A method and apparatus for health analysis through timbre tonality

Abstract

The present application provides a kind of by timbre tone health analysis method and equipment, comprising: the voice signal of user is collected, and the voice signal is analog-digital converted to obtain electric signal, and electric signal is stored to specified area, and electric signal is uniformly transmitted;The electric signal transmitted is signal processing, and the electric signal after processing is timbre tone analysis, and the timbre tone characteristics of user are obtained;According to the timbre tone characteristics of user, the physical health status of user is evaluated, and the intention reference opinion is determined in combination with the analysis intention of user, the technology based on microphone sensor monitoring voice and physical health has the advantages of non-invasive, high precision, good real-time, large data volume and easy to use, can provide more scientific monitoring and analysis means for relevant medical, rehabilitation, sports and other fields, has wide application prospect and market potential.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of speech recognition, in particular to a health analysis method and device through timbre and tone. BACKGROUND

[0002] The health degree of a person can be preliminarily judged from different performances, and there is a close relationship between emotion and psychology and health. However, the performance of emotion and psychology can be preliminarily judged from the timbre and tone of a person in communication with people. People with different timbres can also judge the personality characteristics of a person from the timbres. People with high timbres have a personality of being warm and confident, and people with low timbres have a stable heart and a personality of being calm and rational. However, a low timbre may be caused by fatigue or breathing difficulty due to illness. The tone and pitch of a healthy person are more inclined to be high, and most people have a low timbre and tone when they are ill, and their mood is also more depressed. Traditional Chinese medicine often has four steps when treating a patient, that is, looking, smelling, asking and cutting. The smelling is to listen to the voice and breathing sound of the patient, and the breathing sound and voice can preliminarily judge the physical condition of the patient.

[0003] The purpose of the present project is to collect the normal communication voice information of a user through a designated device, extract the timbre and tone information of the voice information, analyze the breathing frequency of the user, and then determine the health of the user according to the overall information. SUMMARY

[0004] The present application provides a health analysis method and device through timbre and tone, which is used to determine the health status of a user through the voice of the user.

[0005] A health analysis method through timbre and tone, comprising:

[0006] S1: collecting the voice signal of the user, and performing analog-to-digital conversion on the voice signal to obtain an electric signal, storing the electric signal in a designated area, and uniformly transmitting the electric signal;

[0007] S2: performing signal processing on the transmitted electric signal, performing timbre and tone analysis on the processed electric signal, and obtaining the timbre and tone characteristics of the user;

[0008] S3: evaluating the physical health status of the user according to the timbre and tone characteristics of the user, and determining the intention reference opinion in combination with the analysis intention of the user.

[0009] Preferably, in S1, the voice signal of the user is collected, comprising:

[0010] Two or more microphones are used to collect the voice signal, the voice signal is compared with the historical voice signal of the user stored in advance, if they are consistent, the voice signal is retained, otherwise, the voice signal is rejected.

[0011] The signal strength of the speech signals collected by different microphones in the same time period is analyzed, and the speech signal collected by the microphone with the strongest signal strength in the same time period is taken as the final speech signal of the user.

[0012] Preferably, the speech signal is converted into an electrical signal through analog-to-digital conversion, and the electrical signal is stored in a designated area and uniformly transmitted, including:

[0013] The speech signal is converted into an electrical signal by using an analog-to-digital converter;

[0014] The electrical signal is analyzed for effectiveness to obtain an effective electrical signal, and the effective electrical signal is stored in a designated area;

[0015] When the effective electrical signal reaches a preset storage amount and a data transmission instruction is received, the effective electrical signal is uniformly transmitted.

[0016] Preferably, in S2, the transmitted electrical signal is processed, the processed electrical signal is analyzed for timbre and tone, and the timbre and tone characteristics of the user are obtained, including:

[0017] The transmitted electrical signal is filtered to obtain a plurality of filtered signals in different frequency bands, and the plurality of filtered signals are denoised to obtain an electrical signal to be analyzed;

[0018] The filter response of the electrical signal to be analyzed is obtained, and based on the filter response and a wide dynamic range compression curve corresponding to the frequency band, the filter response gain of the electrical signal to be analyzed is calculated;

[0019] Based on the difference between the filter response gain and the frequency band reference gain, the gain adjustment value of the electrical signal to be analyzed is determined, and the electrical signal to be analyzed is adjusted based on the gain adjustment value to obtain a target electrical signal;

[0020] The target electrical signal is input into a trained timbre and tone analysis model to obtain the timbre and tone characteristics of the user.

[0021] Preferably, the target electrical signal is input into a trained timbre and tone analysis model to obtain the timbre and tone characteristics of the user, including:

[0022] The historical speech data of different historical users is extracted to obtain the first acoustic feature of the historical speech data, and a deep neural network is trained based on the first acoustic feature to obtain an initial speech recognition model;

[0023] The feature difference between the first acoustic features of different historical users is obtained, and based on the feature difference, the personalized weight of the historical user is determined, and based on the historical user speech feature and the corresponding personalized weight, a user feature recognition model is established;

[0024] The user feature recognition model and the initial speech recognition model are model fused to obtain an initial timbre and tone analysis model;

[0025] The first acoustic features of the same historical user are maximized in similarity to obtain second acoustic features, and the first acoustic features of different historical users are maximized in similarity based on feature differences to obtain third acoustic features;

[0026] The second acoustic features are used to train and verify the initial timbre and tone analysis model until the first verification result is met, and the training is stopped to obtain a second timbre and tone analysis model;

[0027] The third acoustic features are used to train and verify the second timbre and tone analysis model until the second verification result is met, and the training is stopped to obtain a trained timbre and tone analysis model;

[0028] The user features are input into the user feature recognition model, and based on the feature recognition result, a target recognition channel of the trained timbre and tone analysis model for the target electrical signal is determined. The target electrical signal is input into the target recognition channel of the trained timbre and tone analysis model to obtain the user's timbre and tone features.

[0029] Preferably, the user feature recognition model and the initial speech recognition model are model fused to obtain an initial timbre and tone analysis model, comprising:

[0030] Based on the classification of the recognition features of the user feature recognition model, the user feature recognition model is connected to multiple recognition channels of the initial speech recognition model, wherein one type of recognition feature corresponds to one recognition channel.

[0031] Training and verifying the initial timbre and tone analysis model using the second acoustic features and the third acoustic features specifically involves training and verifying the multiple recognition channels.

[0032] Preferably, in S3, the user's physical health status is evaluated based on the user's timbre and tone features, comprising:

[0033] Based on the timbre and tone features, the user's timbre change curve and tone change curve are determined;

[0034] Based on the breathing frequency feature, the timbre change curve and the tone change curve are analyzed to determine the user's breathing frequency trend, and based on the sound emotion feature, the timbre change curve and the tone change curve are analyzed to determine the user's emotion trend;

[0035] Based on the breathing frequency trend, the user's breathing state is evaluated, and based on the evaluation result, the user's breathing balance score and breathing coordination score are determined;

[0036] Based on the emotion change trend, the mood state of the user is evaluated, and the good or bad mood score of the user is determined according to the evaluation result;

[0037] Based on the respiratory balance score, the respiratory coordination score and the good or bad mood score, the physical health state of the user is determined.

[0038] Preferably, based on the respiratory balance score, the respiratory coordination score and the good or bad mood score, the physical health state of the user is determined, including:

[0039] The respiratory balance score, the respiratory coordination score and the good or bad mood score are weighted and added to obtain a comprehensive score, and the health level of the user's physical health state is determined according to the comprehensive score.

[0040] Preferably, in S3, the intent reference opinion is determined in combination with the analysis intent of the user, including:

[0041] Based on the analysis intent of the user, the analysis field is determined, the physical evaluation index in the analysis field is determined, and the index weight of the physical evaluation index is determined based on the importance of the analysis field to the physical evaluation index;

[0042] According to the standard respiratory characteristics and standard emotion characteristics of the physical evaluation index under the standard state, in combination with the respiratory frequency change trend and the emotion change trend of the user, the physical evaluation index value is determined, and the physical evaluation weighted index value is determined based on the index weight;

[0043] Based on the physical evaluation weighted index value, the physical evaluation feature map of the user in the analysis field is constructed;

[0044] Based on the analysis intent of the user, the intent reply focus to the user is determined, the intent reply focus is associated with the physical evaluation feature map, the marking of the physical evaluation feature map is determined according to the association result, the feature extraction is performed from the physical evaluation feature map based on the marking result, and the key feature map is obtained;

[0045] The evaluation features in the key feature map are input into the intelligent opinion reply model, and the corresponding initial reference opinion is output;

[0046] Based on the physical health state of the user, the feasibility of the initial reference opinion is judged;

[0047] If feasible, the initial reference opinion is taken as the intent reference opinion;

[0048] Otherwise, based on the physical health state, the key feature map is modified to obtain the latest key feature map, until the physical health state of the user meets the reference opinion determined by the latest key feature map, and the reference opinion determined by the latest key feature map is taken as the intent reference opinion.

[0049] A health analysis device through timbre tone, comprising:

[0050] A collection transmission module is used for collecting voice signals of a user, performing analog-digital conversion on the voice signals to obtain electric signals, storing the electric signals to a specified area, and uniformly transmitting the electric signals.

[0051] A signal analysis module is used for performing signal processing on the transmitted electric signals, performing timbre tone analysis on the processed electric signals, and obtaining timbre tone characteristics of the user.

[0052] A health evaluation module is used for evaluating a physical health state of the user according to the timbre tone characteristics of the user, and determining an intention reference opinion in combination with an analysis intention of the user.

[0053] Compared with the prior art, the present application has the following beneficial effects:

[0054] The scheme uses voice recording to collect normal communication voice of a user, performs filtering and frequency and time domain and semantic analysis on the recorded voice through network transmission, analyzes the voice color and tone and basic information such as breathing through the height of the frequency, can obtain useful information about the voice and physical health, and the core technology mainly includes the following three aspects: first, voice data collection, a voice signal is collected through a specified microphone, then the collected sound signal is converted into an electric signal, then the electric signal is stored to a specified device, and then the collected data is transmitted to a mobile device or a cloud server or other data processing device; second, voice data processing, the collected electric signal data is filtered, denoised and processed, the data quality and accuracy are improved, and the data also needs to be analyzed and processed to extract useful timbre, tone and breathing and physical health information, for example, the size of the timbre, the height of the tone, the breathing frequency trend and the like can be analyzed, and then whether the person's breathing is stable, balanced and coordinated is evaluated. Third, data analysis and application, the processed data is analyzed and applied to evaluate the physical health status of the person, and reference opinions are provided for related medical, rehabilitation, sports and other fields, for example, the voice data can be analyzed to evaluate the emotional, stress, psychological characteristics and other physical indicators of the person, and the risk of sports injury, falling and the like can also be monitored to provide scientific data support and reference opinions for related fields.

[0055] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be achieved and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0056] The technical solutions of the present application will be further described in detail below with the help of the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0057] The accompanying drawings are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and are used to explain the application, but are not intended to limit the application. In the drawings:

[0058] Figure 1 A flow chart of a health analysis method through timbre tone in an embodiment of the application;

[0059] Figure 2 A flow chart of collecting a user's voice signal in an embodiment of the application;

[0060] Figure 3 A structural diagram of a health analysis device through timbre tone in an embodiment of the application. DETAILED DESCRIPTION

[0061] The preferred embodiments of the application are described below in conjunction with the accompanying drawings, and it should be understood that the preferred embodiments described here are only used to explain and illustrate the application, and are not used to limit the application.

[0062] Embodiment 1:

[0063] The embodiment of the application provides a health analysis method through timbre tone, as shown in the accompanying drawings, which comprises the following steps: Figure 1

[0064] S1: Collect a user's voice signal, and perform analog-to-digital conversion on the voice signal to obtain an electric signal, store the electric signal in a designated area, and uniformly transmit the electric signal;

[0065] S2: Perform signal processing on the transmitted electric signal, analyze the timbre tone of the processed electric signal, and obtain the timbre tone characteristics of the user;

[0066] S3: Evaluate the physical health status of the user according to the timbre tone characteristics of the user, and determine the reference opinion of the user's analysis intention in combination with the user's analysis intention.

[0067] ​The working principle of the above design scheme is as follows: This scheme uses voice recording to collect the user's normal conversational voice. The recorded voice is then transmitted over a network and filtered, and subjected to frequency, temporal, and semantic analysis. By analyzing the frequency levels, basic information such as timbre, tone, and breathing is obtained, useful information about voice and physical health can be acquired. The core technologies mainly include the following three aspects: First, voice data acquisition: Voice signals are collected through a designated microphone, then the collected sound signals are converted from digital to electrical signals, stored in a designated device, and then transmitted to mobile devices or cloud servers for data processing. Second, voice data processing: The collected electrical signal data is filtered and denoised to improve data quality and accuracy. Simultaneously, the data needs to be analyzed and processed to extract useful timbre, tone, breathing, and physical health information. For example, the loudness of the timbre, the pitch, and the trend of breathing frequency changes can be analyzed to assess whether a person's breathing is stable, balanced, and coordinated. The third aspect is data analysis and application. The processed data is analyzed and applied to assess a person's physical health and provide reference opinions for related medical, rehabilitation, sports and other fields. For example, by analyzing voice data, a person's emotions, stress, psychological characteristics and other physical indicators can be assessed. At the same time, it can also monitor the risks of sports injuries, falls and other risks, providing scientific data support and reference opinions for related fields.

[0068] The beneficial effects of the above design scheme are as follows: by collecting the user's voice signal, converting the voice signal into an electrical signal through analog-to-digital conversion, storing the electrical signal in a designated area, and transmitting the electrical signal uniformly, the system achieves the acquisition of the user's voice signal, performs signal processing on the transmitted electrical signal, analyzes the timbre and pitch of the processed electrical signal to obtain the user's timbre and pitch characteristics, realizes the analysis and management of the voice signal, and, based on the user's timbre and pitch characteristics and combined with the user's analytical intent, assesses the user's physical health status and intention reference opinions, thereby determining the user's physical condition and providing scientific data support and reference opinions for related fields.

[0069] Example 2:

[0070] Based on Example 1, this embodiment of the invention provides a health analysis method using timbre and tone, wherein in S1, as... Figure 2 As shown, the user's voice signal is collected, including:

[0071] The system uses two or more microphones to collect voice signals, compares the voice signals with the user's pre-stored historical voice signals, and retains the voice signals if they match; otherwise, the voice signals are discarded.

[0072] The signal strength of the speech signals collected by different microphones in the same time period is analyzed, and the speech signal collected by the microphone with the strongest signal strength in the same time period is taken as the final speech signal of the user.

[0073] In this embodiment, two or more microphones are placed or worn at different positions of the user, so as to collect the speech signal of the user from multiple directions and ensure the completeness of the collected speech signal.

[0074] In this embodiment, the purpose of comparing the speech signal with the historical speech signal of the user is to ensure that the collected speech signal is the speech of the user.

[0075] The beneficial effects of the above design scheme are as follows: two or more microphones are used to collect speech signals, so as to collect the speech signal of the user from multiple directions and ensure the completeness of the collected speech signal; then the speech signal is compared with the historical speech signal of the user, if they are consistent, the speech signal is retained, otherwise, the speech signal is rejected; the purpose is to ensure that the collected speech signal is the speech of the user and the accuracy of the collected speech signal, the signal strength of the speech signals collected by different microphones in the same time period is analyzed, and the speech signal collected by the microphone with the strongest signal strength in the same time period is taken as the final speech signal of the user, so that the final speech signal is obtained while reducing the amount of speech signal, improving the transmission and analysis efficiency of the speech signal.

[0076] Embodiment 3:

[0077] Based on the basis of embodiment 1, the application provides a health analysis method based on tone and timbre, S1, the speech signal is converted into an electrical signal by an analog-to-digital converter, and the electrical signal is stored in a designated area, and the electrical signal is uniformly transmitted, including:

[0078] The speech signal is converted into an electrical signal by an analog-to-digital converter;

[0079] The effectiveness of the electrical signal is analyzed to obtain an effective electrical signal, and the effective electrical signal is stored in a designated area;

[0080] When the effective electrical signal reaches a preset storage amount, and after receiving a data transmission instruction, the effective electrical signal is uniformly transmitted.

[0081] The beneficial effects of the above design scheme are: through the use of the analog-to-digital converter to convert the voice signal into an electrical signal, the effectiveness of the electrical signal is analyzed, the effective electrical signal is obtained, and the effective electrical signal is stored in a specified area, the effectiveness of the stored electrical signal is guaranteed, the signal quality of the stored voice signal is improved, when the effective electrical signal reaches the preset storage amount, and after receiving the data transmission instruction, the effective electrical signal is uniformly transmitted, the transmission efficiency and accuracy are guaranteed.

[0082] Embodiment 4:

[0083] Based on the basis of Embodiment 1, the present embodiment provides a timbre pitch analysis method, in S2, the transmitted electrical signal is processed, the processed electrical signal is analyzed for timbre pitch, and the user's timbre pitch feature is obtained, including:

[0084] The transmitted electrical signal is filtered to obtain a plurality of filtered signals in different frequency bands, and the plurality of filtered signals are denoised to obtain a to-be-analyzed electrical signal;

[0085] The filtered loudness of the to-be-analyzed electrical signal is obtained, the filtered loudness and the wide dynamic range compression curve corresponding to the frequency band are used to calculate the filtered response gain of the to-be-analyzed signal;

[0086] The gain adjustment value of the to-be-analyzed signal is determined based on the difference between the filtered response gain and the frequency band reference gain, and the to-be-analyzed electrical signal is adjusted based on the gain adjustment value to obtain a target electrical signal;

[0087] The target electrical signal is input into the trained timbre pitch analysis model to obtain the user's timbre pitch feature.

[0088] In this embodiment, the frequency band reference gain is obtained based on historical experience.

[0089] The beneficial effects of the above design scheme are: through the use of the analog-to-digital converter to convert the voice signal into an electrical signal, the effectiveness of the electrical signal is analyzed, the effective electrical signal is obtained, and the effective electrical signal is stored in a specified area, the effectiveness of the stored electrical signal is guaranteed, the signal quality of the stored voice signal is improved, when the effective electrical signal reaches the preset storage amount, and after receiving the data transmission instruction, the effective electrical signal is uniformly transmitted, the transmission efficiency and accuracy are guaranteed.

[0090] Embodiment 5:

[0091] Based on the basis of embodiment 4, the embodiment of the application provides a health analysis method through timbre tone, inputting a target electric signal into a trained timbre tone analysis model to obtain timbre tone characteristics of a user, including:

[0092] The historical voice data of different historical users is extracted to obtain the first acoustic characteristics of the historical voice data, and the deep neural network is trained by using the first acoustic characteristics to obtain an initial voice recognition model.

[0093] The feature difference of the first acoustic characteristics between different historical users is obtained, the personalized weight of the historical user is determined based on the feature difference, and the user feature recognition model is established based on the historical user voice feature and the corresponding personalized weight.

[0094] The user feature recognition model and the initial voice recognition model are fused to obtain an initial timbre tone analysis model.

[0095] The first acoustic characteristics of the same historical user are maximized in similarity to obtain second acoustic characteristics, the first acoustic characteristics of different historical users are maximized in similarity based on the feature difference to obtain third acoustic characteristics.

[0096] The second acoustic characteristics are used to train and verify the initial timbre tone analysis model until the first verification result is met, and the training is stopped to obtain a second timbre tone analysis model.

[0097] The third acoustic characteristics are used to train and verify the second timbre tone analysis model until the second verification result is met, and the training is stopped to obtain a trained timbre tone analysis model.

[0098] The user feature is input into the user feature recognition model, the target recognition channel of the trained timbre tone analysis model is determined according to the feature recognition result, the target electric signal is input into the target recognition channel of the trained timbre tone analysis model, and the timbre tone characteristics of the user are obtained.

[0099] In this embodiment, the initial voice recognition model is used to realize preliminary recognition of different historical voice data.

[0100] In this embodiment, the user feature recognition model is used to realize recognition and classification of different voice users.

[0101] In this embodiment, the user feature recognition model and the initial speech recognition model are model fused to obtain an initial timbre and tone analysis model, including: based on the recognition feature classification of the user feature recognition model, connecting the user feature recognition model with multiple recognition channels of the initial speech recognition model, wherein one type of recognition feature corresponds to one recognition channel; and training and verifying the initial timbre and tone analysis model by using the second acoustic feature and the third acoustic feature, specifically training and verifying the multiple recognition channels.

[0102] In this embodiment, the second acoustic feature is a standard acoustic feature of the same historical user.

[0103] In this embodiment, the third acoustic feature is an acoustic feature highlighting the difference in user features.

[0104] In this embodiment, the trained timbre and tone analysis model realizes the analysis accuracy of different user timbre and tone analysis, and avoids the use of the initial speech recognition model in a general manner, thereby causing the user timbre and tone recognition to be not targeted.

[0105] The beneficial effects of the above design scheme are: by fusing and training the initial speech recognition model according to the historical speech data of different historical users, considering the acoustic feature difference of the same historical user, and considering the acoustic feature difference of different historical users, an ultimate timbre and tone analysis model is obtained, the timbre and tone analysis model is guaranteed to be targeted for different users, the accuracy of the ultimate timbre and tone feature is guaranteed, and accurate data basis is provided for user physical health assessment.

[0106] Embodiment 6:

[0107] Based on the basis of embodiment 5, the present embodiment provides a health analysis method based on timbre and tone, which model fuses a user feature recognition model and an initial speech recognition model to obtain an initial timbre and tone analysis model, including:

[0108] Based on the recognition feature classification of the user feature recognition model, connecting the user feature recognition model with multiple recognition channels of the initial speech recognition model, wherein one type of recognition feature corresponds to one recognition channel;

[0109] Training and verifying the initial timbre and tone analysis model by using the second acoustic feature and the third acoustic feature, specifically training and verifying the multiple recognition channels.

[0110] The beneficial effects of the above design scheme are: based on the identification feature classification of the user feature recognition model, the user feature recognition model is connected with multiple identification channels of the initial speech recognition model, wherein one type identification feature corresponds to one identification channel, the model fusion of the user feature recognition model and the initial speech recognition model is realized, the model basis is provided for acoustic analysis of different users, the second acoustic feature and the third acoustic feature are used for training and verifying the initial timbre pitch analysis model, specifically for training and verifying multiple identification channels, and the analysis accuracy of the trained timbre pitch analysis model is ensured.

[0111] Embodiment 7:

[0112] Based on the basis of embodiment 1, the application provides a health analysis method through timbre pitch, in S3, according to the timbre pitch feature of the user, the physical health state of the user is evaluated, including:

[0113] Based on the timbre pitch feature, the timbre change curve and the pitch change curve of the user are determined;

[0114] Based on the breathing frequency feature, the timbre change curve and the pitch change curve are analyzed to determine the breathing frequency trend of the user, based on the sound emotion feature, the timbre change curve and the pitch change curve are analyzed to determine the emotion trend of the user;

[0115] Based on the breathing frequency trend, the breathing state of the user is evaluated, and the breathing balance score and the breathing coordination score of the user are determined according to the evaluation result;

[0116] Based on the emotion trend, the mood state of the user is evaluated, and the good or bad mood score of the user is determined according to the evaluation result;

[0117] Based on the breathing balance score, the breathing coordination score and the good or bad mood score, the physical health state of the user is determined.

[0118] The beneficial effects of the above design scheme are: the timbre and pitch characteristics are used to determine the timbre change curve and the pitch change curve of the user; the timbre change curve and the pitch change curve are analyzed based on the breathing frequency characteristics to determine the breathing frequency change trend of the user, and the timbre change curve and the pitch change curve are analyzed based on the sound emotion characteristics to determine the emotion change trend of the user; the breathing state of the user is evaluated based on the breathing frequency change trend, and the breathing balance score and the breathing coordination score of the user are determined according to the evaluation result; the mood state of the user is evaluated based on the emotion change trend, and the good or bad mood score of the user is determined according to the evaluation result; the physical health state of the user is determined based on the breathing balance score, the breathing coordination score and the good or bad mood score, which realizes effective analysis of the timbre and pitch characteristics, determines the physical health state of the user by using the timbre and pitch characteristics, and ensures the accuracy of determining the physical health state.

[0119] Embodiment 8:

[0120] Based on the basis of embodiment 7, the embodiment of the application provides a health analysis method based on timbre and pitch, which determines the physical health state of the user based on the breathing balance score, the breathing coordination score and the good or bad mood score, comprising:

[0121] The breathing balance score, the breathing coordination score and the good or bad mood score are weighted and added to obtain a comprehensive score, and the health level of the physical health state of the user is determined according to the comprehensive score.

[0122] In this embodiment, the weighting values of the breathing balance score, the breathing coordination score and the good or bad mood score are set according to actual conditions.

[0123] In this embodiment, the higher the comprehensive score is, the higher the health level of the physical health state of the corresponding user is.

[0124] The beneficial effects of the above design scheme are: the breathing balance score, the breathing coordination score and the good or bad mood score are weighted and added to obtain a comprehensive score, and the health level of the physical health state of the user is determined according to the comprehensive score, which ensures the accuracy of determining the health level of the physical health state of the user.

[0125] Embodiment 9:

[0126] Based on the basis of embodiment 7, the embodiment of the application provides a health analysis method based on timbre and pitch, in S3, the intention reference opinion is determined in combination with the analysis intention of the user, comprising:

[0127] Based on the analysis intention of the user, the analysis field is determined, the physical evaluation index in the analysis field is determined, and the index weight of the physical evaluation index is determined based on the importance of the analysis field to the physical evaluation index;

[0128] According to the standard respiratory characteristics and the standard emotional characteristics of the physical evaluation index in the standard state, in combination with the respiratory frequency change trend and the emotional change trend of the user, a physical evaluation index value is determined, and a physical evaluation weighted index value is determined based on an index weight;

[0129] A physical evaluation feature map of the user in the analysis field is constructed based on the physical evaluation weighted index value;

[0130] Based on the analysis intention of the user, an intention reply focus is determined, the intention reply focus is associated with the physical evaluation feature map, a mark of the physical evaluation feature map is determined according to an association result, feature extraction is performed from the physical evaluation feature map based on a mark result, and a focus feature map is obtained;

[0131] The evaluation features in the focus feature map are input into an intelligent opinion reply model, and a corresponding initial reference opinion is output;

[0132] Based on the physical health state of the user, the feasibility of the initial reference opinion is judged;

[0133] If the initial reference opinion is feasible, the initial reference opinion is taken as an intention reference opinion;

[0134] Otherwise, based on the physical health state, the focus feature map is corrected to obtain a latest focus feature map, until the physical health state of the user meets a reference opinion determined by the latest focus feature map, and the reference opinion determined by the latest focus feature map is taken as an intention reference opinion.

[0135] In this embodiment, the analysis field is, for example, a medical field, a rehabilitation field, a sports field, and the like.

[0136] In this embodiment, the physical evaluation index is, for example, emotion, stress, psychological characteristics, and the like, and different analysis fields focus on different physical evaluation indexes, for example, the sports analysis field focuses more on the respiratory frequency, and the corresponding index weight is also large, and the medical analysis field focuses more on the psychological characteristics, and the corresponding index weight is also large.

[0137] In this embodiment, for example, the analysis intention of the user is to monitor and warn the exercise process, and the corresponding intention reply focus is the stability degree of the physical evaluation index change trend in the exercise process.

[0138] In this embodiment, the intelligent opinion reply model is obtained according to analysis field related information.

[0139] In this embodiment, based on the physical health state of the user, the feasibility of the initial reference opinion is specifically whether the physical health state of the user can be safely executed according to the initial reference opinion.

[0140] In this embodiment, the correction of the key feature spectrum based on the physical health status is specifically fine-tuning the evaluation features in the key feature spectrum according to the physical health status.

[0141] The beneficial effects of the above design scheme are: by replying to the key according to the analysis field and the intention of the user, combining the timbre and tone features to determine the physical evaluation weighted index value, establishing the physical evaluation feature spectrum according to the physical evaluation weighted index value, combining the intelligent opinion reply model to determine the initial reference opinion, and then combining the physical health status of the user to judge the feasibility of the initial reference opinion, the matching of the final intention reference opinion and the user is ensured, and reasonable reference opinion is provided for the user.

[0142] Embodiment 10:

[0143] The embodiment of the present application provides a health analysis equipment through timbre and tone, as shown in the figure, comprising: Figure 3

[0144] The acquisition and transmission module is used for acquiring the voice signal of the user, performing analog-digital conversion on the voice signal to obtain an electric signal, storing the electric signal to a specified area, and uniformly transmitting the electric signal.

[0145] The signal analysis module is used for signal processing on the electric signal transmitted, timbre and tone analysis on the processed electric signal, and obtaining the timbre and tone features of the user.

[0146] The health evaluation module is used for evaluating the physical health status of the user according to the timbre and tone features of the user, and determining the intention reference opinion in combination with the analysis intention of the user.

[0147] ​The working principle of the above design scheme is: the scheme uses voice recording to collect the normal communication voice of the user, filters and analyzes the frequency and time domain and semantics of the recorded voice through network transmission, analyzes the basic information such as tone and intonation and breathing through the high and low frequency, and can obtain useful information about voice and physical health. The core technology mainly includes the following three aspects: first, voice data collection, the voice signal is collected through a specified microphone, then the collected sound signal is converted into an electrical signal through a digital-to-analog converter, then the electrical signal is stored in a specified device, and then the collected data is transmitted to a mobile device or a cloud server or other data processing device; second, voice data processing, the collected electrical signal data is filtered, denoised and processed to improve data quality and accuracy, and the data also needs to be analyzed and processed to extract useful voice, tone and breathing and physical health information, such as analyzing the size of the voice, the high and low of the tone, the trend of the breathing frequency change, and then evaluating whether the person's breathing is stable, balanced and coordinated. Third, data analysis and application, the processed data is analyzed and applied to evaluate the person's physical health status and provide reference opinions for related medical, rehabilitation, sports and other fields, such as analyzing voice data to evaluate a person's mood, stress, psychological characteristics and other physical indicators, and also monitoring exercise injuries, falls and other risks to provide scientific data support and reference opinions for related fields.

[0148] The beneficial effects of the above design scheme are: by collecting the voice signal of the user, converting the voice signal into an electrical signal through analog-to-digital conversion, storing the electrical signal in a specified area, and uniformly transmitting the electrical signal, the voice signal of the user is collected, the transmitted electrical signal is processed, the processed electrical signal is analyzed for tone and intonation, the tone and intonation characteristics of the user are obtained, the voice signal is analyzed and managed, the tone and intonation characteristics of the user are combined with the analysis intention of the user to evaluate the physical health status and intention reference opinion of the user, the physical status of the user is determined, and scientific data support and reference opinions are provided for related fields.

[0149] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A health analysis device based on timbre and tone, characterized in that, include: The acquisition and transmission module is used to acquire the user's voice signal, convert the voice signal into an electrical signal through analog-to-digital conversion, store the electrical signal in a designated area, and transmit the electrical signal uniformly. The signal analysis module is used to process the transmitted electrical signals, analyze the timbre and pitch of the processed electrical signals, and obtain the user's timbre and pitch characteristics. The health assessment module is used to assess a user's physical health status based on their vocal timbre and tone characteristics, and to determine the intended reference opinions based on the user's analytical intent. The signal analysis module includes: The transmitted electrical signal is filtered to obtain multiple filtered signals in different frequency bands, and the multiple filtered signals are then denoised to obtain the electrical signal to be analyzed. Obtain the filtered loudness of the electrical signal to be analyzed, and calculate the filtered response gain of the signal to be analyzed based on the filtered loudness and the wide dynamic range compression curve corresponding to the frequency band. Based on the difference between the filter response gain and the frequency band reference gain, the gain adjustment value of the signal to be analyzed is determined. Based on the gain adjustment value, the electrical signal to be analyzed is adjusted to obtain the target electrical signal. The target electrical signal is input into the trained timbre and pitch analysis model to obtain the user's timbre and pitch characteristics, specifically: Extract the first acoustic features from the historical voice data of different historical users, and use the first acoustic features to train a deep neural network to obtain an initial speech recognition model. The feature differences of the first acoustic features between different historical users are obtained. Based on the feature differences, the personalized weights of historical users are determined. Based on the voice features of historical users and their corresponding personalized weights, a user feature recognition model is established. The user feature recognition model and the initial speech recognition model are fused to obtain the initial timbre and tone analysis model. The similarity of the first acoustic features of the same historical user is maximized to obtain the second acoustic feature. The similarity of the first acoustic features of different historical users is maximized based on the feature differences to obtain the third acoustic feature. The initial timbre and tone analysis model is trained and validated using the second acoustic feature until the first validation result is met, at which point the training stops and the second timbre and tone analysis model is obtained. The second timbre and tone analysis model is trained and validated using the third acoustic feature until the second validation result is met, at which point training stops and the trained timbre and tone analysis model is obtained. The user's features are input into the user feature recognition model. Based on the feature recognition results, the target recognition channel of the target electrical signal in the trained timbre and tone analysis model is determined. The target electrical signal is then input into the target recognition channel of the trained timbre and tone analysis model to obtain the user's timbre and tone features.

Images