Design And Application Research Of Speech Feature Factor Recognition Technology For Traditional Chinese Medicine Voice Diagnosi

Background:Auscultation is an important part of the TCM diagnostic system of "look,smell,ask,and cut",reflecting the characteristics of TCM clinical syndrome differentiation.Among them,traditional Chinese medicine sound diagnosis is the main content of smell diagnosis.Revealing the objective laws of traditional Chinese medicine acoustic diagnosis based on modern information intelligence technology is an important topic in the modernization of traditional Chinese medicine research,which is conducive to excavating the scientific connotation of traditional Chinese medicine acoustic diagnosis and promoting the construction of an intelligent diagnosis and decision-making system.The range of human auditory receptors is 20Hz-2000Hz.Although it is relatively sensitive,the long-term phonological memory is poor.For some subtle auditory experiences,it is more difficult to express in words,or even lost.Based on the current acoustic sensing and signal analysis technology,some of the characteristics of acoustic diagnosis can be rediscovered.The focus of this paper is to explore the emotional-speech physiology supported by speech analysis technology,and based on traditional Chinese medicine classics such as"Five Tibetan Xiangyin,Can Consciousness" To realize the digitization of theories and indicators related to traditional Chinese medicine sound diagnosis and five tones.Based on the mining of objective indicators of acoustic diagnosis in traditional Chinese medicine,this study preliminarily discussed the application performance of the acoustic diagnosis system by taking the study of depression in emotional diseases as an example.Depression is one of the most prevalent serious mental illnesses in the 21st century.Its socialized morbidity trend,as well as the high recurrence rate,high fatality rate,low recognition rate,low medical consultation rate and other cognitive biases of the disease,have implications for the patient himself,his family,and his family.Even society will have a profound impact.At the same time,the pathogenesis of depression is still unclear,and it is difficult to cure.As a feature of TCM diagnosis and treatment,the theory of "preventive treatment" has a relatively systematic scheme for the identification,prevention and control of depression in the pre-depression stage.At the same time,modern psychology also provides positive counseling and intervention under the preliminary diagnosis results.Therefore,the application of convenient means to explore the biological characteristics of pre-depression plays an important role in TCM disease differentiation and Western medicine diagnosis.Based on the physiological characteristics of pre-depression pronunciation,this paper explores the physiological barriers of pre-depression under the effect of speech-resonance by exploring various macro and micro characteristics produced by the speech process and restoring their objective indicators.Purposes:1.Through the feature analysis of the voice signal,extract the voice characteristic factors that conform to the diagnosis of traditional Chinese medicine and the auditory characteristics of the human ear;2.Through data analysis and calculation,explore the quantitative identification method of depression based on speech features;3.Include the verification sample data to preliminarily judge the accuracy of the quantitative identification formula for depression;4.Based on the feature analysis of speech signal,the state identification system of"sound-image-viscera" in traditional Chinese medicine is initially constructed.Methods:1.Innovatively adopts the resonance model R(z)to extract the infrasound band signal in the speech;completes the digitization of the speech signal through sampling,quantization and short-term window processing;through short-term energy analysis,short-term average zero-crossing rate,short-term auto Correlation function and short-term average amplitude difference function for time domain analysis of speech signal;Fourier spectrum,autocorrelation function,power spectrum and cepstrum are obtained through Fourier transform and its inverse transform,and frequency domain analysis of speech signal is performed Incorporating the syndrome differentiation methods of Sanjiao,Qi,blood and body fluids,eight principles,and five-tone zang-fu syndrome,etc.,the characteristic factors of speech that are in line with the diagnostic characteristics of traditional Chinese medicine are obtained.2.Collect 32 depressive emotional samples for 15 days in the morning,fall asleep,and before and after lunch in a fixed pattern,and quantitatively evaluate their mental state under the guidance of traditional Chinese medicine physicians and psychiatrists.The speech characteristic factors with significant differences before and after getting up in the morning,falling asleep,and lunch were extracted respectively,and the difference between the characteristic factors and the quantified depressive emotional state were used for regression analysis to explore the speech characteristic factors related to depression,and establish their quantitative identification formula.3.Collected 9 cases of depression samples in a continuous state for 7 days in the morning,falling asleep and before and after lunch in a fixed pattern,and quantitatively evaluated their mental state under the guidance of traditional Chinese medicine physicians and psychiatrists.The voice data sample is included in the above quantitative identification formula,and the quantitative value of depression under the prediction of the quantitative identification formula is obtained.The accuracy of the identification formula is judged.4.Expand the above voice signal feature analysis,include the TCM consultation module,the "Five Yun and Six Qi" operator transformation module,and the MBTI personality test as the emotional feature evaluation module,and add voice signal processing and input design and identification result output.module.Results:1.Through the characteristic analysis of the speech signal,399 primary speech characteristic factors and 69 secondary speech characteristic factors are obtained;2.The analysis results of voice data samples in the morning and before going to bed show that there is a negative correlation between the presence of voice intensity and average proportion and the quantitative evaluation results of the mental state of the sample,and the sense of the main voice area is positively related to the quantitative evaluation results of the mental state of the sample.Can be negatively correlated with the quantitative evaluation results of the mental state of the sample.The quantitative identification formula is:Depressed mood degree(D1)=-1.567 × voice intensity and mean proportion+3.938 × voice main tone area sense-1.571 × voice main tone area energy+57.578,P<0.01;the analysis results of the speech data samples before and after lunch showed that the number of Chinese characters with phonetic syllables was positively correlated with the quantitative evaluation results of the mental state of the samples,and the average number of frames with the maximum auditory loudness per word among the 131 characters was positively correlated with the quantitative evaluation results of the mental state of the samples.It is negatively correlated,and the maximum energy frequency of the most sustained frame is positively correlated with the quantitative evaluation results of the mental state of the sample.The quantitative identification formula is:Depression degree(D2)=0.257 × number of Chinese characters in phonetic syllables-0.066 × 131 words per word.Word auditory loudness maximum frame number mean+0.066×the most sustained frame maximum energy frequency ratio+59.492,P<0.05.3.Validation data analysis results show that the correlation coefficients between the degree of depression D1 and D2 and the quantitative evaluation results of the mental state of the sample are:r1=0.513,P1<0.01 and r2=0.327,P2<0.01,4.Incorporate the TCM consultation module,the "five transport and six qi" operator transformation module,and the MBTI emotional feature evaluation module to further verify and calibrate the syndrome differentiation results based on the analysis of speech signal features.At the same time,voice signal processing and input terminal design are added to standardize the collection of voice data;a recognition result output module is added,and the output results include the direct feature components of voice and the identification results of TCM health status.The identification results of TCM health status include the multi-dimensional state reasoning results of "Five Sounds-Five Elements" based on audio analysis,the characterization results of pronunciation parts and Sanjiao syndrome differentiation,the analysis results of infrasound and viscera firmness,and the correlation between intonation and viscera syndrome differentiation.sexual outcome.Conclusions:1.Based on the characteristic analysis of speech signals,a total of 468 speech characteristic factors that conform to the diagnosis of traditional Chinese medicine and the auditory characteristics of the human ear are obtained,which can be combined with data analysis and processing to classify diseases;2.In this study,the depressive emotion-related speech characteristic factor was extracted,which has a certain quantitative correlation with depressive emotion,which has a positive effect on finding out the depression-related speech characteristic factor that can be used for objective syndrome differentiation;3.The verification data basically conforms to the dynamic prediction model,and the model has certain accuracy for the discrimination of depression;4.The TCM "audio-image-viscera" state identification system based on voice signal provides a possible digital diagnosis method for the automation of TCM diagnosis,and can be used as a TCM intelligent auxiliary diagnosis method.Innovations:1.Technological innovation:innovative use of voice linear generation model-resonance model A(z);2.Innovation in research methods:explore the differences in individual voice characteristics in different states;3.Data processing and analysis innovation:data processing and analysis combined with TCM syndrome differentiation method and speech signal characteristics;4.System innovation:comprehensively objective TCM sound image,multi-modal four diagnostic information and inquiries,and "five transports and six qi" operator data to construct a TCM "acoustic image-viscera" state identification system based on voice signals.