Intelligent interactive voice explanation system based on AIGC large model

Abstract

The invention relates to the technical field of voice explanation, in particular to an intelligent interactive voice explanation system based on an AIGC large model, which comprises an intelligent interactive voice explanation device main body 1, a memory and a processor are carried in the intelligent interactive voice explanation device main body 1, and a computer program is stored in the memory. The processor executes a computer program to realize the intelligent interactive voice explanation system based on the AIGC large model. The system comprises a tourist intention recognition module, a standard commentary matching module, an explanation content supplementing module and an explanation content broadcasting module. According to the invention, through converting the voice of the tourist into the text and analyzing the demand of the tourist, the matched standard explanation content can be identified in time, the correlation and pertinence of the explanation content are improved by analyzing the information coverage rate of the explanation and searching for the supplementary content, and the naturalness of voice output is enhanced by adjusting the speech speed and tone of the voice, so that the user experience is improved. And the interactive experience of tourists and the convenience of information acquisition are enriched.

Description

Technical Field

[0001] The present invention relates to the field of voice explanation technology, and in particular to an intelligent interactive voice explanation system based on an AIGC large model. Background Art

[0002] The field of voice explanation technology involves delivering information and knowledge through voice output. Voice explanation technology is used to achieve the conversion from text to speech, as well as the functions of understanding and responding to user commands. With the development of artificial intelligence, voice explanation technology has been widely used in various intelligent assistants, educational tools and customer service systems, and natural language generation technology is used to enhance the naturalness and fluency of the interactive experience. It aims to improve the convenience and effectiveness of information acquisition, especially in driving navigation, virtual assistants and barrier-free technology applications.

[0003] Among them, the intelligent interactive voice guide is a system implemented using voice explanation technology, which is designed to provide information and guidance through interactive voice response driven by artificial intelligence. The system usually includes a central processing unit for processing input voice commands and generating corresponding voice feedback. Uses include providing users with instant information explanations, such as museum exhibits, public place guidance, or educational content explanations. The system design focuses on the real-time and accuracy of user interaction, and can provide customized information output based on user inquiries, enriching the user's learning and experience process.

[0004] Traditional guides rely on preset commentaries, which make it difficult to adjust and update information in real time, causing users to feel that the content is outdated or does not match the current environment when receiving information. Traditional guides are not accurate enough in understanding users' specific needs and cannot accurately capture users' intentions and changes in needs, which limits the naturalness of interaction and the continuity of user experience, resulting in limitations in information transmission and a gap in user experience, affecting users' tour experience. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent interactive voice explanation system based on the AIGC large model.

[0006] In order to achieve the above-mentioned object, the present invention adopts the following technical scheme: an intelligent interactive voice explanation system based on an AIGC large model, comprising an intelligent interactive voice explanation device body (1), wherein the intelligent interactive voice explanation device body (1) is internally equipped with a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the intelligent interactive voice explanation system based on the AIGC large model, wherein the system comprises: The tourist intention recognition module converts the voice content into text based on the questions and requirements raised by tourists, extracts the key words and phrases in the interpretation requirements, and identifies the interpretation requirements of tourists by analyzing the voices of tourists, thus obtaining the identification results of the interpretation requirements of tourists; The standard commentary matching module searches a preset standard commentary database based on the identification result of the tourist's commentary demand, identifies the matching standard commentary content, and obtains the standard commentary matching result; The explanation content supplement module analyzes the information coverage in the existing explanation words based on the matching results of the standard explanation words and the identification results of the tourists' explanation needs, determines whether there are any missing explanation contents in the standard explanation words, uses the knowledge base of the AIGC large model to search for matching explanation supplement contents, evaluates the effectiveness of the supplement contents, selects effective supplement contents, and obtains a supplement content list; The explanation content broadcasting module selects the supplementary content matching the current environment based on the supplementary content list and the current environment of the tourists, arranges and integrates the texts in combination with the matching results of the standard explanation words, converts the texts into voices, adjusts the voice speed and pitch, performs voice explanation, and obtains the interactive explanation results; The intelligent interactive voice explanation system based on the AIGC large model is implemented by executing a computer program through a processor. The computer program is stored in a memory. The memory and the processor are mounted in the intelligent interactive voice explanation device body (1).

[0007] The present invention is improved in that the method for extracting keywords and phrases in the explanation requirements is: Based on the questions and language of the tourists, the voice content is converted into text through voice recognition, and the text content is segmented into separate words and phrases to obtain word and phrase segmentation information; Based on the vocabulary and phrase segmentation information, the formula is:

[0008] Calculate vocabulary weights ,in, It is The number of times a word appears in the text, It is The length of the word, is the total number of words in the text, is the total number of texts in the historical question dataset, It includes The number of texts with words, , and is the weight parameter, It is The weight of each word; Based on the weight of the vocabulary , compared with the preset keyword threshold, Words exceeding the preset keyword threshold are marked as keywords, and the interpretation requirement keyword recognition results are obtained.

[0009] The present invention is improved in that the steps of obtaining the identification result of the tourist interpretation demand are as follows: Based on the interpretation demand keyword recognition results, combined with the tourists' emotion recognition data, the formula is:

[0010] Calculating the tourist interpretation demand index ,in, Representative The emotional intensity index of the words, It is The semantic relevance score of each word, is the adjustment factor, It is The average emotional intensity of the words, Indicates the total number of words, An index of tourist interpretation needs; Based on the interpretation needs index of the tourists ,according to The size of the value is used to judge the intensity of the interpretation demand and obtain the identification result of tourists' interpretation needs.

[0011] The present invention is improved in that the steps of obtaining the standard commentary matching result are as follows: Based on the identification result of the tourist interpretation demand, according to the identified interpretation demand keywords, the preset standard interpretation vocabulary is retrieved, and the formula is used:

[0012] Calculate the matching score between standard commentary and commentary requirements ,in, It is Keyword weight factor, It's a commentary and query keywords The similarity score between is the total number of keywords, Score the match; Based on the matching score , by comparison The size of the value, select The standard commentary with the largest value is used as the matching standard commentary to obtain the standard commentary matching result.

[0013] The present invention is improved in that the method for judging whether the standard commentary has missing commentary content is as follows: Based on the matching results of the standard commentary and the identification results of tourists' commentary needs, the formula is:

[0014] Calculate the coverage of the standard commentary content ,in, Indicates the coverage of the explanation content. The standard commentary The vector of keywords, For tourists’ needs The vector of keywords, is the exponential parameter, is the normalization constant, The total number of keywords involved in the calculation; Based on the coverage of the commentary content , compared with the preset content coverage threshold, if If the value is lower than the content coverage threshold, it is judged that additional interpretation content is needed, and by comparing the keywords in the standard interpretation words with the keywords of tourists' interpretation needs, the keywords missing in the standard interpretation words are identified to obtain the supplementary content demand information.

[0015] The present invention is improved in that the step of obtaining the supplementary content list is: Based on the supplementary content demand information, the knowledge base of the AIGC large model is used to search for matching commentary supplementary content to obtain supplementary content retrieval results; Based on the supplementary content retrieval results, by formula:

[0016] Calculating the effectiveness score of supplement content in, It is The relevance score of content items, It is The richness score of the content items, is a weighted index, is the normalization factor, is the total number of content items, score effectiveness; Based on the effectiveness score , compared with the preset validity threshold, Supplementary content whose value exceeds a preset validity threshold is marked as valid, and a supplementary content list is obtained.

[0017] The present invention is improved in that the method for selecting supplementary content matching the current environment is: Based on the supplementary content list, the physical and cultural environment in which the tourists are currently located, including geographical location, cultural background and environmental factors, is collected to obtain real-time environmental data; Based on the real-time environmental data, the formula is:

[0018] Calculating the match score for supplemental content ,in It is The environmental adaptability weight of each content item, It is A binary indicator of the match between a content item and the environment, is the total number of content items, Score the matching of supplementary content; Match score based on the supplemental content , compared with the preset matching threshold, select The supplementary content whose value exceeds the matching degree threshold is selected to obtain the supplementary content selection result.

[0019] The present invention is improved in that the step of obtaining the interactive explanation result is: Based on the supplementary content selection result and the standard commentary matching result, the text is arranged and integrated, and the text is converted into speech to obtain a speech conversion result; Based on the speech conversion result, by formula:

[0020] Calculate the adjusted speaking rate and use the formula:

[0021] Calculate the adjusted pitch, where is the baseline speaking speed, is the reference tone, is the speech rate adjustment coefficient, is the pitch adjustment coefficient, is the environmental impact factor, is the content type factor, is the normalization factor of speech rate, is the base of natural logarithms, is the adjusted speaking speed, is the adjusted pitch; Based on the adjusted speech speed and the adjusted pitch, voice interpretation is performed to obtain an interactive interpretation result.

[0022] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by converting tourists' voice into text, extracting key information in tourists' needs, and analyzing tourists' needs, it is possible to timely identify matching standard interpretation content, and by analyzing the information coverage of the interpretation words, searching for supplementary content, enriching the detailedness and educational value of the interpretation materials, and improving the relevance and pertinence of the interpretation content, and by adjusting the speech speed and tone of the voice, the naturalness of the voice output is enhanced, creating a more immersive and personalized experience for tourists, and enriching the tourists' interactive experience and the convenience of information acquisition. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a system flow chart of the present invention; Figure 2 A flowchart for extracting keywords and phrases from the interpretation requirements for the present invention; Figure 3 A flow chart of obtaining identification results of tourist interpretation needs in the present invention; Figure 4 A flow chart of obtaining standard commentary matching results for the present invention; Figure 5 A flow chart of the present invention for determining whether a standard commentary has missing commentary content; Figure 6 A flowchart for obtaining a list of supplementary contents for the present invention; Figure 7 A flow chart for selecting supplementary content matching the current environment for the present invention; Figure 8 A flow chart for obtaining interactive explanation results for the present invention; Fig. 9 It is a structural schematic diagram of the main body of the intelligent interactive voice interpreter of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0025] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating positions or positional relationships, are based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0026] See also Figure 1 and Fig. 9 The present invention provides a technical solution: an intelligent interactive voice explanation system based on an AIGC large model, the system is implemented by a processor executing a computer program, the computer program is stored in a memory, and the memory and the processor are mounted in an intelligent interactive voice explanation device body (1); the system includes: The tourist intention recognition module converts the voice content into text based on the questions and requirements raised by tourists, performs semantic analysis on the text, extracts the key words and phrases in the interpretation requirements, and identifies the interpretation requirements of tourists by analyzing the voices of tourists to obtain the identification results of the interpretation requirements of tourists; The standard commentary matching module searches the preset standard commentary database based on the identification result of the tourist's commentary demand, identifies the matching standard commentary content, and obtains the standard commentary matching result; The interpretation content supplement module analyzes the information coverage of the existing interpretation words based on the matching results of the standard interpretation words and the identification results of the interpretation needs of tourists, determines whether there are any missing interpretation contents in the standard interpretation words, uses the knowledge base of the AIGC large model to search for matching interpretation supplement contents, and evaluates the effectiveness of the supplement contents according to the richness of the supplement contents, selects the effective supplement contents, and obtains the supplement content list; The interpretation content broadcasting module selects supplementary content that matches the current environment based on the supplementary content list and the current environment of tourists, and arranges and integrates the text in combination with the matching results of the standard interpretation words, converts the text into speech, adjusts the speech speed and pitch, and performs voice interpretation to obtain interactive interpretation results; The results of identifying tourists' interpretation needs include emotion type, intention category and urgency; the results of matching standard interpretation words include selected interpretation words, matching scores and interpretation word sources; the supplementary content list includes supplementary content items, content suitability assessment and content richness assessment; the interactive interpretation results include interpretation voice files and content adaptation information.

[0027] See also Figure 2 , the method to extract keywords and phrases in the explanation requirements is: Based on the questions and language of the tourists, the voice content is converted into text through voice recognition, and the text content is segmented into separate words and phrases to obtain word and phrase segmentation information; Based on vocabulary and phrase segmentation information, the formula is:

[0028] Calculate vocabulary weights in, It is The number of times a word appears in the text, It is The length of the word, is the total number of words in the text, is the total number of texts in the historical question dataset, It includes The number of texts with words, , and is the weight parameter, It is The weight of each word; Word-based weights , compared with the preset keyword threshold,

[0029] Words exceeding the preset keyword threshold are marked as keywords, and the interpretation requirement keyword recognition results are obtained.

[0030] formula:

[0031] Parameter meaning and acquisition method: : No. The number of times a word appears in the text. The parameter is usually obtained through text scanning statistics, that is, the text is read word by word and counted. When a word is found, the count of that word is increased.

[0032] No. The parameter is determined by measuring the number of characters or words in the word, with each character being uniformly counted according to the encoding standard.

[0033] : The total number of words in the text. This is the total number obtained by counting all the words after performing tokenization on the entire text.

[0034] : The total number of texts in the historical question dataset. This parameter is obtained by counting the number of text files in the entire dataset.

[0035] : It includes The number of texts in the vocabulary. The parameter is calculated by searching all the texts in the dataset and counting the texts containing the first The text of each vocabulary is determined.

[0036] , and : Weight adjustment parameters. The parameters are used to fine-tune the weight calculation formula. They are set according to different text processing requirements and can be preset by experts.

[0037] Calculation example: Set up the following situation: The word "example" appears 3 times, with a total of 100 words. There are 50 texts in the historical question text set, 10 of which contain "example". , , .

[0038] Calculate the weight of "example" : , (length of "example"), , , .

[0039] Substituting into the formula, the process uses natural logarithms:

[0040] The calculation results show that the weight of the word "example" is 0.0951 according to the given text and parameter settings. The weight can be used to evaluate the importance or criticality of a word in a text. Words with high weights have a greater impact on the description of the text topic.

[0041] See also Figure 3 ,The steps to obtain the identification results of tourists’ interpretation needs are: Based on the interpretation demand keyword recognition results, combined with the tourist emotion recognition data, the formula is:

[0042] Calculating the tourist interpretation demand index ,in, Representative The emotional intensity index of the words, It is The semantic relevance score of each word, is the adjustment factor, It is The average emotional intensity of the words, Indicates the total number of words, An index of tourist interpretation needs; Visitors’ interpretation needs index ,according to The size of the value is used to judge the intensity of the interpretation demand and obtain the identification result of tourists' interpretation needs.

[0043] formula:

[0044] Parameter details and how to obtain them: : Emotional intensity index, obtained by analyzing the tone, intensity and rhythm of tourists' voices. The analysis usually uses speech recognition and sentiment analysis software to infer emotional states by analyzing the audio characteristics of speech.

[0045] : Semantic relevance score, obtained by semantically analyzing the text content using natural language processing technology. The analysis includes keyword extraction, context understanding, and relevance assessment to determine the degree of relevance of the vocabulary to the visitor's query topic.

[0046] : Adjustment coefficient, used to balance the impact of sentiment variation on the correlation calculation. The coefficient is determined based on historical data and statistical analysis, and aims to optimize the prediction accuracy and stability of the model.

[0047] : The average value of emotional intensity, calculated as This is obtained by summing the intensity values ​​of all participating emotions and dividing by the number of emotion samples.

[0048] : The total number of data points analyzed, i.e. the total number of words or phrases involved in the calculation. The number is determined directly from the size of the dataset and represents the number of all text fragments analyzed and processed.

[0049] Calculation example: Set the parameters as follows: , representing 5 data points.

[0050] , , , , indicating the sentiment intensity of each data point.

[0051] , , , , , which represents the semantic relevance of each data point.

[0052] , , the average value of emotion intensity.

[0053] Calculation process:

[0054] Calculation results , its value is large, indicating that tourists have a strong demand for interpretation and that corresponding interpretation services need to be provided to meet tourists’ detailed needs.

[0055] See also Figure 4 , the steps to obtain the standard commentary matching results are: Based on the identification results of tourists' interpretation needs, according to the identified interpretation demand keywords, the preset standard interpretation vocabulary is retrieved, and the formula is used:

[0056] Calculate the matching score between standard commentary and commentary requirements ,in, It is Keyword weight factor, It's a commentary and query keywords The similarity score between is the total number of keywords, Score the matching degree; Based on matching score , by comparison The size of the value, select The standard commentary with the largest value is used as the matching standard commentary to obtain the standard commentary matching result.

[0057] formula:

[0058] Parameter meaning and acquisition method: :It is The weight factor of a keyword in the matching process. The weight is obtained through statistical analysis based on the influence of each keyword on the interpretation effect and the frequency of occurrence in the historical data. The purpose of setting the weight is to enhance the influence of important keywords and ensure that the keywords occupy a more important position in the final matching score.

[0059] :Indicates the Commentary With query The semantic similarity between words. The similarity is usually calculated by using the cosine similarity algorithm in natural language processing technology to ensure accurate assessment of the semantic association between words.

[0060] : The total number of keywords involved in the scoring. The number is determined directly from the number of keywords involved in the commentary entry, and is usually preset based on the level of detail and richness of the commentary entry.

[0061] Calculation example: Set the following parameters and queries: The commentary entry contains three keywords: .

[0062] The keyword weight is set to .

[0063] The similarities between the three keywords and the query are , , .

[0064] Use the formula to calculate:

[0065] result It means that according to the given weight and similarity calculation, the overall semantic similarity score between the commentary entry and the query is 0.22. The score is used to evaluate the matching degree between different commentary entries and tourists' needs. The higher the score, the higher the matching degree between the commentary and tourists' needs. In actual application, the commentary with the highest similarity score will be selected to meet tourists' interpretation needs.

[0066] See also Figure 5 , the method to determine whether there is any missing commentary content in the standard commentary is: Based on the matching results of standard commentary and identification results of tourist commentary needs, the formula is:

[0067] Calculate the coverage of the standard commentary content ,in, Indicates the coverage of the explanation content. The standard commentary The vector of keywords, For tourists’ needs The vector of keywords, Exponential parameters, is the normalization constant, The total number of keywords involved in the calculation; Based on the coverage of the commentary content , compared with the preset content coverage threshold, if If the value is lower than the content coverage threshold, it is judged that additional interpretation content is needed, and by comparing the keywords in the standard interpretation words with the keywords of tourists' interpretation needs, the keywords missing in the standard interpretation words are identified to obtain the supplementary content demand information.

[0068] formula:

[0069] Parameter details and how to obtain them: :The first The vector representation of the keyword. The vector is obtained by applying word embedding technology to the commentary text. Word embedding models such as Word2Vec or BERT can convert words into vectors in semantic space for mathematical operations and similarity calculations.

[0070] :The first in tourists’ needs The vector representation of each keyword is extracted from the natural language text input by the user through the word embedding model, ensuring that the semantics of each word is accurately captured.

[0071] :The total number of keywords involved in the calculation depends on the number of keywords extracted from the standard commentary and tourist needs. Usually it is the sum of all words in the commentary text and user needs, or the total number of key words that have been screened.

[0072] : An exponential parameter that emphasizes differences, used to amplify the impact of differences in keyword vectors. The choice of parameter depends on the desired sensitivity, with higher values ​​making large differences more significant.

[0073] : A regularization constant used to adjust the overall scale of the difference measure, usually based on The value of is adjusted dynamically to accommodate data sets of different sizes.

[0074] Calculation example: Set the following parameters: Three keywords: ,set up and , set the keyword vector difference as follows:

[0075] The sum of squares of the differences is calculated as follows:

[0076] calculate :

[0077] Calculation results The coverage rate of the commentary content is very high, almost completely covering the needs of tourists. This shows that the current commentary almost completely meets the expectations of tourists and only requires little or no supplementary content.

[0078] See also Figure 6 , the steps to obtain the supplementary content list are: Based on the supplementary content demand information, the AIGC large model knowledge base is used to search for matching supplementary content and obtain supplementary content retrieval results; Based on the supplementary content retrieval results, the formula is:

[0079] Calculating the effectiveness score of supplement content ,in, It is The relevance score of content items, It is The richness score of the content items, is a weighted index, is the normalization factor, is the total number of content items, score effectiveness; Based on effectiveness score , compared with the preset validity threshold, Supplementary content whose value exceeds a preset validity threshold is marked as valid, and a supplementary content list is obtained.

[0080] formula:

[0081] Parameter details and how to obtain them: : Relevance scoring of content items, achieved by comparing how well the content items match the keywords in the requirement, using standard text similarity measures such as TF-IDF or cosine similarity.

[0082] : Content richness score, which evaluates the amount of information and level of detail of supplementary content. The score is calculated based on the length, complexity, and variety of information provided by the content, and is usually evaluated by automated text analysis tools to assess the depth and breadth of the content.

[0083] : A weighted index used to amplify differences in ratings. The parameters are set based on previous experiments and optimizations to ensure that the system's response to changes in content quality is both sensitive and reasonable.

[0084] : Normalization factor used to adjust the scale of the score based on the number of content items The expected maximum score is used to calculate the score to maintain the uniformity and comparability of the scoring results.

[0085] : Total number of content items, indicating the total number of item types included from the supplementary content.

[0086] Calculation example: Set the following data: Number of supplementary content items , relevance score The content richness scores are 0.8, 0.6, and 0.9 respectively.

[0087] The weighted index is 0.7, 0.5, and 0.8 respectively. , normalization factor , the effect evaluation threshold is 0.7.

[0088]

[0089] result , which means that the score is relatively low, far below the set threshold of 0.7, indicating that the effectiveness of the current supplementary content is insufficient and that it is necessary to optimize or add higher-quality supplementary content to meet the tourists' interpretation needs.

[0090] See also Figure 7 , the method to select supplementary content that matches the current environment is: Based on the supplementary content list, collect the tourists' current physical and cultural environment, including geographical location, cultural background and environmental factors, to obtain real-time environmental data; Based on real-time environmental data, through the formula:

[0091] Calculating the match score for supplemental content ,in It is The environmental adaptability weight of each content item, It is A binary indicator of the match between a content item and the environment, is the total number of content items, Score the matching of supplementary content; Match score based on supplementary content , compared with the preset matching threshold, select The supplementary content whose value exceeds the matching degree threshold is selected to obtain the supplementary content selection result.

[0092] formula:

[0093] Parameter details and how to obtain them: :Environmental adaptability weight is a parameter that measures the adaptability of a content item to a specific environment. The weight is preset based on the type of content (such as history, culture, nature, etc.) and the needs of the environment. The weight value is usually adjusted by analyzing past visitor feedback and content performance to ensure that it effectively reflects the adaptability of the content.

[0094] : A binary indicator of the match between the content item and the environment. The indicator indicates whether the content is suitable for the current environment, where 1 represents a match and 0 represents a mismatch. This is determined by analyzing the relevance of the content's keywords and themes to the current environment (such as geographic location, cultural background, etc.).

[0095] : The number of supplementary content items evaluated. This number is determined based on the length of the list of supplementary content that is actually available for evaluation, which usually depends on how relevant the content retrieved from the big data model is to the current context.

[0096] Calculation example: Set the following parameter settings: Environmental adaptability weight 0.8, 0.6, 0.9 respectively, content matching index 1, 0, 1 respectively, the number of supplementary content items evaluated , the matching threshold is 0.7.

[0097] Calculate the overall match:

[0098] Calculation results , the result is lower than the threshold value of 0.7, which means that the matching degree of the currently selected supplementary content does not meet the preset standard and other supplementary content needs to be selected.

[0099] See also Figure 8 , the steps to obtain the interactive explanation results are: Based on the results of the supplementary content selection and the standard commentary matching results, the text is arranged and integrated, and the text is converted into speech to obtain the speech conversion results; Based on the speech conversion results, through the formula:

[0100] Calculate the adjusted speaking rate and use the formula:

[0101] Calculate the adjusted pitch, where is the baseline speaking speed, is the reference tone, is the speech rate adjustment coefficient, is the pitch adjustment coefficient, is the environmental impact factor, is the content type factor, is the normalization factor of speech rate, is the base of natural logarithms, is the adjusted speaking speed, is the adjusted pitch; Based on the adjusted speaking speed and the adjusted pitch, voice interpretation is performed to obtain an interactive interpretation result.

[0102] formula: and

[0103] Parameter details and how to obtain them: : Baseline speaking speed, preset according to the average speaking speed of a standard audience, suitable for most audience groups, especially in non-technical or general commentary content.

[0104] : Baseline pitch, set based on average adult speech frequency.

[0105] : The speech speed adjustment coefficient depends on the urgency or complexity of the content. For urgent information that needs to be delivered quickly, the speech speed adjustment coefficient needs to be increased.

[0106] : Tone adjustment coefficient, adjusted according to the emotional needs of the content. For example, if serious content needs to be expressed, the tone adjustment coefficient needs to be increased.

[0107] :Environmental impact factor, adjusted according to the noise level of the interpretation environment, and adjusted and set according to the current interpretation environment (such as noise level, indoor and outdoor environment).

[0108] Content type factor, adjusted according to the nature of the content (such as educational or entertainment).

[0109] Normalization factor for speech rate, used to ensure that the speech rate remains within the acceptable range after adjustment.

[0110] Calculation example: Set the following parameters: Standard speaking speed , reference tone , speech rate adjustment coefficient , pitch adjustment coefficient , environmental impact factors (slightly noisy environment), content type factor , the regularization factor .

[0111] Calculate adjusted speech rate and Tone , the process uses natural logarithms:

[0112] The calculation results show that the speech rate is increased to about 167wpm and the pitch is increased to about 130Hz to adapt to the slightly noisy environment, ensuring that the content of the explanation is clear and easy to understand, while enhancing the efficiency of the audience's reception and understanding.

[0113] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent interactive voice explanation system based on AIGC large model, characterized in that: The invention comprises an intelligent interactive voice interpreter body (1), wherein the intelligent interactive voice interpreter body (1) is internally equipped with a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement an intelligent interactive voice interpreter system based on an AIGC large model, wherein the system comprises: The tourist intention recognition module converts the voice content into text based on the questions and requirements raised by tourists, extracts the key words and phrases in the interpretation requirements, and identifies the interpretation requirements of tourists by analyzing the voices of tourists, thus obtaining the identification results of the interpretation requirements of tourists; The standard commentary matching module searches a preset standard commentary database based on the identification result of the tourist's commentary demand, identifies the matching standard commentary content, and obtains the standard commentary matching result; The explanation content supplement module analyzes the information coverage in the existing explanation words based on the matching results of the standard explanation words and the identification results of the tourists' explanation needs, determines whether there are any missing explanation contents in the standard explanation words, uses the knowledge base of the AIGC large model to search for matching explanation supplement contents, evaluates the effectiveness of the supplement contents, selects effective supplement contents, and obtains a supplement content list; The explanation content broadcasting module selects the supplementary content matching the current environment based on the supplementary content list and the current environment of the tourists, arranges and integrates the texts in combination with the matching results of the standard explanation words, converts the texts into voices, adjusts the voice speed and pitch, performs voice explanation, and obtains the interactive explanation results; The intelligent interactive voice explanation system based on the AIGC large model is implemented by executing a computer program through a processor. The computer program is stored in a memory. The memory and the processor are mounted in the intelligent interactive voice explanation device body (1).

2. The intelligent interactive voice explanation system based on the AIGC large model according to claim 1 is characterized in that: The method for extracting keywords and phrases in the explanation requirements is: Based on the questions and language of the tourists, the voice content is converted into text through voice recognition, and the text content is segmented into separate words and phrases to obtain word and phrase segmentation information; Based on the vocabulary and phrase segmentation information, the formula is: ； Calculate vocabulary weights ,in, It is The number of times a word appears in the text, It is The length of the word, is the total number of words in the text, is the total number of texts in the historical question dataset, It includes The number of texts with words, , and is the weight parameter, It is The weight of each word; Based on the weight of the vocabulary Compare with the preset keyword threshold and Words that exceed the preset keyword threshold are marked as keywords, and the interpretation requirement keyword recognition results are obtained.

3. The intelligent interactive voice explanation system based on the AIGC large model according to claim 2 is characterized in that: The steps for obtaining the identification result of the tourist interpretation demand are as follows: Based on the interpretation demand keyword recognition results, combined with the tourists' emotion recognition data, the formula is: ； Calculating the tourist interpretation demand index ,in, Representative The emotional intensity index of the words, It is The semantic relevance score of each word, is the adjustment factor, It is The average emotional intensity of the words, Indicates the total number of words, An index of tourist interpretation needs; Based on the interpretation needs index of the tourists ,according to The size of the value is used to judge the intensity of the interpretation demand and obtain the identification result of tourists' interpretation needs.

4. The intelligent interactive voice explanation system based on the AIGC large model according to claim 1 is characterized in that: The steps for obtaining the standard commentary matching results are as follows: Based on the identification result of the tourist interpretation demand, according to the identified interpretation demand keywords, the preset standard interpretation vocabulary is retrieved, and the formula is used: ; Calculate the matching score between standard commentary and commentary requirements ,in, It is Keyword weight factor, It's a commentary and query keywords The similarity score between is the total number of keywords, Score the match; Based on the matching score , by comparison The size of the value, select The standard commentary with the largest value is used as the matching standard commentary to obtain the standard commentary matching result.

5. The intelligent interactive voice explanation system based on the AIGC large model according to claim 1 is characterized in that: The method for judging whether the standard commentary contains any missing commentary content is as follows: Based on the matching results of the standard commentary and the identification results of tourists' commentary needs, the formula is: ； Calculate the coverage of the standard commentary content ,in, Indicates the coverage of the explanation content. The standard commentary The vector of keywords, For tourists’ needs The vector of keywords, is the exponential parameter, is the normalization constant, The total number of keywords involved in the calculation; Based on the coverage of the commentary content , compared with the preset content coverage threshold, if If the value is lower than the content coverage threshold, it is judged that additional interpretation content is needed, and by comparing the keywords in the standard interpretation words with the keywords of tourists' interpretation needs, the keywords missing in the standard interpretation words are identified to obtain the supplementary content demand information.

6. The intelligent interactive voice explanation system based on the AIGC large model according to claim 5 is characterized in that: The steps of obtaining the supplementary content list are: Based on the supplementary content demand information, the knowledge base of the AIGC large model is used to search for matching commentary supplementary content to obtain supplementary content retrieval results; Based on the supplementary content retrieval results, by formula: ; Calculating the effectiveness score of supplement content in, It is The relevance score of content items, It is The richness score of the content items, is a weighted index, is the normalization factor, is the total number of content items, Score effectiveness; Based on the effectiveness score , compared with the preset validity threshold, Supplementary content whose value exceeds a preset validity threshold is marked as valid, and a supplementary content list is obtained.

7. The intelligent interactive voice explanation system based on the AIGC large model according to claim 1 is characterized in that: The method for selecting supplementary content matching the current environment is: Based on the supplementary content list, the physical and cultural environment in which the tourists are currently located, including geographical location, cultural background and environmental factors, is collected to obtain real-time environmental data; Based on the real-time environmental data, the formula is: ; Calculating the match score for supplemental content ,in It is The environmental adaptability weight of each content item, It is A binary indicator of the match between a content item and the environment, is the total number of content items, Score the matching of supplementary content; Match score based on the supplemental content , compared with the preset matching threshold, select The supplementary content whose value exceeds the matching degree threshold is selected to obtain the supplementary content selection result.

8. The intelligent interactive voice explanation system based on the AIGC large model according to claim 7 is characterized in that: The steps for obtaining the interactive explanation result are: Based on the supplementary content selection result and the standard commentary matching result, the text is arranged and integrated, and the text is converted into speech to obtain a speech conversion result; Based on the speech conversion result, by formula: ; Calculate the adjusted speaking rate and use the formula: ; Calculate the adjusted pitch, where is the baseline speaking speed, is the reference tone, is the speech rate adjustment coefficient, is the pitch adjustment coefficient, is the environmental impact factor, is the content type factor, is the normalization factor of speech rate, is the base of natural logarithms, is the adjusted speaking speed, is the adjusted pitch; Based on the adjusted speech speed and the adjusted pitch, voice interpretation is performed to obtain an interactive interpretation result.

Images