A method and device for identifying vocal cord disease status based on AI and audio

Abstract

The present application relates to a method and device for identifying vocal cord disease status based on AI and audio. The method includes: collecting the user's vocal audio data and historical medical information, and identifying the user's tone data and sound quality data based on the vocal audio data; analyzing the user's historical vocal cord disease information, and identifying the user's vocal cord abnormality data through an artificial neural network. Then, based on the above information, the probability value of each vocal cord disease type of the user is identified through a vocal cord disease map, and based on the abnormal data value of each abnormal type in the user's vocal cord abnormality data, the user's disease severity information in each vocal cord disease type is identified; the probability value and disease severity information of each vocal cord disease type of the user are used as the user's vocal cord disease status reference information, and then sent to the medical staff client. The use of this method can comprehensively improve the comprehensiveness and accuracy of the identified user's vocal cord disease types and vocal cord disease severity.

Description

Technical Field

[0001] The present application relates to the technical field of auxiliary intelligent analysis for vocal cord disease identification, and in particular to a method and device for identifying vocal cord disease status based on AI and audio. Background Art

[0002] Vocal cord diseases are mainly caused by vocal cord lesions, which include vocal cord hypertrophy, vocal cord nodules, and vocal cord paralysis. In order to better treat vocal cord lesions, better detection and analysis of vocal cord lesions are needed. Therefore, how to improve the detection effect of vocal cord diseases is the current research focus.

[0003] Traditional technology uses medical instruments to detect and medical staff to conduct a comprehensive analysis and judgment of the user's vocal cord condition through empty ears to determine the state of vocal cord disease. However, this identification method is affected by personal experience, resulting in low accuracy in the identification of vocal cord disease types and the severity of vocal cord diseases. Summary of the invention

[0004] Based on this, it is necessary to provide a method and device for identifying vocal cord disease status based on AI and audio to address the above technical problems.

[0005] In a first aspect, the present application provides a method for identifying vocal cord disease states based on AI and audio. The method comprises:

[0006] Collecting the user's voice audio data and the user's historical medical information, and identifying the user's tone data and the user's sound quality data based on the voice audio data;

[0007] Based on the user's historical medical information, analyzing the user's historical vocal cord disease information, and based on the tone data and the sound quality data, identifying the user's vocal cord abnormality data through an artificial neural network;

[0008] Based on the historical vocal cord disease information of the user and the vocal cord abnormality data of the user, the probability value of each vocal cord disease type of the user is identified through a vocal cord disease atlas, and based on the abnormal data value of each abnormal type in the vocal cord abnormality data of the user, the disease degree information of the user in each vocal cord disease type is identified;

[0009] The probability value of each vocal cord disease type of the user and the disease severity information of the user in each vocal cord disease type are used as the vocal cord disease status reference information of the user, and the vocal cord disease status reference information is sent to the medical staff client.

[0010] Optionally, the identifying the tone data of the user and the sound quality data of the user based on the utterance audio data includes:

[0011] Performing Fourier transform processing on the utterance audio data to obtain an audio spectrum of the utterance audio data, and converting the audio spectrum of the utterance audio data into a waveform graph;

[0012] Collect the user's pronunciation text and the regular waveform corresponding to the normal pronunciation of the pronunciation text, and identify the sub-waveform corresponding to the audio segment of each text content in the pronunciation text of the user in the waveform converted from the audio spectrum of each of the utterance audio data;

[0013] Based on the sub-waveforms corresponding to the audio segments, identifying the audio amplitude range of the user and the audio frequency distribution information of each sub-waveform, and based on the audio amplitude range of the user, identifying the tone data of the user;

[0014] Based on the audio frequency distribution information, the audio features of the user are extracted through an audio feature extraction network, and based on the audio features, the sound quality data of the user is identified through a feature analysis strategy.

[0015] Optionally, the identifying the sound quality data of the user through a feature analysis strategy based on each of the audio features includes:

[0016] Based on each of the audio features, identifying the abnormal audio frequency band data of the user and the abnormal audio fluctuation data of the user;

[0017] Arrange each of the audio abnormal frequency band data according to the position information of each of the audio abnormal frequency band data in the voiced audio data to obtain the audio abnormal frequency band sequence of the user, and use each of the audio abnormal fluctuation data and the audio abnormal frequency band sequence as the sound quality data of the user.

[0018] Optionally, analyzing the user's historical vocal cord disease information based on the user's historical medical information includes:

[0019] In the historical medical information, each vocal cord medical information is screened, and the historical vocal cord disease type corresponding to each vocal cord medical information and the historical disease severity information corresponding to each historical vocal cord disease type are identified;

[0020] Each of the historical vocal cord disease types and the historical disease severity information corresponding to each of the historical vocal cord disease types are used as the historical vocal cord disease information of the user.

[0021] Optionally, identifying abnormal vocal cord data of the user through an artificial neural network based on the tone data and the sound quality data includes:

[0022] Based on the pitch data, identifying the pitch range of the user, and based on the pitch range of the user, querying a vocal cord abnormality database for a vocal cord abnormality parameter corresponding to the pitch range;

[0023] Based on the vocal cord abnormality parameters, an initial artificial neural network is adjusted to obtain an artificial neural network, and each of the audio abnormal fluctuation data and the audio abnormal frequency band sequence is input into the artificial neural network to analyze the first vocal cord abnormality information corresponding to each of the audio abnormal fluctuation data and the second vocal cord abnormality information corresponding to the audio abnormal frequency band sequence;

[0024] Collecting the first weight corresponding to each of the audio abnormal fluctuation data and the second weight corresponding to the audio abnormal frequency band sequence, and performing weighted processing on each of the first vocal cord abnormal information and each of the second vocal cord abnormal information based on the first weight and the second weight to obtain a probability value of each of the vocal cord abnormal information;

[0025] Based on the probability value of the same vocal cord abnormality information and the number of the same vocal cord abnormality information, the target probability value of each vocal cord abnormality information is calculated, and the vocal cord abnormality information corresponding to the maximum target probability value is selected as the vocal cord abnormality data corresponding to the pitch range.

[0026] Optionally, the identifying the probability value of each vocal cord disease type of the user through a vocal cord disease atlas based on the historical vocal cord disease information of the user and the vocal cord abnormality data of the user includes:

[0027] Extracting each data feature of the vocal cord abnormality data, and based on each of the historical vocal cord disease types, adjusting the weight coefficient of the sub-branch atlas corresponding to each of the vocal cord disease types in the vocal cord disease atlas to obtain a target weight coefficient of each of the sub-branch atlases;

[0028] Based on the target weight coefficient of each sub-branch map, the similarity between each data feature and the disease feature of the branch node of each sub-branch map is calculated, and the branch node to which the disease feature having a similarity greater than a similarity threshold belongs is selected as the target branch node;

[0029] Based on the vocal cord disease type to which the target branch node corresponding to each of the data features belongs, the vocal cord disease types of the user are identified, and based on the similarity corresponding to each of the data features, the probability value of each of the vocal cord disease types is calculated.

[0030] Optionally, the identifying the disease severity information of the user in each vocal cord disease type based on the abnormal data value of each abnormal type in the vocal cord abnormal data of the user includes:

[0031] Identify a characteristic value corresponding to each data feature, and based on the disease feature corresponding to the data feature, identify a disease impact parameter corresponding to the characteristic value;

[0032] In the database, query the parameter range to which the disease influencing parameter corresponding to each characteristic value belongs, and the initial disease degree information corresponding to each parameter range, and collect the weight value of each disease feature corresponding to each vocal cord disease type;

[0033] Based on the weight value of each of the disease characteristics and the initial disease severity information corresponding to each of the disease characteristics, the disease severity information of the user in each vocal cord disease type is calculated.

[0034] In a second aspect, the present application also provides a device for identifying vocal cord disease states based on AI and audio. The device comprises:

[0035] A collection module, used to collect the user's voice audio data and the user's historical medical information, and identify the user's tone data and the user's sound quality data based on the voice audio data;

[0036] An analysis module, configured to analyze the user's historical vocal cord disease information based on the user's historical medical information, and identify the user's vocal cord abnormality data through an artificial neural network based on the tone data and the sound quality data;

[0037] an identification module, for identifying the probability value of each vocal cord disease type of the user based on the historical vocal cord disease information of the user and the vocal cord abnormality data of the user through a vocal cord disease atlas, and identifying the disease degree information of the user in each vocal cord disease type based on the abnormal data value of each abnormal type in the vocal cord abnormality data of the user;

[0038] A determination module is used to use the probability value of each vocal cord disease type of the user and the disease severity information of the user in each vocal cord disease type as the vocal cord disease status reference information of the user, and send the vocal cord disease status reference information to the medical staff client.

[0039] Optionally, the acquisition module is specifically used to:

[0040] Performing Fourier transform processing on the utterance audio data to obtain an audio spectrum of the utterance audio data, and converting the audio spectrum of the utterance audio data into a waveform graph;

[0041] Collect the user's pronunciation text and the regular waveform corresponding to the normal pronunciation of the pronunciation text, and identify the sub-waveform corresponding to the audio segment of each text content in the pronunciation text of the user in the waveform converted from the audio spectrum of each of the utterance audio data;

[0042] Based on the sub-waveforms corresponding to the audio segments, identifying the audio amplitude range of the user and the audio frequency distribution information of each sub-waveform, and based on the audio amplitude range of the user, identifying the tone data of the user;

[0043] Based on the audio frequency distribution information, the audio features of the user are extracted through an audio feature extraction network, and based on the audio features, the sound quality data of the user is identified through a feature analysis strategy.

[0044] Optionally, the acquisition module is specifically used to:

[0045] Based on each of the audio features, identifying the abnormal audio frequency band data of the user and the abnormal audio fluctuation data of the user;

[0046] Arrange each of the audio abnormal frequency band data according to the position information of each of the audio abnormal frequency band data in the voiced audio data to obtain the audio abnormal frequency band sequence of the user, and use each of the audio abnormal fluctuation data and the audio abnormal frequency band sequence as the sound quality data of the user.

[0047] Optionally, the analysis module is specifically used to:

[0048] In the historical medical information, each vocal cord medical information is screened, and the historical vocal cord disease type corresponding to each vocal cord medical information and the historical disease severity information corresponding to each historical vocal cord disease type are identified;

[0049] Each of the historical vocal cord disease types and the historical disease severity information corresponding to each of the historical vocal cord disease types are used as the historical vocal cord disease information of the user.

[0050] Optionally, the analysis module is specifically used to:

[0051] Based on the pitch data, identifying the pitch range of the user, and based on the pitch range of the user, querying a vocal cord abnormality database for a vocal cord abnormality parameter corresponding to the pitch range;

[0052] Based on the vocal cord abnormality parameters, an initial artificial neural network is adjusted to obtain an artificial neural network, and each of the audio abnormal fluctuation data and the audio abnormal frequency band sequence is input into the artificial neural network to analyze the first vocal cord abnormality information corresponding to each of the audio abnormal fluctuation data and the second vocal cord abnormality information corresponding to the audio abnormal frequency band sequence;

[0053] Collecting the first weight corresponding to each of the audio abnormal fluctuation data and the second weight corresponding to the audio abnormal frequency band sequence, and performing weighted processing on each of the first vocal cord abnormal information and each of the second vocal cord abnormal information based on the first weight and the second weight to obtain a probability value of each of the vocal cord abnormal information;

[0054] Based on the probability value of the same vocal cord abnormality information and the number of the same vocal cord abnormality information, the target probability value of each vocal cord abnormality information is calculated, and the vocal cord abnormality information corresponding to the maximum target probability value is selected as the vocal cord abnormality data corresponding to the pitch range.

[0055] Optionally, the identification module is specifically used to:

[0056] Extracting each data feature of the vocal cord abnormality data, and based on each of the historical vocal cord disease types, adjusting the weight coefficient of the sub-branch atlas corresponding to each of the vocal cord disease types in the vocal cord disease atlas to obtain a target weight coefficient of each of the sub-branch atlases;

[0057] Based on the target weight coefficient of each sub-branch map, the similarity between each data feature and the disease feature of the branch node of each sub-branch map is calculated, and the branch node to which the disease feature having a similarity greater than a similarity threshold belongs is selected as the target branch node;

[0058] Based on the vocal cord disease type to which the target branch node corresponding to each of the data features belongs, the vocal cord disease types of the user are identified, and based on the similarity corresponding to each of the data features, the probability value of each of the vocal cord disease types is calculated.

[0059] Optionally, the identification module is specifically used to:

[0060] Identify a characteristic value corresponding to each data feature, and based on the disease feature corresponding to the data feature, identify a disease impact parameter corresponding to the characteristic value;

[0061] In the database, query the parameter range to which the disease influencing parameter corresponding to each characteristic value belongs, and the initial disease degree information corresponding to each parameter range, and collect the weight value of each disease feature corresponding to each vocal cord disease type;

[0062] Based on the weight value of each of the disease characteristics and the initial disease severity information corresponding to each of the disease characteristics, the disease severity information of the user in each vocal cord disease type is calculated.

[0063] In a third aspect, the present application provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0064] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0065] In a fifth aspect, the present application provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0066] The above-mentioned AI and audio-based vocal cord disease status recognition method and device collect the user's voice audio data and the user's historical medical information, and recognize the user's tone data and the user's sound quality data based on the voice audio data; analyze the user's historical vocal cord disease information based on the user's historical medical information, and recognize the user's vocal cord abnormality data through an artificial neural network based on the tone data and the sound quality data; recognize the user's historical vocal cord disease information and the user's vocal cord abnormality data through a vocal cord disease map based on the user's historical vocal cord disease information and the user's vocal cord abnormality data, and recognize the user's probability value of each vocal cord disease type based on the abnormal data value of each abnormal type in the user's vocal cord abnormality data. The probability value of each vocal cord disease type of the user and the disease degree information of the user in each vocal cord disease type are used as the vocal cord disease state reference information of the user, and the vocal cord disease state reference information is sent to the medical staff client. This solution uses an artificial neural network to identify the abnormal vocal cord data corresponding to the user's voice audio data, avoiding the problem that the accuracy of manual identification is affected by the experience of medical staff. Then, based on the user's abnormal vocal cord data, combined with the user's historical vocal cord disease information, the user's abnormal vocal cord data and the historical vocal cord disease information are identified through the vocal cord disease map. The associated information and implicit associated information of the current vocal cord disease type are identified, thereby determining the possible vocal cord disease type of the user and the probability value of each vocal cord disease type. Finally, based on the abnormal data value of each abnormal type in the abnormal vocal cord data, the disease degree information of the user in each vocal cord disease type is analyzed, thereby avoiding the problem that the user's vocal cord disease state cannot be directly judged based on the user's voice, or the accuracy of the judged vocal cord disease state is low. Compared with the existing technology, this solution uses the current voice audio data and historical vocal cord disease information as the basis for data analysis from multiple angles. It comprehensively analyzes the user's vocal cord abnormality data, the probability value of the type of vocal cord disease that the user may develop, and the disease severity information of each type of vocal cord disease through the artificial neural network and the knowledge graph corresponding to the vocal cord disease graph, thereby comprehensively improving the comprehensiveness and accuracy of the identified user's vocal cord disease types and vocal cord disease severity. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 is a flowchart of a method for identifying vocal cord disease states based on AI and audio in one embodiment;

[0068] Figure 2 is a structural block diagram of a device for identifying vocal cord disease states based on AI and audio in one embodiment;

[0069] Figure 3 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application 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 application and are not used to limit the present application.

[0071] The method for identifying vocal cord disease states based on AI and audio provided in the embodiment of the present application can be applied to the application environment of auxiliary intelligent analysis of vocal cord disease identification. The method can be applied to a terminal, a server, or a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, etc. Among them, the terminal identifies the vocal cord abnormality data corresponding to the user's voice audio data through an artificial neural network, avoiding the problem that the accuracy of manual identification is affected by the experience of medical staff, and then based on the user's vocal cord abnormality data, combined with the user's historical vocal cord disease information, through the vocal cord disease atlas, the user's vocal cord abnormality data, and the historical vocal cord disease information and the current vocal cord disease type The associated information and implicit associated information are identified, thereby determining the user's vocal cord disease type. Finally, based on the abnormal data values ​​of each abnormal type in the vocal cord abnormality data, the disease degree information of the user's vocal cord disease type is analyzed, thereby avoiding the problem that the user's vocal cord disease state cannot be directly judged based on the user's voice, or the accuracy of the judged vocal cord disease state is low. Compared with the existing technology, this solution takes the current voice audio data and historical vocal cord disease information as the basis for data analysis from multiple angles, and comprehensively analyzes the user's vocal cord abnormality data, vocal cord disease type, and disease severity information through artificial neural networks and the knowledge graph corresponding to the vocal cord disease graph, thereby comprehensively improving the accuracy of identifying the type of vocal cord disease and the severity of vocal cord disease.

[0072] In one embodiment, Figure 1 As shown, a method for identifying vocal cord disease status based on AI and audio is provided, and the method is applied to a terminal as an example for explanation, including the following steps:

[0073] Step S101 , collecting the user's voice audio data and the user's historical medical information, and identifying the user's tone data and the user's sound quality data based on the voice audio data.

[0074] In this embodiment, the terminal collects the audio data obtained by the user reading aloud based on the text information provided by the medical staff through the audio collection device, and then the terminal queries the user's historical medical information in the patient database of the medical staff, wherein the historical medical information includes the user's historical disease type and the severity of the disease of each historical disease type. Then, the terminal identifies the user's tone data and sound quality data through the audio recognition strategy, wherein the user's tone data is the tone range corresponding to the user's audio amplitude, and the user's sound quality data includes the audio abnormal fluctuation data in the user's voice audio data, and the audio abnormal frequency band sequence, which is used to characterize the user's sound quality state. Among them, when the user has a vocal cord disease, the user's sound quality state is poor, the audio abnormal fluctuation data is more, and the distribution of the audio abnormal frequency band sequence is irregular. When the user does not have a vocal cord disease, the user's sound quality state is normal, the audio abnormal fluctuation data is less, and the distribution of the audio abnormal frequency band sequence is irregular. The abnormal audio frequency band sequence includes a plurality of abnormal audio frequency band data, which are audio frequency band data whose similarity between the audio features corresponding to the waveform of the normal audio frequency band data is lower than the similarity threshold. For example, based on the audio features corresponding to the audio waveform of the user's pronunciation of the text field "today", and the audio features corresponding to the audio waveform of the user's pronunciation of the text field "today" under normal vocal cord conditions, the terminal calculates the audio similarity between the two audio features through a similarity algorithm, and when the audio similarity is lower than the preset similarity threshold, the audio frequency band data of the audio waveform is used as abnormal audio frequency band data.

[0075] Step S102, based on the user's historical medical information, analyze the user's historical vocal cord disease information, and based on the pitch data and sound quality data, identify the user's vocal cord abnormality data through an artificial neural network.

[0076] In this embodiment, the terminal analyzes the user's historical vocal cord disease information based on the user's historical medical information, and identifies the user's vocal cord abnormality data through an artificial neural network based on the tone data and the sound quality data. The artificial neural network is used to analyze the tone data and the sound quality data to identify the vocal cord abnormality information corresponding to the tone data and the sound quality data. The tone range to which each tone data belongs corresponds to a different vocal cord abnormality parameter, and the vocal cord abnormality parameter is used to adjust the output weight of the artificial neural network, thereby ensuring that the vocal cord abnormality data output by the artificial neural network is the vocal cord abnormality data that may exist in the tone range. The specific identification process will be described in detail later.

[0077] Step S103, based on the user's historical vocal cord disease information and the user's vocal cord abnormality data, the probability value of each vocal cord disease type of the user is identified through the vocal cord disease atlas, and based on the abnormal data value of each abnormal type in the user's vocal cord abnormality data, the disease severity information of the user in each vocal cord disease type is identified.

[0078] In this embodiment, the terminal identifies the probability value of each vocal cord disease type of the user through the vocal cord disease map based on the user's historical vocal cord disease information and the user's vocal cord abnormality data, and identifies the disease degree information of the user in each vocal cord disease type based on the abnormal data value of each abnormal type in the user's vocal cord abnormality data. Wherein, the vocal cord disease map is a knowledge map of all vocal cord diseases constructed by medical staff based on each historical vocal cord abnormality data in the treatment database, the vocal cord disease type corresponding to each historical vocal cord abnormality data, the disease degree information of the vocal cord disease type corresponding to each historical vocal cord abnormality data, and the probability value range of the vocal cord disease type corresponding to each historical vocal cord abnormality data, wherein the upper layer nodes in the knowledge map are each vocal cord abnormality data node, the middle layer of the knowledge map corresponds to multiple sub-vocal cord disease maps, each sub-vocal cord disease map belongs to a vocal cord disease type node, and the lower layer of the knowledge map corresponds to multiple disease feature nodes. Each vocal cord abnormality data node can be associated with one or more vocal cord disease types, each vocal cord disease type node is associated with each other through the similarity of the vocal cord disease type, and each disease feature node under each vocal cord disease type node can have associated information with multiple vocal cord disease type nodes. The specific identification process will be described in detail later.

[0079] Step S104, using the probability value of each vocal cord disease type of the user and the disease severity information of each vocal cord disease type of the user as the vocal cord disease status reference information of the user, and sending the vocal cord disease status reference information to the medical staff client.

[0080] In this embodiment, the terminal uses the probability value of each vocal cord disease type of the user and the disease severity information of each vocal cord disease type as the vocal cord disease status reference information of the user, and sends the vocal cord disease status reference information to the medical staff client. It can be understood that this solution is an intelligent auxiliary technology that assists medical staff in diagnosing the vocal cord disease of the patient. Through the vocal cord disease status reference information of the intelligent auxiliary technology, the doctor is assisted in analyzing the target vocal cord disease type of the user and the disease severity of the target vocal cord disease type of the user, and it cannot directly diagnose and confirm the vocal cord disease for the medical staff.

[0081] Based on the above scheme, the vocal cord abnormality data corresponding to the user's voice audio data is identified through an artificial neural network, which avoids the problem that the accuracy of manual identification is affected by the experience of medical staff. Then, based on the user's vocal cord abnormality data, combined with the user's historical vocal cord disease information, the user's vocal cord abnormality data, as well as the historical vocal cord disease information and the current vocal cord disease type's associated information and implicit associated information are identified through the vocal cord disease map, thereby determining the user's vocal cord disease type. Finally, based on the abnormal data values ​​of each abnormal type in the vocal cord abnormality data, the disease degree information of the user's vocal cord disease type is analyzed, thereby avoiding the problem that the user's vocal cord disease state cannot be directly judged based on the user's voice, or the accuracy of the judged vocal cord disease state is low. Compared with the prior art, this scheme uses the current voice audio data and historical vocal cord disease information as the data analysis basis, and comprehensively analyzes the vocal cord abnormality data, vocal cord disease type, and disease degree information of the user's vocal cord state through the knowledge map corresponding to the artificial neural network and the vocal cord disease map, thereby comprehensively improving the accuracy of the identified vocal cord disease type and the severity of the vocal cord disease.

[0082] Optionally, based on the vocal audio data, the user's tone data and the user's sound quality data are identified, including: performing Fourier transform processing on the vocal audio data to obtain an audio spectrum of the vocal audio data, and converting the audio spectrum of the vocal audio data into a waveform graph; collecting the user's pronunciation text, and a regular waveform graph corresponding to the normal pronunciation of the pronunciation text, and identifying the sub-waveform graph corresponding to the audio segment of each text content in the user's pronunciation text in the waveform graph converted from the audio spectrum of each vocal audio data; based on the sub-waveform graph corresponding to each audio segment, the user's audio amplitude range and the audio frequency distribution information of each sub-waveform graph are identified, and based on the user's audio amplitude range, the user's tone data is identified; based on each audio frequency distribution information, the user's audio features are extracted through an audio feature extraction network, and based on each audio feature, the user's sound quality data is identified through a feature analysis strategy.

[0083] In this embodiment, the terminal performs Fourier transform processing on the audible audio data to obtain an audio spectrum of the audible audio data, and converts the audio spectrum of the audible audio data into a waveform graph. The audible audio data may be an audio signal, and the terminal converts the audio signal into an audio spectrum through a Fourier transform strategy, and converts the audio spectrum into a waveform graph, thereby improving the precision of the audio analysis and avoiding the low-precision problem of the naked ear analyzing the audio.

[0084] Then, the terminal collects the user's pronunciation text and the conventional waveform corresponding to the normal pronunciation of the pronunciation text, and identifies the sub-waveform corresponding to the audio segment of each text content in the pronunciation text in each waveform. The text content corresponding to each audio segment can be a word content or a sentence content, such as "today", "weather", "beautiful", etc. The sentence content can be "123456789", "I am very happy today", etc. The division method of each text content is the division method preset by the staff, that is, the staff divides the pronunciation text in advance, and obtains each text content of the pronunciation text. Then, the terminal obtains the sub-waveform corresponding to the audio segment of each text content by identifying the audio segment corresponding to each text content.

[0085] Then, the terminal identifies the user's audio amplitude range and the audio frequency distribution information of each sub-waveform based on the sub-waveform corresponding to each audio segment, and identifies the user's tone data based on the user's audio amplitude range. The tone data includes the tone data corresponding to the conventional waveform and the tone data corresponding to the waveform of the vocal audio data. Then, based on the audio frequency distribution information, the terminal extracts the user's audio features through an audio feature extraction network, and identifies the user's sound quality data through a feature analysis strategy based on each audio feature. The specific feature analysis process will be described in detail later, and the audio features include the audio features corresponding to the conventional waveform and the audio features corresponding to the waveform of the vocal audio data. The audio feature extraction network is a neural network based on a combination of a convolutional neural network (CNN) and a short-time Fourier transform (STFT).

[0086] Based on the above solution, by converting the audio signal into a waveform graph and then identifying the user's tone data and sound quality data, the accuracy of identifying the user's tone data and sound quality data is improved.

[0087] Optionally, based on each audio feature, a feature analysis strategy is used to identify the user's sound quality data, including: based on each audio feature, identifying the user's audio abnormal frequency band data and the user's audio abnormal fluctuation data; arranging each audio abnormal frequency band data according to the position information of each audio abnormal frequency band data in the voiced audio data to obtain the user's audio abnormal frequency band sequence, and using each audio abnormal fluctuation data and the audio abnormal frequency band sequence as the user's sound quality data.

[0088] In this embodiment, the terminal identifies the user's audio abnormal frequency band data and the user's audio abnormal fluctuation data based on various audio features. Specifically, the terminal calculates the similarity between the audio features of the waveforms of the same text content, and then screens the audio frequency band data to which the audio features corresponding to the similarity below the preset similarity threshold belong as the audio abnormal frequency band data. Among them, the algorithm for calculating the similarity can be but is not limited to the Euclidean distance algorithm, the Mahalanobis distance algorithm, and the cosine similarity algorithm. Specifically, the terminal identifies the audio abnormal fluctuation data in a manner that the terminal identifies the amplitude range of each waveform and the amplitude change slope of each waveform for two waveforms of the same text content, and calculates the range deviation between the amplitude ranges of the two waveforms and the slope deviation between the amplitude change slopes of the two waveforms. When the range deviation is greater than the range deviation threshold, or the slope deviation is greater than the slope deviation threshold, the terminal uses the waveform corresponding to the uttered audio data as the audio abnormal fluctuation data.

[0089] Then, the terminal arranges the audio abnormal frequency band data according to the position information of each audio abnormal frequency band data in the voiced audio data to obtain the user's audio abnormal frequency band sequence, and uses the audio abnormal fluctuation data and the audio abnormal frequency band sequence as the user's sound quality data.

[0090] Based on the above scheme, the similarity and deviation between two waveforms of the same text content are identified, so as to filter out the abnormal frequency band data of each audio and the abnormal fluctuation data of each audio, thereby improving the accuracy and efficiency of the screening.

[0091] Optionally, based on the user's historical medical information, the user's historical vocal cord disease information is analyzed, including: screening each vocal cord medical information in the historical medical information, and identifying the historical vocal cord disease type corresponding to each vocal cord medical information, and the historical disease severity information corresponding to each historical vocal cord disease type; using each historical vocal cord disease type, and the historical disease severity information corresponding to each historical vocal cord disease type, as the user's historical vocal cord disease information.

[0092] In this embodiment, the terminal screens each vocal cord medical consultation information in the historical medical consultation information, and identifies the historical vocal cord disease type corresponding to each vocal cord medical consultation information, and the historical disease severity information corresponding to each historical vocal cord disease type. Finally, the terminal uses each historical vocal cord disease type and the historical disease severity information corresponding to each historical vocal cord disease type as the user's historical vocal cord disease information.

[0093] Based on the above solution, by querying the user's historical vocal cord disease information in the historical medical information, the reference analysis accuracy of the current user's vocal cord disease identification is improved.

[0094] Optionally, based on the pitch data and the sound quality data, an artificial neural network is used to identify the user's vocal cord abnormality data, including: based on the pitch data, identifying the user's pitch range, and based on the user's pitch range, querying the vocal cord abnormality parameters corresponding to the pitch range in the vocal cord abnormality database; based on the vocal cord abnormality parameters, adjusting the initial artificial neural network to obtain an artificial neural network, and inputting each audio abnormality fluctuation data and the audio abnormality frequency band sequence into the artificial neural network, analyzing the first vocal cord abnormality information corresponding to each audio abnormality fluctuation data and the second vocal cord abnormality information corresponding to the audio abnormality frequency band sequence; collecting the first weight corresponding to each audio abnormality fluctuation data and the second weight corresponding to the audio abnormality frequency band sequence, and based on the first weight and the second weight, weighted processing is performed on each first vocal cord abnormality information and each second vocal cord abnormality information to obtain the probability value of each vocal cord abnormality information; based on the probability value of the same vocal cord abnormality information and the number of the same vocal cord abnormality information, the target probability value of each vocal cord abnormality information is calculated, and the vocal cord abnormality information corresponding to the maximum target probability value is selected as the vocal cord abnormality data corresponding to the pitch range.

[0095] In this embodiment, the terminal identifies the user's pitch range based on the pitch data, and based on the user's pitch range, queries the vocal cord abnormality parameter corresponding to the pitch range in the vocal cord abnormality database. The vocal cord abnormality parameter is a constant parameter of the initial artificial neural network when identifying the vocal cord abnormality information corresponding to each sound quality data.

[0096] Then, the terminal adjusts the initial artificial neural network based on the vocal cord abnormality parameters to obtain an artificial neural network, and inputs each audio abnormality fluctuation data and the audio abnormality frequency band sequence into the artificial neural network to analyze the first vocal cord abnormality information corresponding to each audio abnormality fluctuation data and the second vocal cord abnormality information corresponding to the audio abnormal frequency band sequence. Each audio abnormality fluctuation data corresponds to one vocal cord abnormality information, and the audio abnormal frequency band sequence can correspond to one or more vocal cord abnormality information.

[0097] The terminal collects the first weight corresponding to each audio abnormal fluctuation data and the second weight corresponding to the audio abnormal frequency band sequence. Among them, each weight is a weight sent to the terminal by the staff. Then, based on the first weight and the second weight, the terminal performs weighted processing on each first vocal cord abnormal information and each second vocal cord abnormal information to obtain the probability value of each vocal cord abnormal information. Among them, the first vocal cord abnormal information may be the same as the second vocal cord abnormal information, the first vocal cord abnormal information may be the same as each other, and the second vocal cord abnormal information may be the same as each other.

[0098] Then, the terminal calculates the target probability value of each vocal cord abnormality information based on the probability value of the same vocal cord abnormality information and the number of the same vocal cord abnormality information, and selects the vocal cord abnormality information corresponding to the maximum target probability value as the vocal cord abnormality data corresponding to the pitch range.

[0099] Based on the above scheme, after calculating the probability and then performing weighted calculation on the target probability value of each vocal cord abnormality information, the vocal cord abnormality data is screened, thereby improving the accuracy of screening the vocal cord abnormality data.

[0100] Optionally, based on the user's historical vocal cord disease information and the user's vocal cord abnormality data, the probability values ​​of each vocal cord disease type of the user are identified through the vocal cord disease atlas, including: extracting each data feature of the vocal cord abnormality data, and based on each historical vocal cord disease type, in the vocal cord disease atlas, adjusting the weight coefficient of the sub-branch atlas corresponding to each vocal cord disease type to obtain the target weight coefficient of each sub-branch atlas; based on the target weight coefficient of each sub-branch atlas, calculating the similarity between each data feature and the disease feature of the branch node of each sub-branch atlas, and screening the branch nodes to which the disease features greater than the similarity threshold belong as the target branch nodes; based on the vocal cord disease type to which the target branch nodes corresponding to each data feature belong, identifying each vocal cord disease type of the user, and calculating the probability value of each vocal cord disease type based on the similarity corresponding to each data feature.

[0101] In this embodiment, the terminal extracts each data feature of the abnormal vocal cord data, and based on each historical vocal cord disease type, in the vocal cord disease atlas, adjusts the weight coefficient of the sub-branch atlas corresponding to each vocal cord disease type, and obtains the target weight coefficient of each sub-branch atlas. That is, when the terminal calculates the similarity between each data feature and each disease feature, when the vocal cord disease type to which the disease feature belongs is a historical vocal cord disease type, the terminal needs to multiply it by the target weight coefficient corresponding to the historical vocal cord disease type to obtain the similarity between each data feature and the disease feature. Specifically, the terminal identifies the number of onsets of each historical vocal cord disease type and the interval between onsets of each historical vocal cord disease type. Then, the terminal calculates the recurrence probability of each historical vocal cord disease type through the vocal cord weight algorithm, and uses each recurrence probability as the weight coefficient of each historical vocal cord disease type.

[0102] Specifically, the calculation formula of the vocal cord weight algorithm is:

[0103]

[0104] In the above formula, μ i is the recurrence probability, δ1 is the weight value of the number of occurrences of the i-th historical vocal cord disease type, M iis the number of occurrences of the i-th historical vocal cord disease type, δ2 is the weight value of the interval between the occurrences of the i-th historical vocal cord disease type, K ij is the jth onset interval of the i-th historical vocal cord disease type, is the historical average recurrence probability of the i-th historical vocal cord disease type, i is the virtual number of the historical vocal cord disease type, and j is the virtual number of the onset interval.

[0105] Based on the target weight coefficient of each sub-branch map, the terminal calculates the similarity between each data feature and the disease feature of the branch node of each sub-branch map, and selects the branch nodes to which the disease features greater than the similarity threshold belong as the target branch nodes. Among them, each sub-branch map is a subordinate map corresponding to each vocal cord disease type, and each target score node is a lower-level node of the vocal cord disease map. Among them, the method of calculating the similarity is to calculate it through a similarity algorithm, which is the same as the similarity algorithm described above, and no redundant description is given here.

[0106] The terminal identifies the vocal cord disease types of the user based on the vocal cord disease types to which the target branch nodes corresponding to the data features belong, and calculates the probability value of each vocal cord disease type based on the similarity corresponding to each data feature. The probability value algorithm is to average the similarities corresponding to all data features to which each vocal cord disease type belongs, and the probability value of each vocal cord disease type is obtained.

[0107] Based on the above scheme, the vocal cord disease map is adjusted by using the historical vocal cord disease types and the disease severity information of the historical vocal cord disease types, so as to screen the various vocal cord disease types that the user may have and the probability values ​​of each vocal cord disease type, thereby improving the accuracy of identifying the probability values ​​of each vocal cord disease type.

[0108] Optionally, based on the abnormal data values ​​of each abnormal type in the user's vocal cord abnormal data, the disease severity information of the user in each vocal cord disease type is identified, including: identifying the characteristic value corresponding to each data feature, and based on the disease feature corresponding to the data feature, identifying the disease influence parameter corresponding to the characteristic value; querying in a database the parameter range to which the disease influence parameter corresponding to each characteristic value belongs, and the initial disease severity information corresponding to each parameter range, and collecting the weight value of each disease feature corresponding to each vocal cord disease type; based on the weight value of each disease feature, and the initial disease severity information corresponding to each disease feature, calculating the disease severity information of the user in each vocal cord disease type.

[0109] In this embodiment, the terminal identifies the characteristic value corresponding to each data feature, and identifies the disease impact parameter corresponding to the characteristic value based on the disease feature corresponding to the data feature. Each disease feature corresponds to a disease impact parameter range, and each data feature corresponds to a characteristic value. The terminal queries the disease impact parameter corresponding to each characteristic value by querying the correspondence between each characteristic value and the disease impact parameter preset in the terminal.

[0110] In the database, the terminal queries the parameter range to which the disease influencing parameter corresponding to each characteristic value belongs, as well as the initial disease severity information corresponding to each parameter range, and collects the weight value of each disease feature corresponding to each vocal cord disease type. Then, based on the weight value of each disease feature and the initial disease severity information corresponding to each disease feature, the terminal calculates the disease severity information of the user in each vocal cord disease type. Among them, there is a corresponding relationship between each parameter range and each disease severity information. The terminal obtains the disease severity information of the vocal cord disease type by multiplying the weight value by the initial disease severity information.

[0111] Based on the above scheme, the disease influencing parameters of the disease characteristics of each vocal cord disease type are identified, so as to calculate the disease severity information of each vocal cord disease type, thereby improving the accuracy and comprehensiveness of calculating the disease severity information of each vocal cord disease type.

[0112] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0113] Based on the same inventive concept, the embodiment of the present application also provides an AI and audio-based vocal cord disease state recognition device for implementing the above-mentioned AI and audio-based vocal cord disease state recognition method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in one or more AI and audio-based vocal cord disease state recognition device embodiments provided below can refer to the limitations of the AI ​​and audio-based vocal cord disease state recognition method above, and will not be repeated here.

[0114] In one embodiment, Figure 2As shown, a device for identifying vocal cord disease states based on AI and audio is provided, comprising: a collection module 210, an analysis module 220, an identification module 230 and a determination module 240, wherein:

[0115] The collection module 210 is used to collect the user's voice audio data and the user's historical medical information, and identify the user's tone data and the user's sound quality data based on the voice audio data;

[0116] The analysis module 220 is used to analyze the historical vocal cord disease information of the user based on the historical medical information of the user, and identify the abnormal vocal cord data of the user through an artificial neural network based on the tone data and the sound quality data;

[0117] The identification module 230 is used to identify the probability value of each vocal cord disease type of the user based on the historical vocal cord disease information of the user and the vocal cord abnormality data of the user through the vocal cord disease atlas, and identify the disease degree information of the user in each vocal cord disease type based on the abnormal data value of each abnormal type in the vocal cord abnormality data of the user;

[0118] The determination module 240 is used to use the probability value of each vocal cord disease type of the user and the disease severity information of the user in each vocal cord disease type as the vocal cord disease status reference information of the user, and send the vocal cord disease status reference information to the medical staff client.

[0119] Optionally, the acquisition module 210 is specifically used for:

[0120] Performing Fourier transform processing on the utterance audio data to obtain an audio spectrum of the utterance audio data, and converting the audio spectrum of the utterance audio data into a waveform graph;

[0121] Collect the user's pronunciation text and the regular waveform corresponding to the normal pronunciation of the pronunciation text, and identify the sub-waveform corresponding to the audio segment of each text content in the pronunciation text of the user in the waveform converted from the audio spectrum of each of the utterance audio data;

[0122] Based on the sub-waveforms corresponding to the audio segments, identifying the audio amplitude range of the user and the audio frequency distribution information of each sub-waveform, and based on the audio amplitude range of the user, identifying the tone data of the user;

[0123] Based on the audio frequency distribution information, the audio features of the user are extracted through an audio feature extraction network, and based on the audio features, the sound quality data of the user is identified through a feature analysis strategy.

[0124] Optionally, the acquisition module 210 is specifically used for:

[0125] Based on each of the audio features, identifying the abnormal audio frequency band data of the user and the abnormal audio fluctuation data of the user;

[0126] Arrange each of the audio abnormal frequency band data according to the position information of each of the audio abnormal frequency band data in the voiced audio data to obtain the audio abnormal frequency band sequence of the user, and use each of the audio abnormal fluctuation data and the audio abnormal frequency band sequence as the sound quality data of the user.

[0127] Optionally, the analysis module 220 is specifically used for:

[0128] In the historical medical information, each vocal cord medical information is screened, and the historical vocal cord disease type corresponding to each vocal cord medical information and the historical disease severity information corresponding to each historical vocal cord disease type are identified;

[0129] Each of the historical vocal cord disease types and the historical disease severity information corresponding to each of the historical vocal cord disease types are used as the historical vocal cord disease information of the user.

[0130] Optionally, the analysis module 220 is specifically used for:

[0131] Based on the pitch data, identifying the pitch range of the user, and based on the pitch range of the user, querying a vocal cord abnormality database for a vocal cord abnormality parameter corresponding to the pitch range;

[0132] Based on the vocal cord abnormality parameters, an initial artificial neural network is adjusted to obtain an artificial neural network, and each of the audio abnormal fluctuation data and the audio abnormal frequency band sequence is input into the artificial neural network to analyze the first vocal cord abnormality information corresponding to each of the audio abnormal fluctuation data and the second vocal cord abnormality information corresponding to the audio abnormal frequency band sequence;

[0133] Collecting the first weight corresponding to each of the audio abnormal fluctuation data and the second weight corresponding to the audio abnormal frequency band sequence, and performing weighted processing on each of the first vocal cord abnormal information and each of the second vocal cord abnormal information based on the first weight and the second weight to obtain a probability value of each of the vocal cord abnormal information;

[0134] Based on the probability value of the same vocal cord abnormality information and the number of the same vocal cord abnormality information, the target probability value of each vocal cord abnormality information is calculated, and the vocal cord abnormality information corresponding to the maximum target probability value is selected as the vocal cord abnormality data corresponding to the pitch range.

[0135] Optionally, the identification module 230 is specifically configured to:

[0136] Extracting each data feature of the vocal cord abnormality data, and based on each of the historical vocal cord disease types, adjusting the weight coefficient of the sub-branch atlas corresponding to each of the vocal cord disease types in the vocal cord disease atlas to obtain a target weight coefficient of each of the sub-branch atlases;

[0137] Based on the target weight coefficient of each sub-branch map, the similarity between each data feature and the disease feature of the branch node of each sub-branch map is calculated, and the branch node to which the disease feature having a similarity greater than a similarity threshold belongs is selected as the target branch node;

[0138] Based on the vocal cord disease type to which the target branch node corresponding to each of the data features belongs, the vocal cord disease types of the user are identified, and based on the similarity corresponding to each of the data features, the probability value of each of the vocal cord disease types is calculated.

[0139] Optionally, the identification module 230 is specifically configured to:

[0140] Identify a characteristic value corresponding to each data feature, and based on the disease feature corresponding to the data feature, identify a disease impact parameter corresponding to the characteristic value;

[0141] In the database, query the parameter range to which the disease influencing parameter corresponding to each characteristic value belongs, and the initial disease degree information corresponding to each parameter range, and collect the weight value of each disease feature corresponding to each vocal cord disease type;

[0142] Based on the weight value of each of the disease characteristics and the initial disease severity information corresponding to each of the disease characteristics, the disease severity information of the user in each vocal cord disease type is calculated.

[0143] Each module in the above-mentioned AI and audio-based vocal cord disease state recognition device can be implemented in whole or in part by software, hardware, and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0144] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for identifying a vocal cord disease state based on AI and audio is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0145] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0146] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0147] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0148] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps of any one of the methods in the first aspect.

[0149] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0150] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0151] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0152] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for identifying vocal cord disease states based on AI and audio, characterized in that: The method comprises: Collecting the user's voice audio data and the user's historical medical information, and identifying the user's tone data and the user's sound quality data based on the voice audio data; Based on the user's historical medical information, analyzing the user's historical vocal cord disease information, and based on the tone data and the sound quality data, identifying the user's vocal cord abnormality data through an artificial neural network; Based on the historical vocal cord disease information of the user and the vocal cord abnormality data of the user, the probability value of each vocal cord disease type of the user is identified through a vocal cord disease atlas, and based on the abnormal data value of each abnormal type in the vocal cord abnormality data of the user, the disease degree information of the user in each vocal cord disease type is identified; Using the probability value of each vocal cord disease type of the user and the disease degree information of the user in each vocal cord disease type as the vocal cord disease state reference information of the user, and sending the vocal cord disease state reference information to the medical staff client; The identifying, based on the utterance audio data, the tone data of the user and the sound quality data of the user comprises: Performing Fourier transform processing on the utterance audio data to obtain an audio spectrum of the utterance audio data, and converting the audio spectrum of the utterance audio data into a waveform graph; Collect the user's pronunciation text and the regular waveform corresponding to the normal pronunciation of the pronunciation text, and identify the sub-waveform corresponding to the audio segment of each text content in the pronunciation text of the user in the waveform converted from the audio spectrum of each of the utterance audio data; Based on the sub-waveforms corresponding to the audio segments, identifying the audio amplitude range of the user and the audio frequency distribution information of each sub-waveform, and based on the audio amplitude range of the user, identifying the tone data of the user; Based on the audio frequency distribution information, extract the audio features of the user through an audio feature extraction network, and based on the audio features, identify the sound quality data of the user through a feature analysis strategy; The identifying the sound quality data of the user based on each of the audio features through a feature analysis strategy includes: Based on each of the audio features, identifying the abnormal audio frequency band data of the user and the abnormal audio fluctuation data of the user; Arrange each of the audio abnormal frequency band data according to the position information of each of the audio abnormal frequency band data in the voiced audio data to obtain the audio abnormal frequency band sequence of the user, and use each of the audio abnormal fluctuation data and the audio abnormal frequency band sequence as the sound quality data of the user.

2. The method according to claim 1, characterized in that The analyzing the user's historical vocal cord disease information based on the user's historical medical information includes: In the historical medical information, each vocal cord medical information is screened, and the historical vocal cord disease type corresponding to each vocal cord medical information and the historical disease severity information corresponding to each historical vocal cord disease type are identified; Each of the historical vocal cord disease types and the historical disease severity information corresponding to each of the historical vocal cord disease types are used as the historical vocal cord disease information of the user.

3. The method according to claim 1, characterized in that The identifying the abnormal vocal cord data of the user through an artificial neural network based on the tone data and the sound quality data includes: Based on the pitch data, identifying the pitch range of the user, and based on the pitch range of the user, querying a vocal cord abnormality database for a vocal cord abnormality parameter corresponding to the pitch range; Based on the vocal cord abnormality parameters, an initial artificial neural network is adjusted to obtain an artificial neural network, and each of the audio abnormal fluctuation data and the audio abnormal frequency band sequence is input into the artificial neural network to analyze the first vocal cord abnormality information corresponding to each of the audio abnormal fluctuation data and the second vocal cord abnormality information corresponding to the audio abnormal frequency band sequence; Collecting the first weight corresponding to each of the audio abnormal fluctuation data and the second weight corresponding to the audio abnormal frequency band sequence, and performing weighted processing on each of the first vocal cord abnormal information and each of the second vocal cord abnormal information based on the first weight and the second weight to obtain a probability value of each of the vocal cord abnormal information; Based on the probability value of the same vocal cord abnormality information and the number of the same vocal cord abnormality information, the target probability value of each vocal cord abnormality information is calculated, and the vocal cord abnormality information corresponding to the maximum target probability value is selected as the vocal cord abnormality data corresponding to the pitch range.

4. The method according to claim 2, characterized in that: The method of identifying the probability value of each vocal cord disease type of the user based on the historical vocal cord disease information of the user and the vocal cord abnormality data of the user through a vocal cord disease atlas includes: Extracting each data feature of the vocal cord abnormality data, and based on each of the historical vocal cord disease types, adjusting the weight coefficient of the sub-branch atlas corresponding to each of the vocal cord disease types in the vocal cord disease atlas to obtain a target weight coefficient of each of the sub-branch atlases; Based on the target weight coefficient of each sub-branch map, the similarity between each data feature and the disease feature of the branch node of each sub-branch map is calculated, and the branch node to which the disease feature having a similarity greater than a similarity threshold belongs is selected as the target branch node; Based on the vocal cord disease type to which the target branch node corresponding to each of the data features belongs, the vocal cord disease types of the user are identified, and based on the similarity corresponding to each of the data features, the probability value of each of the vocal cord disease types is calculated.

5. The method according to claim 4, characterized in that The identifying the disease degree information of the user in each vocal cord disease type based on the abnormal data value of each abnormal type in the vocal cord abnormal data of the user includes: Identify a characteristic value corresponding to each data feature, and based on the disease feature corresponding to the data feature, identify a disease impact parameter corresponding to the characteristic value; In the database, query the parameter range to which the disease influencing parameter corresponding to each characteristic value belongs, and the initial disease degree information corresponding to each parameter range, and collect the weight value of each disease feature corresponding to each vocal cord disease type; Based on the weight value of each of the disease characteristics and the initial disease severity information corresponding to each of the disease characteristics, the disease severity information of the user in each vocal cord disease type is calculated.

6. A device for identifying vocal cord disease states based on AI and audio, characterized in that: The device comprises: A collection module, used to collect the user's voice audio data and the user's historical medical information, and identify the user's tone data and the user's sound quality data based on the voice audio data; An analysis module, configured to analyze the user's historical vocal cord disease information based on the user's historical medical information, and identify the user's vocal cord abnormality data through an artificial neural network based on the tone data and the sound quality data; an identification module, for identifying the probability value of each vocal cord disease type of the user based on the historical vocal cord disease information of the user and the vocal cord abnormality data of the user through a vocal cord disease atlas, and identifying the disease degree information of the user in each vocal cord disease type based on the abnormal data value of each abnormal type in the vocal cord abnormality data of the user; A determination module, used for using the probability value of each vocal cord disease type of the user and the disease degree information of the user in each vocal cord disease type as the vocal cord disease state reference information of the user, and sending the vocal cord disease state reference information to a medical staff client; The identifying, based on the utterance audio data, the tone data of the user and the sound quality data of the user comprises: Performing Fourier transform processing on the utterance audio data to obtain an audio spectrum of the utterance audio data, and converting the audio spectrum of the utterance audio data into a waveform graph; Collect the user's pronunciation text and the regular waveform corresponding to the normal pronunciation of the pronunciation text, and identify the sub-waveform corresponding to the audio segment of each text content in the pronunciation text of the user in the waveform converted from the audio spectrum of each of the utterance audio data; Based on the sub-waveforms corresponding to the audio segments, identifying the audio amplitude range of the user and the audio frequency distribution information of each sub-waveform, and based on the audio amplitude range of the user, identifying the tone data of the user; Based on the audio frequency distribution information, extract the audio features of the user through an audio feature extraction network, and based on the audio features, identify the sound quality data of the user through a feature analysis strategy; The specific implementation of identifying the user's sound quality data based on each of the audio features through a feature analysis strategy is: Based on each of the audio features, identifying the abnormal audio frequency band data of the user and the abnormal audio fluctuation data of the user; Arrange each of the audio abnormal frequency band data according to the position information of each of the audio abnormal frequency band data in the voiced audio data to obtain the audio abnormal frequency band sequence of the user, and use each of the audio abnormal fluctuation data and the audio abnormal frequency band sequence as the sound quality data of the user.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images