a computer program / instruction, the computer program / instruction comprising a program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium. When the computer program is executed by a central processing unit, various functions provided by the embodiments of the present application are executed.

[0147] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0148] Fig.14 The structure block diagram of a computer system for implementing an electronic device according to an embodiment of the present application is schematically shown.

[0149] It should be noted that Fig.14 The computer system 1400 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0150] like Fig.14 As shown, the computer system 1400 includes a central processing unit 1401 (CPU), which can perform various appropriate actions and processes according to the program stored in the read-only memory 1402 (ROM) or the program loaded from the storage part 1408 to the random access memory 1403 (RAM). Various programs and data required for system operation are also stored in the random access memory 1403. The central processing unit 1401, the read-only memory 1402 and the random access memory 1403 are connected to each other through a bus 1404. An input / output interface 1405 (Input / Output interface, i.e., I / O interface) is also connected to the bus 1404.

[0151] The following components are connected to the input / output interface 1405: an input section 1406 including a keyboard, a mouse, etc.; an output section 1407 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1408 including a hard disk, etc.; and a communication section 1409 including a network interface card such as a LAN card, a modem, etc. The communication section 1409 performs communication processing via a network such as the Internet. A drive 1410 is also connected to the input / output interface 1405 as needed. A removable medium 1411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1410 as needed so that a computer program read therefrom is installed into the storage section 1408 as needed.

[0152] In particular, according to an embodiment of the present application, the process described in each method flow chart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, and the computer program contains a program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1409, and / or installed from the removable medium 1411. When the computer program is executed by the central processor 1401, various functions defined in the system of the present application are executed.

[0153] According to one aspect of the present application, a computer-readable storage medium is provided, and a processor of a computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device executes the methods provided in the above-mentioned various optional implementations.

[0154] Optionally, in this embodiment, the computer-readable storage medium may be configured to store a computer program for performing the following steps:

[0155] S1, obtaining a first set of media information corresponding to a set of bit rates of a target media resource, wherein each piece of media information in the first set of media information includes a resource identification vector of the target media resource, a first definition of the target media resource, a bit rate corresponding to the set of bit rates, and a first network bandwidth of the target media resource;

[0156] S2, inputting the media information in the first group of media information into the target jamming rate prediction model in sequence, and obtaining the predicted jamming rate corresponding to the input media information, wherein each predicted jamming rate is used to represent the jamming rate when the target media resource is played under the first definition, a corresponding bit rate and the first network bandwidth, and the target jamming rate prediction model is used to determine the predicted jamming rate according to the resource identification vector, the first definition, a corresponding bit rate and the first network bandwidth;

[0157] S3. Determine a first bitrate threshold of the target media resource under the first clarity and the first network bandwidth based on the predicted stuttering rate corresponding to the input media information, wherein the first bitrate threshold is used to indicate a threshold allowed to be adjusted when adjusting the bitrate of the target media resource.

[0158] Optionally, in this embodiment, a person of ordinary skill in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing hardware related to the terminal device through a program, and the program may be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0159] If the integrated units in the above embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling one or more computer devices (which can be personal computers, servers or network devices, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention.

[0160] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0161] In the several embodiments provided in the present application, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0162] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0163] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0164] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for determining a bit rate threshold, characterized in that: include: Acquire sample media information of a sample media resource and an actual sample freeze rate of the sample media resource, wherein the sample media information includes a resource identification vector of the sample media resource, the definition of the sample media resource, the bit rate of the sample media resource, and the network bandwidth of the sample media resource, and the resource identification vector of the sample media resource is used to represent the name of the sample media resource and words related to the name of the sample media resource; Building an original jam rate prediction model, wherein the original jam rate prediction model includes multiple hidden layers, each of which includes a regularization layer and a batch normalization layer, wherein the batch normalization layer is used to perform minimum batch normalization processing on the features input into the hidden layer, and the regularization layer is used to learn different combination relationships between features processed by the batch normalization layer; According to the sample media information and the actual sample jamming rate, the original jamming rate prediction model is trained, wherein the original jamming rate prediction model predicts the predicted sample jamming rate corresponding to the sample media resource by learning the relationship between the resource identification vector of the sample media resource, the definition of the sample media resource, the bit rate of the sample media resource and the network bandwidth of the sample media resource; Determine the original jam rate prediction model at the end of training as the target jam rate prediction model; Acquire a first set of media information corresponding to a set of bit rates of a target media resource, wherein each piece of media information in the first set of media information includes a resource identification vector of the target media resource, a first definition of the target media resource, a corresponding bit rate in the set of bit rates, and a first network bandwidth of the target media resource; Inputting media information in the first group of media information into the target jamming rate prediction model in sequence, and obtaining predicted jamming rates corresponding to the media information in the first group of media information, wherein each predicted jamming rate is used to represent a jamming rate when the target media resource is played at the first definition, a corresponding bit rate, and the first network bandwidth, and the target jamming rate prediction model is used to determine the predicted jamming rate according to the resource identification vector, the first definition, a corresponding bit rate, and the first network bandwidth; Based on the predicted stuttering rate corresponding to the input media information, a first bitrate threshold of the target media resource under the first clarity and the first network bandwidth is determined, wherein the first bitrate threshold is used to represent a threshold allowed to be adjusted when adjusting the bitrate of the target media resource.

2. The method according to claim 1, characterized in that The step of sequentially inputting the media information in the first group of media information into the target jam rate prediction model to obtain a predicted jam rate corresponding to the input media information includes: Inputting the media information in the first group of media information into the target jamming rate prediction model in order from the smallest to the largest bitrate in the group of bitrates, to obtain a predicted jamming rate corresponding to the input media information; The method of determining a first bitrate threshold of the target media resource at the first clarity and the first network bandwidth based on the predicted stuttering rate corresponding to the media information in the first group of media information input includes: obtaining predicted stuttering rates obtained in sequence by the target stuttering rate prediction model; and when it is detected that the predicted stuttering rate obtained currently is greater than a preset stuttering rate threshold, and the predicted stuttering rate obtained last time before the current time is less than or equal to the stuttering rate threshold, determining the target bitrate corresponding to the predicted stuttering rate obtained last time as the first bitrate threshold of the target media resource, wherein the predicted stuttering rate obtained last time is the predicted stuttering rate determined by the target stuttering rate prediction model according to the resource identification vector, the first clarity, the target bitrate and the first network bandwidth.

3. The method according to claim 1, characterized in that The step of sequentially inputting the media information in the first group of media information into the target jamming rate prediction model to obtain a predicted jamming rate corresponding to the media information in the first group of media information includes: sequentially inputting the media information in the first group of media information into the target jamming rate prediction model in descending order of the bit rates in the group of bit rates to obtain a predicted jamming rate corresponding to the input media information; The determining, based on the predicted stuttering rate corresponding to the media information in the first group of media information input, the first bitrate threshold of the target media resource under the first clarity and the first network bandwidth includes: obtaining the predicted stuttering rates obtained in sequence by the target stuttering rate prediction model; when it is detected that the predicted stuttering rate obtained currently is less than or equal to a preset stuttering rate threshold, and the predicted stuttering rate obtained last time before the current time is greater than the stuttering rate threshold, determining the target bitrate corresponding to the predicted stuttering rate obtained currently as the first bitrate threshold of the target media resource, wherein the predicted stuttering rate obtained currently is the predicted stuttering rate determined by the target stuttering rate prediction model according to the resource identification vector, the first clarity, the target bitrate and the first network bandwidth.

4. The method according to claim 1, characterized in that The training of the original jam rate prediction model according to the sample media information and the actual sample jam rate includes: The original stuttering rate prediction model is trained using the sample media information and the actual sample stuttering rate. During the training process, if a loss value between the predicted sample stuttering rate and the actual sample stuttering rate determined by the original stuttering rate prediction model based on the sample media resource does not meet a preset convergence condition, the parameters in the original stuttering rate prediction model are adjusted. If the loss value meets the preset convergence condition, the training is terminated.

5. The method according to claim 4, characterized in that The training of the original jam rate prediction model using the sample media information and the actual sample jam rate includes: The original jam rate prediction model is trained for multiple rounds of iterations using the sample media information and the actual sample jam rate, wherein i is a positive integer greater than or equal to 1, the original jam rate prediction model determined in the 0th round of training is the untrained original jam rate prediction model, and the i-th round of training in the multiple rounds of iterative training includes: Inputting the sample media information into the original jam rate prediction model determined by the i-1th round of training to obtain the predicted sample jam rate determined by the i-th round of training, wherein the hidden layer included in each of the original jam rate prediction models includes a preset number of neurons, and each of the hidden layers includes a Dense layer and a LeakyReLU activation function layer; If the loss value between the predicted sample jamming rate determined by the i-th round of training and the actual sample jamming rate does not meet the preset convergence condition, adjusting the parameters in the original jamming rate prediction model obtained by the i-1-th round of training to obtain the original jamming rate prediction model determined by the i-th round of training; If the loss value between the predicted sample jamming rate and the actual sample jamming rate determined in the i-th round of training meets the preset convergence condition, the training is terminated.

6. The method according to claim 1, characterized in that After determining a first bit rate threshold of the target media resource under the first definition and the first network bandwidth, the method further includes: Acquire a second set of media information corresponding to the set of bit rates of the target media resource, wherein each piece of media information in the second set of media information includes a resource identification vector of the target media resource, a second definition of the target media resource, a corresponding bit rate in the set of bit rates, and the first network bandwidth of the target media resource; Inputting the media information in the second group of media information into the target jamming rate prediction model in sequence to obtain the predicted jamming rate corresponding to the media information in the second group of media information, wherein each predicted jamming rate corresponding to the media information in the second group of media information is used to represent the jamming rate when the target media resource is played under the second definition, a corresponding bit rate and the first network bandwidth, and the target jamming rate prediction model is used to determine the predicted jamming rate according to the resource identification vector, the second definition, a corresponding bit rate and the first network bandwidth; Based on the predicted stuttering rate corresponding to the media information in the second group of media information, determine a second bitrate threshold of the target media resource under the second clarity and the first network bandwidth, wherein the second bitrate threshold is used to indicate a threshold allowed to be adjusted when adjusting the bitrate of the target media resource.

7. The method according to claim 1, characterized in that The method further comprises: Acquire a third group of media information corresponding to the set of bit rates of the target media resource, wherein each piece of media information in the third group of media information includes a resource identification vector of the target media resource, the first definition of the target media resource, a corresponding bit rate in the set of bit rates, and a second network bandwidth of the target media resource; Inputting the media information in the third group of media information into the target jamming rate prediction model in sequence to obtain the predicted jamming rate corresponding to the media information in the third group of media information, wherein each predicted jamming rate corresponding to the media information in the third group of media information is used to represent the jamming rate when the target media resource is played under the first definition, a corresponding bit rate and the second network bandwidth, and the target jamming rate prediction model is used to determine the predicted jamming rate according to the resource identification vector, the first definition, a corresponding bit rate and the second network bandwidth; Based on the predicted stuttering rate corresponding to the media information in the third group of media information, determine a third bitrate threshold of the target media resource under the first clarity and the second network bandwidth, wherein the third bitrate threshold is used to indicate a threshold allowed to be adjusted when adjusting the bitrate of the target media resource.

8. The method according to claim 1, characterized in that The method further comprises: Acquire a fourth group of media information corresponding to the set of bit rates of the target media resource, wherein each piece of media information in the fourth group of media information includes a resource identification vector of the target media resource, a third definition of the target media resource, a corresponding bit rate in the set of bit rates, and a third network bandwidth of the target media resource; Inputting the media information in the fourth group of media information into the target stuttering rate prediction model in sequence to obtain the predicted stuttering rate corresponding to the media information in the fourth group of media information, wherein each predicted stuttering rate corresponding to the media information in the fourth group of media information is used to represent the stuttering rate when the target media resource is played under the third definition, a corresponding bit rate and the third network bandwidth, and the target stuttering rate prediction model is used to determine the predicted stuttering rate according to the resource identification vector, the third definition, a corresponding bit rate and the third network bandwidth; Based on the predicted stuttering rate corresponding to the media information in the fourth group of media information, determine a fourth bitrate threshold of the target media resource at the third clarity and the third network bandwidth, wherein the fourth bitrate threshold is used to indicate a threshold allowed to be adjusted when adjusting the bitrate of the target media resource.

9. The method according to any one of claims 1 to 8, characterized in that After determining a first bitrate threshold of the target media resource under the first definition and the first network bandwidth, the method further includes: When the current bit rate of the target media resource is the original bit rate, the current definition of the target media resource is the first definition, and the current network bandwidth of the target media resource is the first network bandwidth, the bit rate of the target media resource is increased from the original bit rate to the target bit rate, wherein the original bit rate is less than the first bit rate threshold, and the target bit rate is less than or equal to the first bit rate threshold and greater than the original bit rate; or When the current bit rate of the target media resource is the original bit rate, the current definition of the target media resource is the first definition, and the difference between the current network bandwidth of the target media resource and the first network bandwidth meets a preset threshold, the bit rate of the target media resource is increased from the original bit rate to a target bit rate, wherein the original bit rate is less than the first bit rate threshold, and the target bit rate is less than or equal to the first bit rate threshold and greater than the original bit rate.

10. The method according to any one of claims 1 to 8, characterized in that After determining a first bit rate threshold of the target media resource under the first definition and the first network bandwidth, the method further includes: In a case where the current bit rate of the target media resource is the original bit rate, the current definition of the target media resource is a target definition different from the first definition, and the current network bandwidth of the target media resource is a target network bandwidth different from the first network bandwidth, obtaining a target bit rate threshold of the target media resource at the target definition and the target network bandwidth, wherein the target bit rate threshold is a bit rate threshold determined according to a target predicted stuttering rate, and the target predicted stuttering rate is a predicted stuttering rate determined by the target stuttering rate prediction model according to the resource identification vector, the target definition, a bit rate corresponding to the set of bit rates, and the target network bandwidth; The bit rate of the target media resource is increased from the original bit rate to a target bit rate, wherein the original bit rate is less than the target bit rate threshold, and the target bit rate is less than or equal to the target bit rate threshold and greater than the original bit rate.

11. A device for determining a bit rate threshold, characterized in that: include: An acquisition module, configured to acquire a first set of media information corresponding to a set of bit rates of a target media resource, wherein each piece of media information in the first set of media information includes a resource identification vector of the target media resource, a first definition of the target media resource, a corresponding bit rate in the set of bit rates, and a first network bandwidth of the target media resource; An input module, used to input the media information in the first group of media information into a target jamming rate prediction model in sequence, to obtain a predicted jamming rate corresponding to the input media information, wherein each predicted jamming rate is used to represent a jamming rate when the target media resource is played at the first definition, a corresponding bit rate and the first network bandwidth, and the target jamming rate prediction model is used to determine the predicted jamming rate according to the resource identification vector, the first definition, a corresponding bit rate and the first network bandwidth; a determination module, configured to determine, according to the predicted freeze rate corresponding to the input media information, a first bitrate threshold of the target media resource under the first definition and the first network bandwidth, wherein the first bitrate threshold is used to indicate a threshold to which the bitrate of the target media resource is allowed to be adjusted; The device is also used to obtain sample media information of a sample media resource and an actual sample stuttering rate of the sample media resource before obtaining a first group of media information corresponding to a group of bit rates of a target media resource, wherein the sample media information includes a resource identification vector of the sample media resource, the clarity of the sample media resource, the bit rate of the sample media resource and the network bandwidth of the sample media resource, and the resource identification vector of the sample media resource is used to represent the name of the sample media resource and words related to the name of the sample media resource; and build an original stuttering rate prediction model, wherein the original stuttering rate prediction model includes multiple hidden layers, each of the hidden layers includes a regularization layer and a batch normalization layer. layer, the batch normalization layer is used to perform minimum batch normalization processing on the features input into the hidden layer, and the regularization layer is used to learn different combination relationships between the features processed by the batch normalization layer; according to the sample media information and the actual sample stuttering rate, the original stuttering rate prediction model is trained, and the original stuttering rate prediction model is performed by learning the relationship between the resource identification vector of the sample media resource, the clarity of the sample media resource, the bit rate of the sample media resource and the network bandwidth of the sample media resource to predict the predicted sample stuttering rate corresponding to the sample media resource; the original stuttering rate prediction model at the end of training is determined as the target stuttering rate prediction model.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method described in any one of claims 1 to 10 when executed.

13. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

14. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 10 through the computer program.

Images