A method for optimizing outbound voice communication quality based on intelligent AI

Abstract

The present invention relates to the field of intelligent communication technology, and specifically to an outbound voice communication quality optimization method based on intelligent AI, comprising: collecting audio parameters transmitted during outbound voice communication of a communication device, evaluating audio quality based on the audio parameters, optimizing the audio when the audio quality evaluation result is poor, synchronously capturing the instantaneous throughput of the network where the communication device is located, and predicting the audio transmission stability based on the instantaneous throughput of the network; obtaining the audio transmission stability prediction result, and when the prediction result is stable, the communication device performs an audio transmission operation. The present invention evaluates the audio quality to be transmitted by the communication device, decides whether to optimize the audio, and predicts the stability of the communication device when transmitting audio in combination with the instantaneous throughput of the communication network where the communication device is located, and further adaptively switches the communication device to the network where the audio communication data is executed based on the transmission stability prediction result to ensure that the communication device transmits audio stably.

Description

Technical Field

[0001] The present invention relates to the field of intelligent communication technology, and in particular to an outbound voice communication quality optimization method based on intelligent AI. Background Art

[0002] Voice communication refers to the way of exchanging information through sound. Common voice communication devices include mobile phones, voice walkie-talkies, etc.

[0003] The invention patent with application number 201410271058.X discloses a voice quality optimization method, which is characterized in that the method includes: when it is determined that the current channel quality is lower than the set index, selecting a corresponding adaptive multi-rate wideband coding AMR-WB coding mode for the information to be transmitted, and using the selected AMR-WB coding mode to encode the information to be transmitted; using multiple description coding MDC to convert the encoded information to be transmitted into two description information: adjusting the coding rate of the two description information so that the adjusted coding rate is consistent with the AMR-WB coding rate; sending the adjusted two description information; wherein, selecting a corresponding adaptive multi-rate wideband coding AMR-WB coding mode for the information to be transmitted includes: selecting an AMR-WB coding mode whose encoded bit rate is less than the current encoded bit rate for the information to be transmitted.

[0004] The application aims to solve the problem that: "When error concealment is used for encoding and decoding, under low packet loss rate, the error concealment algorithm of AMR-WB can ensure voice quality. However, under high packet loss rate, the effect of improving voice quality by using the error concealment algorithm of AMR-WB will be significantly reduced, and the loss of the current voice packet will also affect the recovery of the subsequent voice packet data;

[0005] When multiple descriptions are used for encoding and decoding, although voice quality can be guaranteed under high packet loss rates, the encoded parameters need to be described multiple times, so that some parameters are repeatedly included in each description, making the total encoding rate of the multiple description output greater than the encoding rate in the original encoding mode, and occupying more resources during transmission. In addition, the encoding rate output by the multiple description algorithm will not be the same as any of the nine original encoding rates of AMR-WB, and a new encoding rate will be generated. When it needs to be applied to an existing communication system, the network architecture of the communication system needs to be updated.

[0006] However, most communication devices do not have adaptive switching control logic for communication networks, resulting in the quality and fluency of communication being completely limited by the communication network they use. When the communication networks available to the communication devices are unstable, audio cannot be transmitted normally, resulting in the information to be conveyed by the audio being unable to be stably heard by the receiving end. Summary of the invention

[0007] In view of the above-mentioned shortcomings of the prior art, the present invention provides an outbound voice communication quality optimization method based on intelligent AI, which solves the technical problems raised in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] An outbound voice communication quality optimization method based on intelligent AI, comprising:

[0010] Collect transmission audio parameters of the communication device during outbound voice communication, evaluate audio quality based on audio parameters, optimize audio when the audio quality evaluation result is poor, synchronously capture the instantaneous throughput of the network where the communication device is located, and predict audio transmission stability based on the instantaneous throughput of the network; obtain audio transmission stability prediction results, when the prediction result is stable, the communication device performs audio transmission operations, when the prediction result is unstable, control the communication device to switch the communication network, and jump again to the stage of capturing the instantaneous throughput of the network where the communication device is located, and predicting audio transmission stability based on the instantaneous throughput of the network, until the prediction result is stable, the communication device uses the currently switched communication network to perform audio transmission operations; when the corresponding audio transmission stability evaluation results of all switched communication networks are negative, obtain audio, convert the audio into text, recognize the pinyin and tone of the text, and further convert the audio converted into text into digital information using the binary method, the communication device transmits the digital information, and restores the digital information to text and plays it on the communication device receiving the audio.

[0011] Furthermore, the transmission audio parameters of the communication device during the outbound voice communication are collected based on the communication device, and the audio parameters include: audio clarity, audio noise level;

[0012]

[0013] Where: is the audio quality judgment value; n is the total number of audio frequency bands; wi is the weight of the i-th audio frequency band; E vi is the effective speech segment energy in the i-th audio frequency band; E ti is the total energy in the i-th audio frequency band; σ is the standard deviation of the noise amplitude; is the average audio amplitude;

[0014] in, Indicates audio clarity. Indicates the audio noise level and audio quality judgment value The larger the value, the better the audio quality. Conversely, the lower the value, the worse the audio quality. The user end sets the judgment threshold and audio quality judgment value. Compare and determine whether the current audio is of poor quality. If the result is yes, optimize the audio.

[0015] Furthermore, before the audio quality assessment phase begins, the audio is evenly divided to obtain no less than 16 groups of frequency bands, and then the audio quality assessment is performed;

[0016] The energy of the audio frequency band is calculated by the following formula:

[0017]

[0018] Where: E s is the energy of the audio at the sth moment; x(s) is the value of the audio at the sth moment; w(s) is the window function; Q is the time domain where the audio frequency band is located; E ALL is the energy in the audio frequency band;

[0019] Among them, the window function N is the frame length.

[0020] Furthermore, the audio optimization processing is to perform noise reduction processing on the audio, and any one of the Spectral Subtraction algorithm and the Wiener Filtering algorithm is applied during the noise reduction processing of the audio;

[0021] After capturing the instantaneous throughput of the network where the communication device is located, a trend graph representing the change of the instantaneous throughput of the network is generated based on the instantaneous throughput of the network, which is recorded as a first trend graph. The audio is further divided based on the instantaneous measurement unit to obtain a plurality of groups of sub-audio, and the data size of each group of sub-audio is identified. Based on the time sequence of each sub-audio, each group of sub-audio is sorted, and a trend graph representing the change of the data size of each sub-audio based on the time sequence and the instantaneous measurement unit in the audio is generated, which is recorded as a second trend graph.

[0022] The instantaneous measurement unit is one second. After the first trend graph and the second trend graph are generated, it is determined whether the second trend graph is within the first trend graph, and then the audio transmission stability is predicted based on the similarity of the two groups of trend graphs.

[0023] Furthermore, the capture time domain of the network instantaneous throughput is the time domain in the historical network instantaneous throughput that is equal to the length of the audio before the communication device transmits the audio;

[0024] k = sim(a, b) × γ;

[0025] Where: k is the predicted value of audio transmission stability; sim(a,b) is the similarity between the two sets of trend graphs; γ is a constant;

[0026] Among them, the constant γ > 0. When the second trend graph is within the first trend graph and the predicted value k of audio transmission stability is k ≥ 99%, the prediction result is that the audio transmission is stable. When the second trend graph is within the first trend graph and the predicted value k of audio transmission stability is k < 99%, the prediction result is that the audio transmission is unstable. When the second trend graph is not within the first trend graph, the prediction result is that the audio transmission is unstable.

[0027] Furthermore, the network instantaneous throughput is expressed as:

[0028] Th q =R q ×T q ×(1 - L q )×(1 - O);

[0029] In the formula: Th q is the network throughput at time q; R q is the effective data transmission rate at time q; T q is the transmission time, which is 1; L q is the packet loss rate at time q; O is the network protocol overhead ratio;

[0030] Among them, the value of the network protocol overhead ratio O is based on obtained, where H0 represents the reference bandwidth and h represents the link bandwidth.

[0031] Furthermore, when converting audio to text, the applied audio is the audio after performing optimization processing;

[0032] The operations of converting audio to text, identifying the pinyin and tones of the text, further converting the audio converted to text into digital information by combining the binary method, and restoring the digital information to text and playing it are all executed by an intelligent AI with the corresponding running program;

[0033] The intelligent AI is deployed inside the communication device that transmits and receives audio.

[0034] Furthermore, the operation logic of further converting the audio converted to text into digital information by combining the binary method is:

[0035] When the audio is converted to text as "Hello", the recognition result of pinyin and tones is "n, i, 3; h, a, o, 3", and the sorting in the English alphabet is "14, 9, 3; 8, 1, 15, 3", which is recorded as the recognition result of pinyin and tones with "14, 9, 3; 8, 1, 15, 3";

[0036] "14, 9, 3; 8, 1, 15, 3" is converted into digital information containing only 1 and 0 using the binary method. The digital information is transmitted based on a communication device. After being received by another communication device, the digital information is restored to text information based on the opposite logic of the above logic, and the intelligent AI deployed in the communication device is used to play it;

[0037] Among them, the audio transmitted by the communication device is converted into digital information through the above logic after being converted into any text;

[0038] When the above “14, 9, 3; 8, 1, 15, 3” is converted into digital information containing only 1 and 0 in binary method, it obeys:

[0039]

[0040] Where: A is a number; u1, u2, ...0 are the largest integer quotients, v1, v2, ...v x+1 is the remainder, in which v1, v2, ...v x+ 1Each value is either 1 or 0;

[0041] All the numbers in "14, 9, 3; 8, 1, 15, 3" are obtained based on the above formula to obtain the remainders, and the arrangement results of the remainders from bottom to top are obtained, that is, the digital information obtained by converting a group of numbers into binary system.

[0042] Furthermore, digital information originating from the same audio is arranged and packaged based on the time sequence of its corresponding text, and transmitted from the outbound communication device to the communication device receiving the audio.

[0043] Compared with the known public technology, the technical solution provided by the present invention has the following beneficial effects:

[0044] The present invention provides an outbound voice communication quality optimization method based on intelligent AI. During the execution of the method, the audio quality to be transmitted through the device is evaluated to decide whether to optimize the audio, and the stability of the communication device when transmitting the audio is predicted in combination with the instantaneous throughput of the communication network where the communication device is located. The communication device is further adaptively switched to the network executing the audio communication data based on the transmission stability prediction result to ensure that the communication device transmits the audio stably.

[0045] At the same time, combined with the binary method, when all networks available to the communication device are unstable, the audio is converted and transmitted in the form of digital information, and then restored, to further ensure the basic communication needs of the communication device when the network is poor, and to ensure that the information to be transmitted by the communication device through audio transmission is stable and exchanged between communication devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0047] Figure 1 A flowchart of an outbound voice communication quality optimization method based on intelligent AI;

[0048] Figure 2 It is a schematic diagram of the similarity analysis process of the first trend graph and the second trend graph in the present invention;

[0049] Figure 3 This is a logical diagram of digital information conversion in the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution 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 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 are within the scope of protection of the present invention.

[0051] The present invention will be further described below in conjunction with the embodiments.

[0052] Embodiment 1:

[0053] In this embodiment, a method for optimizing the quality of outbound voice communication based on intelligent AI is provided. Figure 1 As shown, including:

[0054] Collect the transmission audio parameters of the communication device during outbound voice communication, evaluate the audio quality based on the audio parameters, optimize the audio when the audio quality evaluation result is poor, and simultaneously capture the instantaneous throughput of the network where the communication device is located, and predict the audio transmission stability based on the instantaneous throughput of the network;

[0055] Obtaining the prediction result of audio transmission stability, when the prediction result is stable, the communication device performs the audio transmission operation, when the prediction result is unstable, controlling the communication device to switch the communication network, and jumping again to the stage of capturing the instantaneous throughput of the network where the communication device is located, and predicting the audio transmission stability based on the instantaneous throughput of the network, until the prediction result is stable, the communication device applies the currently switched communication network to perform the audio transmission operation;

[0056] When the evaluation results of the audio transmission stability corresponding to all switched communication networks are all negative, obtain the audio, convert the audio into text, identify the pinyin and tones of the text, and further convert the audio converted into text into digital information by combining the binary method. The communication device transmits the digital information, and restores the digital information to text and plays it on the communication device that receives the audio;

[0057] When converting the audio into text, the applied audio is the audio after performing optimization processing;

[0058] The operations of converting the audio into text, identifying the pinyin and tones of the text, further converting the audio converted into text into digital information by combining the binary method, and restoring the digital information to text and playing it are all executed by the intelligent AI with the corresponding running program;

[0059] The intelligent AI is deployed inside the communication device that transmits and receives the audio;

[0060] The operation logic of further converting the audio converted into text into digital information by combining the binary method is as follows:

[0061] When the audio is converted into text as "Hello", the recognition results of pinyin and tones are "n, i, 3; h, a, o, 3", and the sorting in the English alphabet is "14, 9, 3; 8, 1, 15, 3", which is recorded as the recognition results of pinyin and tones with "14, 9, 3; 8, 1, 15, 3";

[0062] "14, 9, 3; 8, 1, 15, 3" is converted into digital information containing only 1 and 0 by the binary method. The digital information is transmitted based on the communication device, and after being received by another communication device, the digital information is restored to text information based on the logic opposite to the above logic, and is played using the intelligent AI deployed in the communication device;

[0063] Among them, after the audio transmitted by the communication device is converted into any text, it is converted into digital information through the above logic;

[0064] When "14, 9, 3; 8, 1, 15, 3" is converted into digital information containing only 1 and 0 by the binary method, it follows:

[0065]

[0066] In the formula: A is a number; u1, u2,...0 are the maximum integer quotients, v1, v2,...v x+1 are the remainders, and among the remainders v1, v2,...v x+1 each value is either 1 or 0;

[0067] All the numbers in "14, 9, 3; 8, 1, 15, 3" are obtained by calculating the remainders based on the above formula, and the arrangement of the remainders from bottom to top is obtained, that is, the digital information obtained by converting a group of numbers through the binary method;

[0068] Digital information originating from the same audio is arranged and packaged based on the timing of its corresponding text, and is transmitted from the outbound communication device to the communication device receiving the audio.

[0069] In this embodiment, through the execution of the method in the above embodiment, the communication equipment is equipped with intelligent AI operation. When the communication network is not good, it can adaptively switch to a communication network that meets the communication needs. When all communication networks available to the communication equipment cannot meet the communication needs, based on the digital conversion of the communication audio, the communication audio can be transmitted with lower network requirements, ensuring that the information contained in the communication audio is stably transmitted between communication devices, providing guarantee for real-time interaction of communication devices.

[0070] See also Figure 3 As shown, based on the arrows indicated and the text information recorded in the figure, the operational logic process of further converting the audio converted into text into digital information through the binary method is further demonstrated.

[0071] Embodiment 2:

[0072] In terms of specific implementation, based on Example 1, this example refers to Figure 1 A method for optimizing the quality of outbound voice communication based on intelligent AI in Example 1 is further described in detail:

[0073] The transmission audio parameters of the communication device during outbound voice communication are collected based on the communication device. The audio parameters include: audio clarity and audio noise level;

[0074]

[0075] Where: is the audio quality judgment value; n is the total number of audio frequency bands; wi is the weight of the i-th audio frequency band; E vi is the effective speech segment energy in the i-th audio frequency band; E ti is the total energy in the i-th audio frequency band; σ is the standard deviation of the noise amplitude; is the average audio amplitude;

[0076] in, Indicates audio clarity. Indicates the audio noise level and audio quality judgment value The larger the value, the better the audio quality. Conversely, the lower the value, the worse the audio quality. The user end sets the judgment threshold and audio quality judgment value. Compare and determine whether the current audio is of poor quality. If the result is yes, optimize the audio.

[0077] Before the audio quality assessment phase begins, the audio is evenly divided into no less than 16 frequency bands, and then the audio quality assessment is performed;

[0078] The energy of the audio frequency band is calculated by the following formula:

[0079]

[0080] Where: E s is the energy of the audio at the sth moment; x(s) is the value of the audio at the sth moment; w(s) is the window function; Q is the time domain where the audio frequency band is located; E ALL is the energy in the audio frequency band;

[0081] Among them, the window function N is the frame length.

[0082] In this embodiment, through the above settings, the audio quality assessment logic and basis are further limited, providing necessary data support for the execution of the method in Example 1.

[0083] Embodiment 3:

[0084] In terms of specific implementation, based on Example 1, this example refers to Figure 1 A method for optimizing the quality of outbound voice communication based on intelligent AI in Example 1 is further described in detail:

[0085] The network instantaneous throughput is expressed as:

[0086] Th q =R q ×T q ×(1-L q )×(1-O);

[0087] Where: Th q is the network throughput at time q; R q is the effective data transmission rate at time q; T q is the transmission time, which is 1; L q is the packet loss rate at time q; O is the network protocol overhead ratio;

[0088] The value of network protocol overhead ratio O is based on , H0 represents the reference bandwidth, and h represents the link bandwidth.

[0089] Through the above settings, a calculation logic is provided for the instantaneous throughput of the network, which provides necessary data support for the execution of the method in Example 1.

[0090] like Figure 1 As shown, the audio optimization processing is to perform noise reduction on the audio, and any one of the Spectral Subtraction algorithm and the Wiener Filtering algorithm is applied during the noise reduction processing of the audio;

[0091] After capturing the instantaneous throughput of the network where the communication device is located, a trend graph representing the change of the instantaneous throughput of the network is generated based on the instantaneous throughput of the network, which is recorded as a first trend graph. The audio is further divided based on the instantaneous measurement unit to obtain a plurality of groups of sub-audio, and the data size of each group of sub-audio is identified. Based on the time sequence of each sub-audio, each group of sub-audio is sorted, and a trend graph representing the change of the data size of each sub-audio based on the time sequence and the instantaneous measurement unit in the audio is generated, which is recorded as a second trend graph.

[0092] The instantaneous measurement unit is one second. After the first trend graph and the second trend graph are generated, it is determined whether the second trend graph is within the first trend graph, and then the audio transmission stability is predicted based on the similarity of the two sets of trend graphs.

[0093] Through the above settings, combined with Figure 2 As shown, Figure 2 The two groups of images in the upper middle portion, from left to right, represent the first trend graph and the second trend graph respectively. The graphs pointed by the arrows in the graphs are used to obtain the following audio transmission stability prediction values.

[0094] The capture time domain of the network instantaneous throughput is the time domain equal to the length of the audio before the communication device transmits the audio in the historical network instantaneous throughput;

[0095] k = sim(a, b) × γ;

[0096] Where: k is the predicted value of audio transmission stability; sim(a,b) is the similarity between the two sets of trend graphs; γ is a constant;

[0097] Among them, the constant γ>0, when the second trend graph is within the first trend graph, the audio transmission stability prediction value k≥99%, then the prediction result is stable audio transmission; when the second trend graph is within the first trend graph, the audio transmission stability prediction value k<99%, then the prediction result is unstable audio transmission; when the second trend graph is not within the first trend graph, the prediction result is unstable audio transmission.

[0098] Through the above settings, the prediction logic of the audio transmission stability in Example 1 is further limited, ensuring that the audio transmission stability is output in a digital form, providing data support for the further execution of the method in Example 1.

[0099] In summary, during the execution of the method in the above embodiment, by evaluating the audio quality to be transmitted by the communication device, a decision is made whether to optimize the audio, and the stability of the communication device in transmitting audio is predicted in combination with the instantaneous throughput of the communication network where the communication device is located. Further, based on the transmission stability prediction result, the communication device is adaptively switched to the network in which the audio communication data is executed, to ensure that the communication device transmits audio stably. At the same time, combined with the binary method, when all networks available to the communication device are unstable, the audio is converted and transmitted in the form of digital information, and then restored, to further ensure the basic communication needs of the communication device when the network is poor, and to ensure that the information to be transmitted by the communication device in transmitting audio is stably exchanged between communication devices.

[0100] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing outbound voice communication quality based on intelligent AI, characterized in that: include: Collect the transmission audio parameters of the communication device during outbound voice communication, evaluate the audio quality based on the audio parameters, optimize the audio when the audio quality evaluation result is poor, and simultaneously capture the instantaneous throughput of the network where the communication device is located, and predict the audio transmission stability based on the instantaneous throughput of the network; Obtaining the prediction result of audio transmission stability, when the prediction result is stable, the communication device performs the audio transmission operation, when the prediction result is unstable, controlling the communication device to switch the communication network, and jumping again to the stage of capturing the instantaneous throughput of the network where the communication device is located, and predicting the audio transmission stability based on the instantaneous throughput of the network, until the prediction result is stable, the communication device applies the currently switched communication network to perform the audio transmission operation; When the corresponding audio transmission stability assessment results of all switched communication networks are negative, the audio is acquired, converted into text, the text pinyin and tone are recognized, the audio converted into text is further converted into digital information using a binary method, the communication device transmits the digital information, and the digital information is restored to text and played on the communication device receiving the audio.

2. According to claim 1, a method for optimizing outbound voice communication quality based on intelligent AI is characterized in that: The transmission audio parameters of the communication device during the outbound voice communication are collected based on the communication device, and the audio parameters include: audio clarity and audio noise level; Where: θ is the audio quality judgment value; n is the total amount of audio frequency band; w i is the weight of the i-th audio frequency band; E vi is the effective speech segment energy in the i-th audio frequency band; E ti is the total energy in the i-th audio frequency band; σ is the standard deviation of the noise amplitude; is the average audio amplitude; in, Indicates audio clarity. Indicates the audio noise level. The larger the audio quality judgment value θ is, the better the audio quality is. Conversely, the worse the audio quality is. The user end sets the judgment threshold and compares it with the audio quality judgment value θ to determine whether the current audio is of poor quality. If the judgment result is yes, the audio is optimized.

3. The method for optimizing outbound voice communication quality based on intelligent AI according to claim 1 or 2, characterized in that: Before the audio quality assessment phase begins, the audio is evenly divided to obtain no less than 16 frequency bands, and then the audio quality assessment is performed; The energy of the audio frequency band is calculated by the following formula: Where: E s is the energy of the audio at the sth moment; x(s) is the value of the audio at the sth moment; w(s) is the window function; Q is the time domain where the audio frequency band is located; E ALL is the energy in the audio frequency band; Among them, the window function N is the frame length.

4. According to claim 1, a method for optimizing outbound voice communication quality based on intelligent AI is characterized in that: The audio optimization processing is to perform noise reduction on the audio, and any one of the Spectral Subtraction algorithm and the Wiener Filtering algorithm is applied to the audio during the noise reduction processing; After capturing the instantaneous throughput of the network where the communication device is located, a trend graph representing the change of the instantaneous throughput of the network is generated based on the instantaneous throughput of the network, which is recorded as a first trend graph. The audio is further divided based on the instantaneous measurement unit to obtain a plurality of groups of sub-audio, and the data size of each group of sub-audio is identified. Based on the time sequence of each sub-audio, each group of sub-audio is sorted, and a trend graph representing the change of the data size of each sub-audio based on the time sequence and the instantaneous measurement unit in the audio is generated, which is recorded as a second trend graph. The instantaneous measurement unit is one second. After the first trend graph and the second trend graph are generated, it is determined whether the second trend graph is within the first trend graph, and then the audio transmission stability is predicted based on the similarity of the two groups of trend graphs.

5. According to claim 4, a method for optimizing outbound voice communication quality based on intelligent AI is characterized in that: The capture time domain of the network instantaneous throughput is the time domain equal to the length of the audio before the communication device transmits the audio in the historical network instantaneous throughput; k = sim(a, b) × γ; Where: k is the predicted value of audio transmission stability; sim(a,b) is the similarity between the two sets of trend graphs; γ is a constant; Among them, the constant γ>0, when the second trend graph is within the first trend graph, the audio transmission stability prediction value k≥99%, then the prediction result is stable audio transmission; when the second trend graph is within the first trend graph, the audio transmission stability prediction value k<99%, then the prediction result is unstable audio transmission; when the second trend graph is not within the first trend graph, the prediction result is unstable audio transmission.

6. The method for optimizing outbound voice communication quality based on intelligent AI according to claim 1, characterized in that: The instantaneous network throughput is expressed as: Th q =R q ×T q ×(1-L q )×(1-O); Where: Th q is the network throughput at time q; R q is the effective data transmission rate at time q; T q is the transmission time, which is 1; L q is the packet loss rate at time q; O is the network protocol overhead ratio; The value of network protocol overhead ratio O is based on Calculate, H0 represents the reference bandwidth, h represents the link bandwidth.

7. The method for optimizing outbound voice communication quality based on intelligent AI according to claim 1, characterized in that: When converting audio to text, the audio applied is the audio after performing optimization processing. The operations of converting audio to text, identifying the pinyin and tones of the text, further converting the audio converted to text into digital information by combining the binary method, and restoring the digital information to text and playing it are all performed by an intelligent AI with corresponding running programs. The intelligent AI is deployed inside the communication device that transmits and receives audio.

8. The method for optimizing outbound voice communication quality based on intelligent AI according to claim 7, characterized in that: The operation logic of further converting the audio converted to text into digital information by combining the binary method is as follows: When the audio is converted to text "Hello", the pinyin and tone recognition results are "n, i, 3; h, a, o, 3", and their order in the English alphabet is "14, 9, 3; 8, 1, 15, 3", and "14, 9, 3; 8, 1, 15, 3" is recorded as the pinyin and tone recognition results. "14，9，3； "8, 1, 15, 3" is converted into digital information containing only 1 and 0 by the binary method. The digital information is transmitted based on the communication device and received by another communication device. Then, based on the logic opposite to the above, the digital information is restored to text information and played using the intelligent AI deployed in the communication device. Among them, after the audio transmitted by the communication device is converted into any text, it is converted into digital information through the above logic.

9. The method for optimizing outbound voice communication quality based on intelligent AI according to claim 8, characterized in that: When converting "14, 9, 3; 8, 1, 15, 3" into digital information containing only 1 and 0 by the binary method, it follows: Where: A is a number; u1, u2, ...0 are the largest integer quotients, v1, v2, ...v x+1 is the remainder, in which v1, v2, ...v x+1 Each value is either 1 or 0; "14，9，3； All the numbers in "8, 1, 15, 3" obtain remainders based on the above formula, and the arrangement result of obtaining the remainders from bottom to top is the digital information obtained by converting a group of numbers through the binary method.

10. The method for optimizing outbound voice communication quality based on intelligent AI according to claim 8, characterized in that: The digital information from the same audio is arranged and packaged based on the time sequence of its corresponding text and transmitted from the calling communication device to the communication device receiving the audio.

Images