Intelligent voice interaction method and device based on large language model

Abstract

The present invention provides an intelligent voice interaction method and device based on a large language model. The method and device receive an interaction instruction sent by a voice interaction terminal and interaction audio data corresponding to the interaction instruction, perform timbre analysis on the interaction audio data based on the large language model, generate a predicted value segment, and retrieve a broadcast sentence corresponding to the interaction instruction; retrieve basic information of a binding person set by a corresponding binding terminal based on the predicted value segment, display and customize the broadcast sentence according to the basic information, and obtain a display font size; obtain environmental noise collected by a collection device, perform broadcast customization adjustment on the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise, and obtain a customized interval duration and a customized volume of adjacent broadcast characters; send the broadcast sentence to a display interface of the voice interaction terminal for display according to the display font size, and control the voice interaction terminal to play the broadcast sentence based on the customized interval duration and the customized volume.

Description

Technical Field

[0001] The present invention relates to data processing technology, and in particular to an intelligent voice interaction method and device based on a large language model. Background Art

[0002] With the rapid development of artificial intelligence technology, intelligent voice interaction technology has become an important part of the field of human-computer interaction. Traditional voice interaction systems mostly rely on rule or template matching methods, which limits the flexibility and intelligence of the system and does not take into account the actual needs of users.

[0003] However, the existing intelligent voice interaction systems based on large language models still have some shortcomings. For example, the systems often only focus on the recognition and response of voice content, but ignore the personalized needs of users and the differences in environments. Different users and environments have different broadcasting methods.

[0004] Therefore, how to further improve the personalization and environmental adaptability of the intelligent voice interaction system according to different users and environments, thereby improving the user experience, has become an urgent problem to be solved. Summary of the invention

[0005] The embodiments of the present invention provide an intelligent voice interaction method and device based on a large language model, which can further improve the personalization and environmental adaptability of the intelligent voice interaction system according to different users and environments, thereby improving the user experience.

[0006] According to a first aspect of an embodiment of the present invention, there is provided an intelligent voice interaction method based on a large language model, comprising:

[0007] Receive the interactive command sent by the voice interactive terminal, and the interactive audio data corresponding to the interactive command, perform timbre analysis on the interactive audio data based on the large language model, generate a predicted value segment, and retrieve the broadcast sentence corresponding to the interactive command;

[0008] Based on the predicted value segment, basic information of the binding personnel set by the corresponding binding terminal is retrieved, and the broadcast sentence is displayed and customized according to the basic information to obtain a display font size;

[0009] The acquisition device collects environmental noise, and customizes the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise, so as to obtain the customized interval duration and customized volume of adjacent broadcast characters;

[0010] The announcement sentence is sent to the display interface of the voice interaction terminal for display according to the display font size, and the voice interaction terminal is controlled to play the announcement sentence based on the customized interval duration and the customized volume.

[0011] Optionally, in a possible implementation of the first aspect, performing timbre analysis on the interactive audio data based on a large language model, generating a predicted value segment, and retrieving a broadcast sentence corresponding to the interactive instruction includes:

[0012] Performing timbre analysis on the interactive audio data based on a large language model to obtain a timbre stage corresponding to the interactive audio data;

[0013] A preset numerical segment corresponding to the timbre stage is retrieved as a predicted numerical segment, the timbre stage having a preset numerical segment corresponding thereto, and a broadcast sentence corresponding to the interactive instruction is retrieved.

[0014] Optionally, in a possible implementation manner of the first aspect, the retrieving basic information of the binding person set by the corresponding binding terminal based on the predicted value segment, displaying and customizing the announcement sentence according to the basic information, and obtaining the display font size includes:

[0015] Obtaining basic information of the bound person set by each of the binding terminals, retrieving age information in the basic information, and when determining that the age information is within the predicted value range, retrieving basic information of the bound person set by the corresponding binding terminal;

[0016] Retrieving age information in the corresponding basic information, and when it is determined that the age information is greater than a maximum value of a preset age interval, obtaining a first age difference according to a difference between the age information and the maximum value of the preset age interval, wherein the preset age interval has a corresponding preset font size;

[0017] The preset font size is increased and adjusted according to the first age difference to obtain a display font size.

[0018] Optionally, in a possible implementation manner of the first aspect, the increasing and adjusting the preset font size according to the first age difference to obtain a display font size includes:

[0019] The display font size is obtained by the following formula:

[0020] ,

[0021] in, To display the font size, For the default font size, For age information, is the maximum value of the preset age range, is the font size weight value.

[0022] Optionally, in a possible implementation of the first aspect, the method further includes:

[0023] Obtaining the interface size of the display interface corresponding to the voice interaction terminal, wherein the interface size includes the interface length and the interface width;

[0024] Retrieving the display length and display width of the display font size, obtaining a first display quantity based on the ratio of the interface length to the display length, and obtaining a second display quantity based on the ratio of the interface width to the display width;

[0025] Obtaining a total display quantity corresponding to the display interface according to the product of the first display quantity and the second display quantity;

[0026] Retrieving the number of characters in the broadcast sentence, and when it is determined that the number of characters is greater than the total display number, obtaining the number of display pages according to the ratio of the number of characters to the total display number;

[0027] Based on the ratio of the number of characters to the number of display pages, the number of page characters corresponding to each display page is obtained;

[0028] The benchmark dwell time is adjusted according to the number of characters on the page and the age information to obtain the dwell time corresponding to the displayed page.

[0029] Optionally, in a possible implementation manner of the first aspect, adjusting the benchmark stay time according to the number of characters on the page and the age information to obtain the stay time corresponding to the displayed page includes:

[0030] Obtaining a character adjustment coefficient according to a ratio of the number of characters on the page to the number of reference characters;

[0031] When it is determined that the age information is not within the preset age range, retrieving the maximum value and the minimum value of the preset age range, the preset age range having a corresponding preset stay time;

[0032] When it is determined that the age information is less than the minimum value, a second age difference is obtained according to a difference between the minimum value and the age information, and an age adjustment coefficient is obtained according to a ratio of the second age difference to the minimum value;

[0033] When it is determined that the age information is greater than the maximum value, a first age difference is obtained according to a difference between the age information and the maximum value, and an age adjustment coefficient is obtained according to a ratio of the first age difference to the maximum value;

[0034] The preset stay time is adjusted according to the age adjustment coefficient and the character adjustment coefficient to obtain the stay time corresponding to the display page.

[0035] Optionally, in a possible implementation manner of the first aspect, adjusting the preset stay time according to the age adjustment coefficient and the character adjustment coefficient to obtain the stay time corresponding to the display page includes:

[0036] The dwell time is obtained by the following formula:

[0037] ,

[0038] in, For age information, is the maximum value of the preset age range, is the minimum value of the preset age range, For the length of stay, To preset the duration of stay, is the number of characters on the page, is the base number of characters, is the weight value of the number of characters, is the age weight value.

[0039] Optionally, in a possible implementation of the first aspect, the acquisition and collection device collects environmental noise, and performs customized broadcast adjustment on the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise, to obtain the customized interval duration and customized volume of adjacent broadcast characters, including:

[0040] Acquiring environmental noise collected by a collection device, retrieving age information and basic disease types in the corresponding basic information, and retrieving disease type quantification values ​​based on the basic disease types;

[0041] Determining the noise volume of the environmental noise, and obtaining a noise adjustment coefficient according to a ratio of the noise volume to a reference volume;

[0042] Retrieving the age adjustment coefficient, and custom-adjusting the character interval duration according to the noise adjustment coefficient and the age adjustment coefficient to obtain a custom interval duration;

[0043] The broadcast volume is customized and adjusted based on the noise adjustment coefficient, the age adjustment coefficient and the disease type quantization value to obtain a customized volume.

[0044] Optionally, in a possible implementation of the first aspect, the method further includes:

[0045] Acquire a noise waveform of the environmental noise, and when determining that the noise waveform is a regular waveform, extract the number of waveforms in the regular waveform;

[0046] When it is determined that the number of waveforms is greater than 1, multiple sub-waveforms in the regular waveform and waveform interval durations between adjacent sub-waveforms are obtained, and the waveform interval duration greater than or equal to the character playback duration is obtained as the positioning placement duration;

[0047] Counting the middle interference durations between adjacent positioning placement durations, the head interference duration before the head positioning placement duration, and / or the tail interference duration after the tail positioning placement duration;

[0048] Obtaining an interference duration according to the first interference duration, the middle interference duration, and the tail interference duration;

[0049] Taking the interference duration longer than the customized interval duration as the interference duration to be adjusted, and obtaining the positioning placement duration before the interference duration to be adjusted as the first placement duration;

[0050] According to the ratio of the first placement time to the character playback time, a first playback word count corresponding to each of the first placement time is obtained, and the first playback word count is counted to obtain a first total word count;

[0051] Determine the corresponding broadcast characters in the broadcast sentence as adjustment characters based on the first total number of characters, and replace the customized interval duration between the adjustment characters with the interference duration to be adjusted;

[0052] When it is determined that the number of waveforms is equal to 1, the interval duration between adjacent regular waveforms is obtained, and when it is determined that the interval duration is greater than or equal to the character playback duration, the duration of the regular waveform is obtained as the regular interference duration, and the customized interval duration is adjusted to the regular interference duration;

[0053] When it is determined that the noise waveform is an irregular waveform, the average of the waveform durations corresponding to multiple sub-waveforms in the irregular waveform is determined based on the preset extraction duration, the average waveform duration corresponding to the irregular waveform is obtained, and the customized interval duration is adjusted to the average waveform duration.

[0054] A second aspect of an embodiment of the present invention provides an intelligent voice interaction device based on a large language model, including:

[0055] An analysis module is used to receive the interaction instruction sent by the voice interaction terminal and the interaction audio data corresponding to the interaction instruction, perform timbre analysis on the interaction audio data based on the large language model, generate a predicted value segment, and retrieve the broadcast sentence corresponding to the interaction instruction;

[0056] A customization module, used for retrieving basic information of a binding person set by a corresponding binding terminal based on the predicted value segment, and displaying and customizing the announcement sentence according to the basic information to obtain a display font size;

[0057] An adjustment module is used to obtain the environmental noise collected by the collection device, and customize the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise, so as to obtain the customized interval duration and customized volume of adjacent broadcast characters;

[0058] A broadcast module is used to send the broadcast sentence to the display interface of the voice interaction terminal for display according to the display font size, and control the voice interaction terminal to play the broadcast sentence based on the customized interval duration and the customized volume.

[0059] According to a third aspect of an embodiment of the present invention, there is provided an electronic device, comprising: a memory, a processor and a computer program, wherein the computer program is stored in the memory, and the processor runs the computer program to execute the first aspect of the present invention and various methods that may be involved in the first aspect.

[0060] According to a fourth aspect of an embodiment of the present invention, a storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the first aspect of the present invention and various methods that may be involved in the first aspect.

[0061] The beneficial effects of the present invention are as follows:

[0062] 1. The present invention significantly improves the personalized interactive experience of users through an intelligent voice interaction method based on a large language model. Specifically, the invention can receive interactive instructions and corresponding interactive audio data sent by the voice interaction terminal, use a large language model to perform timbre analysis on the interactive audio data, generate predicted numerical segments, and retrieve the broadcast sentences that match the interactive instructions accordingly. This process not only realizes the accurate recognition of the user's voice, but also further predicts the user's age and other characteristics through timbre analysis, providing a basis for the customization of subsequent broadcast sentences. In addition, according to the user's basic information, such as age, disease condition, etc., the broadcast sentences are displayed and customized, such as adjusting the display font size to meet the visual needs of different users. These personalized customization measures work together to enhance the user's interactive experience.

[0063] 2. The present invention also shows significant beneficial effects in enhancing environmental adaptability. During the interaction process, the invention can obtain the environmental noise collected by the collection device, and customize the character interval duration and broadcast volume of the broadcast characters in the broadcast sentence according to the environmental noise and user basic information. For example, in a noisy environment, by increasing the broadcast volume and character interval duration, it is ensured that users can receive information clearly. This adjustment process not only improves the anti-interference ability of voice interaction, but also enables users to have a good interactive experience in different environments. Overall, this adaptive adjustment based on environmental noise enhances the environmental adaptability of intelligent voice interaction equipment, allowing the equipment to maintain efficient interactive performance in various environments.

[0064] 3. The present invention has also achieved remarkable results in optimizing the comprehensive interactive effect. On the one hand, through precise timbre analysis and the use of basic user information, the invention realizes the personalized customization of broadcast sentences, which not only meets the visual needs of users, but also adapts to the hearing and comprehension abilities of different users by adjusting the broadcast mode and speed. On the other hand, the invention further improves the clarity and comprehensibility of voice interaction by real-time monitoring and adjusting the impact of environmental noise on broadcast sentences. In addition, the invention also considers the matching problem of the size of the display interface and the display font size to ensure the clear display of the broadcast sentences on the display interface. These measures work together to enhance the comprehensive effect of intelligent voice interaction, so that users can have a smoother, more accurate and comfortable experience during the interaction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 A flowchart of an intelligent voice interaction method based on a large language model provided by the present invention;

[0066] Figure 2 A schematic diagram of the structure of an intelligent voice interaction device based on a large language model provided by the present invention;

[0067] Figure 3 A schematic diagram of the hardware structure of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0068] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0069] The present invention provides an intelligent voice interaction method based on a large language model, such as Figure 1 As shown, including:

[0070] S1, receiving the interaction instructions sent by the voice interaction terminal, and the interaction audio data corresponding to the interaction instructions, performing timbre analysis on the interaction audio data based on the large language model, generating a predicted value segment, and calling the broadcast sentence corresponding to the interaction instruction.

[0071] The voice interaction terminal may be a smart speaker, the interaction command is the command sent by the person who is interacting, the interaction audio data is the audio data sent by the person who is interacting, and the large language model is a technology in the field of artificial intelligence, which is mainly used to process and understand human language. These models are trained through deep learning and a large amount of text data, and can perform a variety of natural language processing tasks, including but not limited to text generation, language translation, sentiment analysis, timbre recognition, etc. The predicted value segment is the predicted age value segment, for example, over 60 years old.

[0072] Therefore, after receiving the interactive command sent by the voice interaction terminal, this solution will use the large language model to perform timbre analysis on the interactive audio data corresponding to the interactive command. For example, the voice of the elderly is usually low and soft. This is because the vocal cords and throat muscles vibrate less frequently during phonation due to aging. The vocal cords of young people are more flexible and elastic, and the timbre is usually brighter and crisper. The age value segment corresponding to the interactive person can be predicted through timbre analysis. The approximate age range of the person analyzed by the model is a prior art and will not be elaborated here.

[0073] The broadcast sentence corresponding to the interactive instruction is retrieved. For example, the interactive instruction is "How is the weather today?" The broadcast sentence is "The weather is sunny, the temperature is 20° C.", etc. This is the prior art and will not be described in detail.

[0074] In some embodiments, the specific implementation of step S1 (the step of performing timbre analysis on the interactive audio data based on the large language model, generating a predicted value segment, and retrieving a broadcast sentence corresponding to the interactive instruction) includes:

[0075] S11, performing timbre analysis on the interactive audio data based on a large language model to obtain a timbre stage corresponding to the interactive audio data.

[0076] It is not difficult to understand that the server will perform timbre analysis on the interactive audio data based on the large language model to analyze the growth stage of the interactive person's timbre, such as the childhood stage (low volume, high pitch, and baby voice), the adult stage (clear, accurate, and crisp pronunciation), and the old age stage (low and hoarse).

[0077] S12, retrieve a preset numerical segment corresponding to the timbre stage as a predicted numerical segment, the timbre stage having a preset numerical segment corresponding thereto, and retrieve a broadcast sentence corresponding to the interactive instruction.

[0078] It is understandable that each timbre stage has a corresponding preset value segment, such as the childhood stage (0-12 years old) and the elderly stage (60 years old and above). Therefore, the age group that the corresponding interactive person may be in will be retrieved according to the growth stage. The broadcast sentence corresponding to the interactive instruction will be retrieved later.

[0079] S2, based on the predicted value segment, retrieve the basic information of the binding personnel set by the corresponding binding terminal, customize the display of the announcement sentence according to the basic information, and obtain the display font size.

[0080] In some embodiments, the specific implementation of step S2 (the step of retrieving basic information of the binding personnel set by the corresponding binding terminal based on the predicted value segment, and displaying and customizing the announcement sentence according to the basic information to obtain the display font size) includes:

[0081] S21, obtaining basic information of the binding personnel set by each of the binding terminals, retrieving age information in the basic information, and when determining that the age information is in the predicted value segment, retrieving basic information of the binding personnel set by the corresponding binding terminal.

[0082] It should be noted that before the smart speaker is used normally, people can bind the smart speaker to their mobile terminal, such as a mobile phone, tablet, etc. During the binding process, people can actively enter their relevant information. For example, all members of a family are bound to the smart speaker, including the elderly, adults, children, etc.

[0083] Among them, the basic information is the basic information corresponding to the bound person, such as name, age, height, and past medical history. For example, the elderly can enter common diseases, high blood pressure, blood sugar, etc., which can be used to remind people to take medicine on time, measure blood pressure, etc. Therefore, people can enter the corresponding historical medical record status according to their own situation. The age information is the age value of the bound person.

[0084] It is understandable that the server will obtain the basic information of the bound persons set by each binding end, and retrieve the age information in the basic information, that is, obtain all the ages of the bound persons, and then check one by one which of the bound persons is in the predicted numerical range. For example, if an elderly family member of 70 years old is over 60 years old, the basic information set by the mobile device corresponding to the elderly person will be retrieved.

[0085] S22, retrieve the age information in the corresponding basic information, and when it is determined that the age information is greater than the maximum value of a preset age range, obtain a first age difference according to the difference between the age information and the maximum value of the preset age range, and the preset age range has a corresponding preset font size.

[0086] It should be noted that as people age, their vision generally declines. This may be because the lens of the eye gradually loses its elasticity and its ability to adjust decreases with age, making it difficult to see objects clearly at close range, a phenomenon known as presbyopia. In addition, eye diseases such as cataracts, glaucoma, and macular degeneration may also affect the vision of the elderly. Therefore, larger fonts can help them see information more clearly.

[0087] The preset age range is an artificially preset age range, for example, 12 to 60 years old, wherein the preset age range has a corresponding preset font size, for example, normally a size 14 font.

[0088] It is understandable that the server will retrieve the age information of the interacting person from the basic information corresponding to the person. When it is determined that the person's age information is greater than the maximum value of the preset age range, it means that the person interacting at this time is an elderly person. The older the age, the higher the degree of vision loss may be.

[0089] Therefore, the present invention obtains a first age difference according to the difference between the age information and the maximum value of the preset age range, and subsequently increases and adjusts the preset font size according to the first age difference to obtain a display font size.

[0090] S23, increasing and adjusting the preset font size according to the first age difference to obtain a display font size.

[0091] In some embodiments, a specific implementation of step S23 (increasing and adjusting the preset font size according to the first age difference to obtain a display font size) includes:

[0092] The display font size is obtained by the following formula:

[0093] ,

[0094] in, To display the font size, For the default font size, For age information, is the maximum value of the preset age range, is the font size weight value.

[0095] It is not difficult to understand that is the first age difference, the first age difference With the corresponding display font size Positively proportional, the first age difference The larger the corresponding display font size The larger the font size, the greater the font weight. It can be a weight value preset manually based on actual conditions.

[0096] In some embodiments, it also includes:

[0097] A1, obtaining the interface size of the display interface corresponding to the voice interaction terminal, where the interface size includes the interface length and the interface width.

[0098] The display interface is the display interface corresponding to the voice interaction terminal. It is not difficult to understand that the existing smart speakers have a corresponding display screen. The interface size is the size of the display interface, for example, 20 cm long and 10 cm wide.

[0099] A2, retrieve the display length and display width of the display font size, obtain a first display quantity based on the ratio of the interface length to the display length, and obtain a second display quantity based on the ratio of the interface width to the display width.

[0100] It is understandable that the server will retrieve the display length and display width of the display font size, for example, the length of font size 14 is 14 mm and the width is 5 mm. Each display font size has a corresponding display length and display width.

[0101] Subsequently, a first display quantity is obtained according to the ratio of the interface length to the display length, and the server can determine the number of fonts that can be placed in the length direction according to the ratio of the interface length to the display length. A second display quantity is obtained according to the ratio of the interface width to the display width, and the server can determine the number of fonts that can be placed in the width direction according to the ratio of the interface width to the display width.

[0102] It is worth mentioning that if the number that can be placed in the length direction and the width direction is not an integer, it will be rounded down.

[0103] A3, obtaining a total display quantity corresponding to the display interface according to the product of the first display quantity and the second display quantity.

[0104] It can be understood that the total display quantity corresponding to the display interface is obtained according to the product of the first display quantity and the second display quantity.

[0105] A4, retrieve the number of characters in the broadcast sentence, and when it is determined that the number of characters is greater than the total display number, obtain the number of display pages according to the ratio of the number of characters to the total display number.

[0106] It is understandable that the server will retrieve the number of characters in the broadcast sentence, for example, if the weather is fine today, the number of characters is 6.

[0107] Therefore, when the number of characters is greater than the total display number, it means that the number of characters corresponding to this font cannot be displayed on the screen at the same time. For example, if the total display number is 2 characters, the corresponding number of display pages is 3.

[0108] It is worth mentioning that if the number cannot be divided evenly, it will be rounded up. For example, when the number of characters is 5 and the total number of displayed characters is 2, the corresponding number of displayed pages is 2.5, which is rounded up to 3.

[0109] A5. Based on the ratio of the number of characters to the number of display pages, the number of page characters corresponding to each display page is obtained.

[0110] It can be understood that the number of page characters corresponding to each of the display pages can be obtained according to the ratio of the number of characters to the number of display pages.

[0111] For example, if the number of display pages is 3 and the number of characters is 6, then the number of characters on each display page is 2.

[0112] It is worth mentioning that when the number cannot be divided evenly, for example, the number of display pages is 3 and the number of characters is 5, then each page should have 1.6 characters. At this time, round up first and put 2 characters on each page until the last page has one of the remaining characters.

[0113] A6, adjusting the reference stay time according to the number of characters on the page and the age information to obtain the stay time corresponding to the displayed page.

[0114] It is understandable that the benchmark dwell time is adjusted according to the age information of the number of characters on the page to obtain the dwell time corresponding to the displayed page.

[0115] Among them, the benchmark stay time is the preset stay time for each display page, and the preset age range has a corresponding benchmark stay time. It is not difficult to understand that after the age range is preset manually, the normal stay time and normal font size corresponding to the range will be set, that is, the benchmark stay time and the preset font size.

[0116] In some embodiments, a specific implementation of step A6 (adjusting the benchmark dwell time according to the number of characters on the page and the age information to obtain the dwell time corresponding to the displayed page) includes:

[0117] A61, obtaining a character adjustment coefficient according to the ratio of the number of characters on the page to the number of reference characters.

[0118] It can be understood that the ratio of the number of characters on the page to the number of reference characters gives the character adjustment coefficient.

[0119] A62, when it is determined that the age information is not within the preset age range, the maximum value and the minimum value of the preset age range are retrieved, and the preset age range has a corresponding preset stay time.

[0120] It is understandable that when it is determined that the age information is not within the preset age range, the maximum and minimum values ​​of the preset age range are retrieved, for example, 12 years old and 60 years old, and the preset age range has a corresponding preset stay time.

[0121] A63, when it is determined that the age information is less than the minimum value, a second age difference is obtained according to the difference between the minimum value and the age information, and an age adjustment coefficient is obtained according to the ratio of the second age difference to the minimum value.

[0122] It can be understood that when it is determined that the age information is less than the minimum value, that is, a child interacts with it, then a second age difference is obtained according to the difference between the age of the bound child and the minimum value.

[0123] Furthermore, a second age difference is obtained based on the difference between the minimum value and the age information, and an age adjustment coefficient is obtained based on the ratio of the second age difference to the minimum value.

[0124] A64, when it is determined that the age information is greater than the maximum value, a first age difference is obtained according to the difference between the age information and the maximum value, and an age adjustment coefficient is obtained according to the ratio of the first age difference to the maximum value.

[0125] It can be understood that when it is judged that the age information is greater than the maximum value, the first age difference is obtained according to the difference between the age information and the maximum value. At this time, the age deviation value can be obtained by the difference between the age of the elderly and the maximum value.

[0126] Subsequently, an age adjustment coefficient is obtained according to the ratio of the first age difference and the maximum value.

[0127] A65, adjusting the preset stay time according to the age adjustment coefficient and the character adjustment coefficient to obtain the stay time corresponding to the display page.

[0128] In some embodiments, a specific implementation of step A65 (adjusting the preset stay time according to the age adjustment coefficient and the character adjustment coefficient to obtain the stay time corresponding to the display page) includes:

[0129] The dwell time is obtained by the following formula:

[0130] ,

[0131] in, For age information, is the maximum value of the preset age range, is the minimum value of the preset age range, For the length of stay, To preset the duration of stay, is the number of characters on the page, is the base number of characters, is the weight value of the number of characters, is the age weight value.

[0132] It should be noted that as they age, the cognitive abilities of the elderly, such as memory, attention, and reaction speed, will decline. This makes them need longer to understand and process information when reading or listening. Therefore, in order for them to better understand and accept information, it is necessary to appropriately extend the duration of displaying paragraph words. Children are in the stage of cognitive development and need longer time to process and absorb information. Therefore, when displaying paragraph words, appropriately extending the duration will help them better understand and remember the content displayed.

[0133] Therefore, this solution will select different stay times according to the age of the interacting personnel, thereby appropriately extending the stay time of each page display.

[0134] It is not difficult to understand that the length of stay Number of characters per page Proportional to the length of stay and The second age difference and The first age difference is proportional to the number of characters on the page. The more words there are, the longer it takes to read, so the corresponding stay time The longer the second age difference is, and the larger the second age difference is than the first age difference, it means that the corresponding interactive person is younger or older, so it takes more time to understand, so the corresponding stay time Larger, where the weight of the number of characters and age weight Can be pre-set, .

[0135] It is not difficult to understand that this solution will adaptively adjust the page dwell time according to the number of words displayed and the age of the corresponding interacting personnel, so as to facilitate personnel to view and understand, and avoid the problem of personnel being unable to view due to too fast display.

[0136] Based on the above embodiment, it also includes:

[0137] The active stay duration input by the user terminal is received, and the character adjustment ratio is obtained according to the ratio of the character quantity weight value and the age weight value.

[0138] Based on the ratio of the age weight value to the character quantity weight value, an age adjustment ratio is obtained.

[0139] When it is determined that the active stay duration is greater than the stay duration, a first adjusted duration is obtained according to a difference between the active stay duration and the stay duration.

[0140] The character quantity weight value is adjusted according to the first adjustment duration and the character adjustment ratio to obtain an increased and adjusted character quantity weight value.

[0141] The age weight value is adjusted based on the first adjustment duration and the age adjustment ratio to obtain an increased and adjusted age weight value.

[0142] The following formula is used to obtain the increased and adjusted character quantity weight value and the increased and adjusted age weight value:

[0143] ,

[0144] in, For active stay duration, To increase the adjusted character count weight value, To adjust the constant value, To increase the adjusted age weight value.

[0145] It is not difficult to understand that in the model for training the duration of subtitles, age is considered an important influencing factor. Older people may read slower, or due to factors such as vision and cognitive ability, it takes longer to read and understand subtitles. Therefore, in the model training process, the weight of age is set to be larger to ensure that the duration of subtitles can meet the needs of audiences of different ages. Although the number of words is also an influencing factor, the number of words in the subtitles usually does not change during the subsequent adjustment process. Once the content of the subtitles is determined, the number of words is fixed, so the direct impact of the number of words on the duration of subtitles is relatively small. In the model training process, the weight of the number of words may therefore be set to be smaller.

[0146] Therefore, if a person feels that the subtitle stays for too short a time, it is necessary to significantly increase the person's age weight value, so that the subsequent stay time is longer, while the word count weight is relatively less increased.

[0147] The present invention will adaptively increase the weights for training according to the proportions of various factors in actual situations, so as to achieve a weight value that fits the actual situation.

[0148] When it is determined that the active stay duration is less than the stay duration, a second adjusted duration is obtained according to a difference between the stay duration and the active stay duration.

[0149] The character quantity weight value is adjusted according to the second adjustment duration and the character adjustment ratio to obtain a reduced and adjusted character quantity weight value.

[0150] The age weight value is adjusted based on the first adjustment duration and the age adjustment ratio to obtain a reduced and adjusted age weight value.

[0151] The following formula is used to obtain the reduced adjusted character quantity weight value and the reduced adjusted age weight value:

[0152] ,

[0153] in, For active stay duration, To reduce the adjusted character weight value, To adjust the constant value, To reduce the adjusted age weight value.

[0154] It is not difficult to understand that in the model for training the duration of subtitles, age is considered an important influencing factor. Older people may read slower, or due to factors such as vision and cognitive ability, it takes longer to read and understand subtitles. Therefore, in the model training process, the weight of age is set to be larger to ensure that the duration of subtitles can meet the needs of audiences of different ages. Although the number of words is also an influencing factor, the number of words in the subtitles usually does not change during the subsequent adjustment process. Once the content of the subtitles is determined, the number of words is fixed, so the direct impact of the number of words on the duration of subtitles is relatively small. In the model training process, the weight of the number of words may therefore be set to be smaller.

[0155] Therefore, if a person feels that the subtitles stay on for too long, it is necessary to significantly reduce the person's age weight value, so that the subsequent stay time is shorter, and the word count weight is reduced relatively less.

[0156] The present invention can adaptively reduce the weights for training according to the proportion of each factor in the actual situation, so as to make the weight values ​​fit the actual situation.

[0157] S3, obtaining environmental noise collected by a collection device, and making customized adjustments to the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise, so as to obtain customized interval duration and customized volume of adjacent broadcast characters.

[0158] In some embodiments, the specific implementation of step S3 (the acquisition and collection device collects environmental noise, and customizes and adjusts the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise to obtain the customized interval duration and customized volume of adjacent broadcast characters) includes:

[0159] S31, obtaining environmental noise collected by a collection device, retrieving age information and basic disease types in the corresponding basic information, and retrieving disease type quantification values ​​based on the basic disease types.

[0160] Among them, the collecting device is a device for collecting environmental noise, the basic disease type is the disease type corresponding to each interacting person, such as hypertension, and the disease type quantification value is the quantification value corresponding to various disease types.

[0161] It should be noted that cardiovascular disease, diabetes, hypertension, etc. may affect the blood circulation of the inner ear, resulting in damage to the hair cells in the cochlea, thereby affecting the hearing function. Therefore, the present invention will quantify the impact of each disease type with the help of existing quantification methods to obtain a corresponding quantified value. For example, the hearing condition of each disease type corresponding to the person is statistically quantified to obtain an impact value. The greater the impact, the greater the corresponding value.

[0162] S32, determining the noise volume of the environmental noise, and obtaining a noise adjustment coefficient according to a ratio of the noise volume to a reference volume.

[0163] It is not difficult to understand that the server will obtain the noise adjustment coefficient based on the ratio of the noise volume to the reference volume, wherein the reference volume can be a basic noise volume preset manually.

[0164] S33, retrieve the age adjustment coefficient, and customize the character interval duration according to the noise adjustment coefficient and the age adjustment coefficient to obtain a customized interval duration.

[0165] It should be noted that the elderly may have a decline in hearing, memory and cognitive abilities due to aging, which may slow down their speed in processing and understanding information. Slowing down the speed of broadcasting can help them better capture and digest information. Similarly, children are in the stage of learning language, and their language comprehension and processing abilities are still developing. A slower speed of broadcasting can help them better understand and imitate, thereby promoting language learning and comprehension.

[0166] It is not difficult to understand that in a noisy environment, speaking too fast may cause the listener to be unable to keep up with the speaker's rhythm, resulting in misunderstanding or missing important information. Slowing down the speed of speech can provide the listener with more time to process and understand the received information, thus ensuring the accuracy and effectiveness of communication.

[0167] Therefore, the present invention adaptively adjusts the pause duration between the played texts according to the age conditions of the interacting personnel of different ages and the environmental noise conditions, thereby obtaining a customized interval duration.

[0168] It is understandable that the server will retrieve the age adjustment coefficient, and customize the character interval duration according to the noise adjustment coefficient and the age adjustment coefficient to obtain a customized interval duration.

[0169] The custom interval duration is obtained by the following formula:

[0170] ,

[0171] in, To customize the interval duration, is the character interval length, is the noise volume, is the base volume, is the noise adjustment factor, is the volume weight value.

[0172] S34, customizing and adjusting the announcement volume based on the noise adjustment coefficient, the age adjustment coefficient and the disease type quantization value to obtain a customized volume.

[0173] It is not difficult to understand that the present invention will adaptively adjust the pause duration between the played texts according to the age, hearing conditions and environmental noise conditions of the interacting personnel of different ages, so as to obtain a customized volume. The volume needs to be increased when hearing becomes poor due to illness.

[0174] The custom volume is obtained by the following formula:

[0175] ,

[0176] in, To customize the volume, is the upper limit of the number of underlying diseases corresponding to the interacting personnel, For the The disease quantification value corresponding to the basic disease category, is the weight value of the disease category quantification value, To announce the volume, It can be a volume value preset manually.

[0177] It is not difficult to understand that the present invention will adaptively adjust the volume and speech speed according to the hearing conditions of different people, environmental noise, etc., so that the volume and speech speed are compatible with the people and the environment.

[0178] In some embodiments, it also includes:

[0179] B1, obtaining a noise waveform of the environmental noise, and when determining that the noise waveform is a regular waveform, extracting the number of waveforms in the regular waveform.

[0180] It is not difficult to understand that during the interaction process, some environmental noises are continuous and relatively regular, for example, the noise of household appliances, such as refrigerators, and the compressor will work every few minutes.

[0181] Therefore, when the server determines that the noise waveform is a regular waveform, it extracts the number of waveforms in the regular waveform. The specific determination method is to check whether the noise waveform is regular. There is existing technology here and it will not be repeated here. Some regular noise waveforms are continuous, and some are composed of many non-continuous segments. Therefore, the number of waveforms in the regular waveform will be obtained.

[0182] B2, when it is determined that the number of waveforms is greater than 1, multiple sub-waveforms in the regular waveform and the waveform interval durations of adjacent sub-waveforms are obtained, and the waveform interval duration greater than or equal to the character playback duration is obtained as the positioning placement duration.

[0183] It is not difficult to understand that when the number of waveforms is greater than 1, it means that it is discontinuous. At this time, multiple sub-waveforms in the regular waveform and the waveform interval duration of adjacent sub-waveforms will be obtained, and the interval duration between discontinuous sub-waveforms will be obtained as the waveform interval duration. Subsequently, the broadcast words can be played between the noise intervals to avoid noise.

[0184] Therefore, this solution will obtain the waveform interval duration equal to the character playback duration as the positioning placement duration, wherein the character playback duration is the playback duration of each character, for example, the playback duration of each character in "Today's weather is fine".

[0185] B3, counting the middle interference durations between adjacent positioning placement durations, the first interference duration before the first positioning placement duration, and / or the last interference duration after the last positioning placement duration.

[0186] It is understandable that the server will count the middle interference durations between adjacent positioning placement durations, the first interference duration before the first positioning placement duration, and / or the tail interference duration after the tail positioning placement duration.

[0187] It is not difficult to understand that, due to the regular noise composed of multiple non-continuous sub-waveforms, when obtaining the interval lengths between adjacent sub-waveforms, some intervals are larger and text can be broadcast, but some intervals are shorter and text cannot be broadcast.

[0188] Therefore, the noise duration during which text cannot be broadcast between the positioning placement durations during which text can be broadcast will be counted, that is, the middle interference duration corresponding to the interference segment, and when there are multiple positioning placement durations, the noise in front and at the end is also the interference duration. For example, for 10 seconds of noise, there are 2 positioning placement durations in the middle, which are 2-3s and 5-6s respectively, and the rest of the time is interference duration.

[0189] B4: Obtain interference duration according to the first interference duration, the middle interference duration and the last interference duration.

[0190] B5: taking the interference duration greater than the customized interval duration as the interference duration to be adjusted, and obtaining the positioning placement duration before the interference duration to be adjusted as the first placement duration.

[0191] It is not difficult to understand that the characters to be played are placed between the interference durations, for example, 2-3s and 5-6s, so there is a pause duration between characters. If the pause duration is less than the interference duration, it will be interfered by noise.

[0192] Therefore, the present invention will adaptively customize and adjust a portion of the character broadcast intervals according to the duration of the regular noise neutron waveform, thereby avoiding the noise stage and making the text broadcast clear.

[0193] Therefore, the server will use the interference duration greater than the customized interval duration as the interference duration to be adjusted, and obtain the positioning placement duration before the interference duration to be adjusted as the first placement duration, that is, the duration that can be broadcast.

[0194] B6, according to the ratio of the first placement time to the character playing time, obtaining a first number of played words corresponding to each of the first placement time, and counting the first number of played words to obtain a first total number of words.

[0195] It is not difficult to understand that according to the ratio of the first placement time and the character playback time, the first number of played words corresponding to each first placement time is obtained, that is, the number of words corresponding to each broadcast time. Subsequently, the number of words that can be played in all the first placement times before the interference time to be adjusted will be counted to obtain the first total number of words.

[0196] B7, based on the first total number of words, determining corresponding broadcast characters in the broadcast sentence as adjustment characters, and replacing the customized interval durations between the adjustment characters with the interference duration to be adjusted.

[0197] It is not difficult to understand that the first total number of words is the number of words that will not be affected by noise, for example, 3 words. Subsequently, the corresponding broadcast characters in the broadcast sentence will be determined as adjustment characters based on the first total number of words. That is, the first 3 words will not be affected, then the 3rd and 4th words will be used as adjustment characters. For example, today the weather is fine, and the weather and air will be used as adjustment characters.

[0198] Subsequently, the customized interval duration between the adjusted characters is replaced with the interference duration to be adjusted, thereby skipping the interference duration and making the text clear.

[0199] B8, when it is determined that the number of waveforms is equal to 1, the interval duration between adjacent regular waveforms is obtained, and when it is determined that the interval duration is greater than or equal to the character playback duration, the duration of the regular waveform is obtained as the regular interference duration, and the customized interval duration is adjusted to the regular interference duration.

[0200] It can be understood that when the server determines that the number of waveforms is equal to 1, it means that the regular waveform is a continuous waveform. At this time, the interval duration between adjacent regular waveforms is obtained. When it is determined that the interval duration is greater than or equal to the character playback duration, it means that the regular waveforms can be broadcast. The duration of the regular waveform itself is used as the regular interference duration, and the customized interval duration is adjusted to the regular interference duration. It can be adjusted according to the regular interference duration to avoid noise first.

[0201] B9, when it is determined that the noise waveform is an irregular waveform, the average of the waveform durations corresponding to multiple sub-waveforms in the irregular waveform is determined based on the preset extraction duration, the average waveform duration corresponding to the irregular waveform is obtained, and the customized interval duration is adjusted to the average waveform duration.

[0202] It can be understood that when the noise waveform is an irregular waveform, it means that the noise is extremely irregular, and then the average of the waveform interval durations corresponding to the multiple sub-waveforms in the irregular waveform within the preset extraction duration is determined based on the preset extraction duration, thereby obtaining the average waveform duration corresponding to the irregular waveform. Among them, the preset extraction duration is a manually preset duration.

[0203] The customized interval duration will be subsequently adjusted to the average waveform duration to avoid most noises, thereby realizing voice broadcasting of emotions.

[0204] S4, sending the announcement sentence to the display interface of the voice interaction terminal for display according to the display font size, and controlling the voice interaction terminal to play the announcement sentence based on the customized interval duration and the customized volume.

[0205] It is understandable that the server will send the broadcast sentence to the display interface of the voice interaction terminal for display according to the display font size, and control the voice interaction terminal to play the broadcast sentence based on the customized interval duration and the customized volume.

[0206] It is not difficult to understand that after obtaining the corresponding font size, volume and interval duration, the broadcast sentences will be customized to adapt to the corresponding interacting personnel.

[0207] See also Figure 2 , is a schematic diagram of the structure of an intelligent voice interaction device based on a large language model provided by an embodiment of the present invention, and the intelligent voice interaction device based on a large language model includes:

[0208] An analysis module is used to receive the interaction instruction sent by the voice interaction terminal and the interaction audio data corresponding to the interaction instruction, perform timbre analysis on the interaction audio data based on the large language model, generate a predicted value segment, and retrieve the broadcast sentence corresponding to the interaction instruction;

[0209] A customization module, used for retrieving basic information of a binding person set by a corresponding binding terminal based on the predicted value segment, and displaying and customizing the announcement sentence according to the basic information to obtain a display font size;

[0210] An adjustment module is used to obtain the environmental noise collected by the collection device, and customize the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise, so as to obtain the customized interval duration and customized volume of adjacent broadcast characters;

[0211] A broadcast module is used to send the broadcast sentence to the display interface of the voice interaction terminal for display according to the display font size, and control the voice interaction terminal to play the broadcast sentence based on the customized interval duration and the customized volume.

[0212] See also Figure 3 , is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention, the electronic device 30 includes: a processor 31, a memory 32 and a computer program; wherein

[0213] The memory 32 is used to store the computer program, which may also be a flash memory. The computer program is, for example, an application program, a functional module, etc. for implementing the above method.

[0214] The processor 31 is used to execute the computer program stored in the memory to implement each step performed by the device in the above method. For details, please refer to the relevant description in the above method embodiment.

[0215] Optionally, the memory 32 may be independent or integrated with the processor 31 .

[0216] When the memory 32 is a device independent of the processor 31, the device may further include:

[0217] The bus 33 is used to connect the memory 32 and the processor 31 .

[0218] The present invention also provides a readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the methods provided by the various embodiments described above.

[0219] Among them, the readable storage medium can be a computer storage medium or a communication medium. The communication medium includes any medium that facilitates the transmission of a computer program from one place to another. The computer storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application-specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). In addition, the ASIC can be located in a user device. Of course, the processor and the readable storage medium can also exist in a communication device as discrete components. The readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0220] The present invention also provides a program product, which includes an execution instruction, which is stored in a readable storage medium. At least one processor of a device can read the execution instruction from the readable storage medium, and at least one processor executes the execution instruction so that the device implements the methods provided in the above various embodiments.

[0221] In the embodiments of the above-mentioned devices, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0222] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent voice interaction method based on a large language model, characterized in that: include: Receive the interactive command sent by the voice interactive terminal, and the interactive audio data corresponding to the interactive command, perform timbre analysis on the interactive audio data based on the large language model, generate a predicted value segment, and retrieve the broadcast sentence corresponding to the interactive command; Based on the predicted value segment, basic information of the binding personnel set by the corresponding binding terminal is retrieved, and the broadcast sentence is displayed and customized according to the basic information to obtain a display font size; The acquisition device collects environmental noise, and customizes the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise, so as to obtain the customized interval duration and customized volume of adjacent broadcast characters; Sending the announcement sentence to the display interface of the voice interaction terminal for display according to the display font size, and controlling the voice interaction terminal to play the announcement sentence based on the customized interval duration and the customized volume; The method of retrieving basic information of the binding personnel set by the corresponding binding terminal based on the predicted value segment, and displaying and customizing the announcement sentence according to the basic information to obtain the display font size includes: Obtaining basic information of the bound person set by each of the binding terminals, retrieving age information in the basic information, and when determining that the age information is within the predicted value range, retrieving basic information of the bound person set by the corresponding binding terminal; Retrieving age information in the corresponding basic information, and when it is determined that the age information is greater than a maximum value of a preset age interval, obtaining a first age difference according to a difference between the age information and the maximum value of the preset age interval, wherein the preset age interval has a corresponding preset font size; Increase and adjust the preset font size according to the first age difference to obtain a display font size; The step of increasing and adjusting the preset font size according to the first age difference to obtain a display font size includes: The display font size is obtained by the following formula: ； in, To display the font size, For the default font size, For age information, is the maximum value of the preset age range, is the font size weight value.

2. The method according to claim 1, characterized in that The performing timbre analysis on the interactive audio data based on the large language model, generating a predicted value segment, and retrieving a broadcast sentence corresponding to the interactive instruction, includes: Performing timbre analysis on the interactive audio data based on a large language model to obtain a timbre stage corresponding to the interactive audio data; A preset numerical segment corresponding to the timbre stage is retrieved as a predicted numerical segment, the timbre stage having a preset numerical segment corresponding thereto, and a broadcast sentence corresponding to the interactive instruction is retrieved.

3. The method according to claim 1, characterized in that Also includes: Obtaining the interface size of the display interface corresponding to the voice interaction terminal, wherein the interface size includes the interface length and the interface width; Retrieving the display length and display width of the display font size, obtaining a first display quantity based on the ratio of the interface length to the display length, and obtaining a second display quantity based on the ratio of the interface width to the display width; Obtaining a total display quantity corresponding to the display interface according to the product of the first display quantity and the second display quantity; Retrieving the number of characters in the broadcast sentence, and when it is determined that the number of characters is greater than the total display number, obtaining the number of display pages according to the ratio of the number of characters to the total display number; Based on the ratio of the number of characters to the number of display pages, the number of page characters corresponding to each display page is obtained; The benchmark dwell time is adjusted according to the number of characters on the page and the age information to obtain the dwell time corresponding to the displayed page.

4. The method according to claim 3, characterized in that The step of adjusting the reference dwell time according to the number of characters on the page and the age information to obtain the dwell time corresponding to the displayed page includes: Obtaining a character adjustment coefficient according to a ratio of the number of characters on the page to the number of reference characters; When it is determined that the age information is not within the preset age range, retrieving the maximum value and the minimum value of the preset age range, the preset age range having a corresponding preset stay time; When it is determined that the age information is less than the minimum value, a second age difference is obtained according to a difference between the minimum value and the age information, and an age adjustment coefficient is obtained according to a ratio of the second age difference to the minimum value; When it is determined that the age information is greater than the maximum value, a first age difference is obtained according to a difference between the age information and the maximum value, and an age adjustment coefficient is obtained according to a ratio of the first age difference to the maximum value; The preset stay time is adjusted according to the age adjustment coefficient and the character adjustment coefficient to obtain the stay time corresponding to the display page.

5. The method according to claim 4, characterized in that The step of adjusting the preset stay time according to the age adjustment coefficient and the character adjustment coefficient to obtain the stay time corresponding to the display page includes: The dwell time is obtained by the following formula: ； in, For age information, is the maximum value of the preset age range, is the minimum value of the preset age range, For the length of stay, To preset the duration of stay, is the number of characters on the page, is the base number of characters, is the weight value of the number of characters, is the age weight value.

6. The method according to claim 5, characterized in that The acquisition and collection device collects environmental noise, and performs customized broadcast adjustment on the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise, to obtain the customized interval duration and customized volume of adjacent broadcast characters, including: Acquiring environmental noise collected by a collection device, retrieving age information and basic disease types in the corresponding basic information, and retrieving disease type quantification values ​​based on the basic disease types; Determining the noise volume of the environmental noise, and obtaining a noise adjustment coefficient according to a ratio of the noise volume to a reference volume; Retrieving the age adjustment coefficient, and custom-adjusting the character interval duration according to the noise adjustment coefficient and the age adjustment coefficient to obtain a custom interval duration; The broadcast volume is customized and adjusted based on the noise adjustment coefficient, the age adjustment coefficient and the disease type quantization value to obtain a customized volume.

7. The method according to claim 6, characterized in that Also includes: Acquire a noise waveform of the environmental noise, and when determining that the noise waveform is a regular waveform, extract the number of waveforms in the regular waveform; When it is determined that the number of waveforms is greater than 1, multiple sub-waveforms in the regular waveform and waveform interval durations between adjacent sub-waveforms are obtained, and the waveform interval duration greater than or equal to the character playback duration is obtained as the positioning placement duration; Counting the middle interference durations between adjacent positioning placement durations, the head interference duration before the head positioning placement duration, and / or the tail interference duration after the tail positioning placement duration; Obtaining an interference duration according to the first interference duration, the middle interference duration, and the tail interference duration; Taking the interference duration longer than the customized interval duration as the interference duration to be adjusted, and obtaining the positioning placement duration before the interference duration to be adjusted as the first placement duration; According to the ratio of the first placement time to the character playback time, a first playback word count corresponding to each of the first placement time is obtained, and the first playback word count is counted to obtain a first total word count; Determine the corresponding broadcast characters in the broadcast sentence as adjustment characters based on the first total number of characters, and replace the customized interval duration between the adjustment characters with the interference duration to be adjusted; When it is determined that the number of waveforms is equal to 1, the interval duration between adjacent regular waveforms is obtained, and when it is determined that the interval duration is greater than or equal to the character playback duration, the duration of the regular waveform is obtained as the regular interference duration, and the customized interval duration is adjusted to the regular interference duration; When it is determined that the noise waveform is an irregular waveform, the average of the waveform durations corresponding to multiple sub-waveforms in the irregular waveform is determined based on the preset extraction duration, the average waveform duration corresponding to the irregular waveform is obtained, and the customized interval duration is adjusted to the average waveform duration.

8. An intelligent voice interaction device based on a large language model, characterized in that: include: An analysis module is used to receive the interaction instruction sent by the voice interaction terminal and the interaction audio data corresponding to the interaction instruction, perform timbre analysis on the interaction audio data based on the large language model, generate a predicted value segment, and retrieve the broadcast sentence corresponding to the interaction instruction; A customization module, used for retrieving basic information of a binding person set by a corresponding binding terminal based on the predicted value segment, and displaying and customizing the announcement sentence according to the basic information to obtain a display font size; An adjustment module is used to obtain the environmental noise collected by the collection device, and customize the character interval duration and the broadcast volume of the broadcast characters in the broadcast sentence according to the corresponding basic information and the environmental noise, so as to obtain the customized interval duration and customized volume of adjacent broadcast characters; A broadcast module, used for sending the broadcast sentence to the display interface of the voice interaction terminal for display according to the display font size, and controlling the voice interaction terminal to play the broadcast sentence based on the customized interval duration and the customized volume; The method of retrieving basic information of the binding personnel set by the corresponding binding terminal based on the predicted value segment, and displaying and customizing the announcement sentence according to the basic information to obtain the display font size includes: Obtaining basic information of the bound person set by each of the binding terminals, retrieving age information in the basic information, and when determining that the age information is within the predicted value range, retrieving basic information of the bound person set by the corresponding binding terminal; Retrieving age information in the corresponding basic information, and when it is determined that the age information is greater than a maximum value of a preset age interval, obtaining a first age difference according to a difference between the age information and the maximum value of the preset age interval, wherein the preset age interval has a corresponding preset font size; Increase and adjust the preset font size according to the first age difference to obtain a display font size; The step of increasing and adjusting the preset font size according to the first age difference to obtain a display font size includes: The display font size is obtained by the following formula: ； in, To display the font size, For the default font size, For age information, is the maximum value of the preset age range, is the font size weight value.

Images