54 results about "Speech sounds" patented technology

An estimation method includes: obtaining a first voice feature group of a plurality of persons who speak a first language; obtaining a second voice feature group of a plurality of persons who speak a second language; obtaining a voice feature of a subject; correcting the voice feature of the subject according to a relationship between the first voice feature group and the second voice feature group; estimating, from the voice feature of the subject that has been corrected, an oral function or a cognitive function of the subject by using an estimation process for an oral function or a cognitive function based on the second language; and outputting a result of estimation of the oral function or the cognitive function of the subject.

The invention relates to the technical field of English pronunciation learning, and discloses an AI (artificial intelligence) English learning system capable of correcting pronunciation. The system comprises a multi-source voice acquisition layer, an acoustic feature analysis layer, a pronunciation quality evaluation layer, a correction decision generation layer and a biological feedback execution layer. The multi-source voice acquisition layer captures a multi-channel sound wave signal and a maxillofacial electromyographic signal in real time when a user pronounces, and generates an original pronunciation data stream; the acoustic feature analysis layer performs time-frequency domain conversion on original data, separates multi-scale acoustic features and constructs a three-dimensional feature tensor; the pronunciation quality evaluation layer compares the standard template library to output scores and phoneme offset vectors; the correction decision generation layer constructs a pronunciation error topological graph and generates a correction instruction set containing tongue position, airflow and vocal cord parameters; and the biological feedback execution layer synchronously monitors the physiological response and updates the weight of the template library. According to the system, accurate acquisition, analysis, evaluation, correction and dynamic optimization of pronunciation learning are realized, and the English pronunciation learning effect is improved.

The invention discloses an intelligent exoskeleton bone auxiliary conduction hearing aid system capable of being remotely controlled and having an electroencephalogram control function, and relates to the technical field of intelligent rehabilitation robots and hearing aid devices. The wearable mechanical part comprises a flexible actuator, an elastic fabric base material, an inertial measurement unit IMU, a plantar pressure sensor array, a surface myoelectricity EMG sensor and an electroencephalogram EEG sensor. And the information processing communication part comprises a central controller, a bone conduction hearing aid module, a multi-mode sensor, an embedded SoC, a DSP / AI accelerator, a 4G / 5G / Wi-Fi module and a power management unit. The flexible exoskeleton walking aid system and the bone conduction hearing aid module are integrated, the electroencephalogram EEG sensor is additionally arranged to achieve the electroencephalogram control function, and cooperation of the three functions of movement assistance, auditory perception and neural intention recognition is achieved. A user can directly control the exoskeleton action mode through electroencephalogram signals, meanwhile, intelligent assistance of lower limb joints is obtained in the walking process, and environment sounds and voice prompts are clearly received in a bone conduction mode. According to the system, the perception ability, the control flexibility and the social participation degree of old people or rehabilitation patients who are inconvenient to move and accompanied with hearing impairment in a complex environment are remarkably improved, and more natural and convenient man-machine interaction experience is provided for users.

The present invention relates to an apparatus (10) for the self-administration of therapies for the treatment of speech disorders comprising a stimulus delivery device (11,12) with a horizontal development substantially curved along an angular portion so as to outline a portion of a circle, comprising at least one panel (11) for the delivery of visual stimuli spatially distributed in an angular range substantially equal to the angular portion of horizontal development of the stimulus delivery device (11,12); at least one acoustic source (12) for the delivery of acoustic stimuli spatially distributed in an angular range substantially equal to the angular portion of horizontal development of the stimulus delivery device (11,12); and a local electronic processing unit configured to control the generation of visual and acoustic stimuli by means of the at least one panel (11) and the at least one acoustic source (12), and characterized in that it comprises at least one first acoustic detector (15) for detecting a reproduction of a verbal element produced by the patient, wherein the at least one first acoustic detector (15) is carried by the at least one panel (11) and positioned so as to detect acoustic signals generated in a space inside the portion of circle outlined by the stimulus delivery device (11,12).

Voice-interaction massage device for massaging at least one penis, comprising the following: a housing comprising a first massage chamber and a second massage chamber, each of the first massage chamber and the second massage chamber being configured to accommodate at least part of the penis; a massage module incorporated in the housing, wherein the massage module is configured to massage at least a part of the at least one penis that is inserted into at least one of the two massage chambers; a loudspeaker; and a first sensor configured to detect movements of the penis in the first massage chamber; a control unit that is electrically connected to the speaker and the first sensor; the control unit, as soon as the first sensor detects movements of the penis in the first massage chamber, activates the speaker to output an initial voice message.

The application provides a model training method and a bipolar disorder personalized treatment dialogue generation method and device, belonging to the technical field of deep learning, aiming at the problems that the current diagnosis accuracy of bipolar mood disorder is not high and the matching of treatment dialogue with patients is poor, the model training can be performed on a multi-modal language model according to a voice signal with significant characteristics of bipolar mood disorder, and a target classification model for bipolar mood disorder classification is obtained. Since the voice signal contains characteristics such as semantics, tone, pronunciation rhythm and other characteristics directly related to bipolar mood disorder expressed by the patient, the target classification model can be used for accurate classification of bipolar mood disorder, and when the target classification model is used for classification of bipolar mood disorder, the excessive dependence on the subjective diagnosis of doctors can be avoided, and the accuracy of the classification prediction of bipolar mood disorder is effectively improved.

The present disclosure relates to a speech disorder detection method, device, equipment and readable storage medium. By acquiring standard audio-visual materials, in response to the pronunciation operation of the to-be-detected object to the standard audio-visual materials, multi-modal pronunciation data is collected, audio acoustic features are extracted based on the pronunciation audio, video visual features are extracted based on the video of the face and oral cavity activity, the audio acoustic features, the video visual features and the demographic information coding data are fused to obtain a fusion feature vector, and based on the fusion feature vector and a pre-trained prediction model, a speech disorder detection result of the to-be-detected object is obtained. Compared with the prior art, the embodiment of the present disclosure can improve the accuracy and comprehensiveness of speech disorder detection, improve the diagnosis efficiency, reduce the dependence on professionals, reduce the burden of medical resources, and clearly determine the specific type of pronunciation problem, thereby providing a scientific basis for subsequent individualized intervention treatment.

The invention discloses a digital human generation system and method based on a language model, and relates to the technical field of artificial intelligence. A complete digital human pronunciation modeling path from voice signal acquisition, formant frequency extraction, deviation analysis and pronunciation stability evaluation to motion compensation control and three-dimensional animation fusion is realized. Compared with the existing mode of driving the digital human pattern only based on energy envelope or mouth shape classification, the method has the advantages that the multi-frame formant frequency of the user voice is extracted and processed, so that the digital human can generate dynamic lingual surface and tongue tip actions according to the sound channel change in the real pronunciation process of the user; therefore, the physical correspondence and linguistic consistency of the pronunciation actions of the digital human are improved. According to the method, the direct mapping between the pronunciation characteristics and the three-dimensional tongue actions is realized, and the authenticity of the pronunciation animation of the digital human and the guidance in a teaching scene are remarkably improved.

An object of the system according to the embodiment is to analyze a cry or an action of a pet and accurately convert an intention or an emotion into a human language.SOLUTION: A system includes a collection unit, an analysis unit, an estimation unit, and a conversion unit. The collection unit collects a cry or an action. The analysis unit analyzes the data collected by the collection unit. The estimation unit estimates an intention or a feeling on the basis of an analysis result obtained by the analysis unit. The conversion unit converts the intention or the feeling estimated by the estimation unit into human language.SELECTED DRAWING: Figure 1

This utility model discloses a voice tremor nursing teaching model in the field of nursing teaching aids. It includes a three-dimensional humanoid model with a left lung model and a right lung model positioned at the chest area. The left lung model is covered with a silicone membrane, forming a pleural cavity model. A catheter is located at the top of the left lung model, with one end extending into the pleural cavity model. Air or liquid is injected into the pleural cavity through the catheter to simulate pneumothorax or pleural effusion. The right lung model includes a detachable and replaceable lung cavity model, lung tumor model, and bronchial obstruction model. A bronchial model is located between the left and right lung models, with a microphone connected to the top of the bronchial model. The two forked ends of the bronchial model at the bottom are connected to the left and right lung models, respectively. This device can replace different models to simulate pathological states, allowing students to intuitively experience the differences in voice tremor under different pathological conditions, improving teaching effectiveness and students' clinical skills training level.

This invention discloses a humanized voice intervention method for inhalation errors in nasal and oral aerosol delivery devices, aiming to address the lack of humanization in existing voice intervention methods, which leads to patient misunderstanding and resistance. The method includes: establishing a patient-doctor corpus by combining expert experience, literature, and guidelines; fine-tuning a language parsing model to obtain an error-sensitive parsing model; using the error-sensitive parsing model to obtain a total intervention feature library and an urgency level sub-library; establishing a two-level query mechanism for patient inhalation errors to query intervention text in the urgency level sub-library; and fine-tuning the voice generation model based on the aforementioned patient-doctor corpus, incorporating new tokens, and evolving through prompting learning into a patient-friendly intervention voice generation model capable of generating intervention voices that are appropriate for the urgency level. The intervention voices generated by this invention can form a humanized voice intervention that approximates the tone of clinical medical staff, and can adapt the tone according to the urgency level of the error, effectively improving the patient's understanding and acceptance of the intervention suggestions.

The invention provides an artificial intelligence-based ear-nose-throat specialized clinical teaching effect evaluation system. The system comprises a natural language processing module, a computer vision analysis module, a virtual patient evaluation module, an otoscope image acquisition device, an image recognition module, an intelligent scoring and feedback module, a literature retrieval scoring module, a data management module and a visual presentation module. According to the system, multi-modal data of students in the processes of voice answering, skill operation, virtual inquiry, otoscope image recognition and literature retrieval are collected, automatic analysis and scoring are carried out, an ability evaluation report and personalized feedback are generated, standardization and intelligence of teaching evaluation are achieved, and the teaching efficiency is improved. The comprehensive ability evaluation device is suitable for comprehensive ability evaluation in the ear-nose-throat specialized medical teaching process. The system has the advantages of objective scoring, efficient processing, visual feedback and the like.

The invention provides an easily-confused emotion feature extraction method based on adaptive wavelet transform, and aims to improve the recognition performance of easily-confused emotions. According to the method, learnable parameters beta, gamma and a are introduced into a wavelet basis function, the time-frequency form of the wavelet basis function is dynamically adjusted, multi-scale time-frequency decomposition is carried out on input voice signals, two-dimensional tensor feature representation is obtained, and extracted features reserve time local changes of the voice signals and differences of different frequency bands in emotion distinguishing. Meanwhile, a projection gradient descent method is adopted to carry out constraint optimization on parameters, the stability and energy convergence of a primary function are ensured, and a gradient return updating mechanism is utilized to enable wavelet parameters to be adaptively optimized in the training process. According to the method, emotional characteristics with higher discrimination and robustness can be extracted, and the accuracy and generalization ability of emotion recognition are improved.

The invention relates to the technical field of rehabilitation assessment, in particular to a cerebral apoplexy rehabilitation assessment method and system based on voice multi-task learning, and the method comprises the following steps: carrying out the psychological emotion recognition of assessment voice data, and obtaining a cerebral apoplexy rehabilitation subjective assessment task used for perceiving the subjective rehabilitation feeling of an assessment object; performing physiological function recognition on the evaluation voice data to obtain a cerebral apoplexy rehabilitation objective evaluation task for perceiving the objective rehabilitation state of the evaluation object; and performing weight adaptive combination on the stroke rehabilitation subjective evaluation task and the stroke rehabilitation objective evaluation task by using a multi-task learning mechanism to obtain a stroke rehabilitation evaluation model. According to the method, objective function states and subjective emotional feelings fed back from the voices of the patient can be taken into consideration during rehabilitation evaluation, pain and discomfort of the stroke patient can be perceived during functional evaluation, the individual accuracy of stroke rehabilitation evaluation is improved, and the user experience is enhanced.

The invention discloses a voice anomaly detection method for pre-driving period screening of Parkinson's disease, which is mainly used for performing non-invasive pre-driving period screening of the Parkinson's disease on voice information of a subject, analyzing by using daily voice, and discovering early signs of the Parkinson's disease without invasive examination such as blood drawing, imaging and the like. The method provides a new way for large-scale population screening, reduces the detection cost, is good in patient compliance, is high in result interpretation, is combined with a knowledge graph or feature importance analysis, provides a basis for an anomaly detection result, enables a clinician to understand the correlation between voice anomaly and Parkinson pathology, and improves the detection accuracy. And the credibility of the AI result is improved.

The invention, which relates to a filter element (1) for respiratory gas conditioning with a speech function, is based on the objective of providing a filter element (1) for use with a tracheotomized or laryngectomized person, wherein the filter element (1) is robust, easy to operate, and inexpensive to manufacture. This objective is achieved by the housing (2) comprising a first housing part (5) and a second housing part (6) connected to the first housing part (5), the first opening (3) being arranged in the second housing part (6), the first housing part (5) being a single piece comprising an outer ring (7), a central upper part (8), and elastically deformable webs (9) arranged between the outer ring (7) and the upper part (8).

1. The name of the design product: morning inspection device. 2. The use of the design product: health detection in schools and other places, with functions such as face recognition, oral cavity photographing and temperature measurement. 3. The design points of the design product: the combination of shape and pattern. 4. The picture or photo that best indicates the design points: perspective view. 5. Other circumstances that need to be explained: application example: the morning inspection device is fixed at the entrance of each class. After sensing the human body information of the students, the instruction triggers the voice module to play the oral temperature measurement prompt. The face image of the student is recognized, and the oral temperature of the student is sensed and the oral temperature information is obtained. If the oral temperature is abnormal, an alarm prompt is played.

Providing a selection of speech features for building a model to detect medical conditions. [Solution] A mathematical model can be trained to diagnose a person's medical condition by processing the acoustic features 1021 and linguistic features 1022 of a person's voice. The performance of the mathematical model can be improved by appropriately selecting the features 1021 and 1022 used with the mathematical model. Features 1021 and 1022 can be selected by calculating a feature selection score 1031 for each acoustic feature 1021 and each linguistic feature 1022, and then using the score 1031 to select the feature 1021 or 1022 that gave the highest score 1031.

The application discloses a post-processing method and device for improving voice perception, equipment and medium. The pre-configured contrast stretching maximum value can limit the degree of contrast stretching, preventing signal distortion caused by excessive stretching. When determining the target contrast stretching parameter of any frequency, the pre-set contrast stretching preset parameter corresponding to the frequency is divided by the pre-determined contrast stretching maximum value, which can effectively reduce the energy of the non-sensitive frequency of the human ear and maintain the energy of the sensitive frequency of the human ear. Subsequently, the amplitude spectrum of the current voice frame is stretched by applying the target contrast stretching parameter corresponding to each frequency to obtain the stretched amplitude spectrum. The stretched amplitude spectrum maintains the energy of the sensitive frequency of the human ear while reducing the energy of the non-sensitive frequency, thereby avoiding excessive stretching of the amplitude of the sensitive frequency of the human ear and the problem of numerical overflow.

The application provides a bionic oral cavity sound production device, a robot and a dynamic bionic oral cavity sound production control method. The bionic oral cavity sound production device comprises a sound production element, the sound production element is arranged in a sound production box, the sound production box comprises a movable component, and the movable component is used for changing the form of the internal space of the sound production box. The dynamic bionic oral cavity sound production control method comprises the following steps: obtaining semantic information of a target pronunciation; determining a target phoneme sequence according to the semantic information; querying a preset phoneme-cavity form mapping database according to the phoneme sequence to obtain a movement parameter sequence of the movable component; and adjusting the spatial form of the sound production box according to the movement parameter sequence. The application drives the sound production box to change the form through the movable component, realizes dynamic adjustment matched with the physiological process of human sound production, and has the advantages of improving the naturalness of speech and the accuracy of pronunciation by controlling the cavity form in real time.

The application relates to the technical field of speech recognition, in particular to a speech-based emotion recognition method, which comprises the following steps: according to an input speech signal, frame division and windowing are carried out. The application can obtain more profound insights into emotional expression by decomposing the input speech signal into two physical sources of glottal excitation and vocal tract response for independent modeling and analysis, obtaining a vocal tract transfer function set via linear predictive coding operation, and applying inverse filtering to the original speech signal to reconstruct an approximate glottal pulse sequence, effectively stripping the influence of vocal tract resonance on the signal, so that perturbation parameters, open quotient and closed quotient and the like representing vocal cord vibration patterns can be directly calculated, at the same time, the vocal tract transfer function set is used to identify and track the dynamic trajectory of the formant, and the phase difference cosine mean between adjacent frames is combined to quantify the sound production stability, and subtle dynamic adjustment and control stability of sound production organs such as the oral cavity and tongue position caused by emotional changes are captured.

The invention discloses an ear-nose-throat disease early screening method. The method comprises the following steps: a voice acquisition step: acquiring a standardized voice sample of a user through an audio acquisition device; a feature extraction step: extracting multi-level acoustic features including vocal cord vibration features, resonance features, pronunciation stability features and depth representation features from the voice sample; and a baseline establishment step. According to the ear-nose-throat disease early-stage screening method and the ear-nose-throat disease early-stage screening system provided by the embodiment of the invention, the influence of individual voice difference on the screening result is effectively overcome by establishing the individual dynamic voiceprint base line; multi-level acoustic feature analysis is adopted, so that the accuracy of disease recognition is improved; continuous monitoring and dynamic risk assessment are realized, and possibility is provided for early lesion discovery; on the premise of protecting user privacy, model optimization is realized through federal learning; an objective and quantitative auxiliary diagnosis basis is provided for clinicians, and the diagnosis and treatment efficiency is improved.

A machine learning based patient voice monitoring and analysis system can reduce the need for patient hospitalization by early detection and treatment of health conditions such as acute decompensated heart failure.

The application discloses a method for fine-grained respiratory interaction based on visual and auditory detection, comprising the following steps: acquiring respiratory parameters based on respiratory types, and constructing a respiratory input method of an interactive vocabulary; based on the respiratory input method of the interactive vocabulary, designing several characteristic parameters of a respiratory vector representing respiratory interaction actions, and acquiring respiratory intensity, respiratory direction, whether inhaling, respiratory time, respiratory frequency, expression information and volume information from the several characteristic parameters; constructing a visual and auditory double-channel detection model, inputting the expression information and the volume information, and completing respiratory detection. The application realizes the recognition of multiple respiratory action characteristic parameters including respiratory action airflow intensity, respiratory airflow direction, respiratory airflow type and the like based on the combination of images and voices.

The invention discloses an intelligent device and method for assisting a vocal cord injury patient in sounding, and the method comprises the steps: collecting a fuzzy voice signal and an airflow pressure signal in the oral cavity of a user through an oral cavity hardware module; the fuzzy voice signal and the airflow pressure signal are converted into digital signals through an external control device and are preprocessed, the preprocessed digital signals pass through a double-branch feature fusion deep learning model based on an attention mechanism built in a mobile phone APP to obtain an adjusted sound signal, and the adjusted sound signal is output through the oral cavity hardware module. The auxiliary sounding of the vocal cord injury user is realized. According to the method, detail information such as mood and emotion of the patient can be accurately transmitted, the real expression intention of the patient can be distinguished and restored, the real intention of the user can be accurately expressed, the user can realize natural and smooth language communication, and the communication difficulty of the vocal cord injury user is solved.

The microstructure artificial laryngeal device provided by the application comprises a micro-porous structure audio sensor, a lower computer and an upper computer, wherein the micro-porous structure audio sensor is used for detecting a vibration signal of a throat and converting the vibration signal into a resistance signal; the lower computer is used for filtering and processing the resistance signal to remove noise interference; and the upper computer is used for accurately identifying audio information in the processed resistance signal. The micro-porous structure audio sensor is attached at a skin surface of a thyroid cartilage, the lower computer is worn at a neck, and the upper computer is worn at a small arm. The microstructure artificial laryngeal device provided by the application can be directly attached to a throat of a patient without being installed in the patient's body, thereby avoiding discomfort of the patient, in addition, the vibration signal of the throat can be better sensed through the micro-porous structure audio sensor, thereby completing sound detection, and in addition, the lower computer for removing noise interference is arranged, and the accuracy of speech recognition can be improved.

The application relates to the field of medical devices, and particularly discloses a tongue cancer postoperative speech-swallowing integrated rehabilitation training ball, which comprises an inflatable silica gel ball, a pressure sensing handle, a speech collecting unit, a control module and a timing linkage control unit. The timing linkage control unit acquires the pressure value generated by the tongue body compressing the ball, triggers speech collection and analysis when the pressure reaches a first threshold value, calculates the speech similarity score, releases the swallowing training lock and generates a prompt signal after the score reaches a clarity threshold value, and then identifies the swallowing trigger waveform characteristics and records the effective swallowing times during the ball-containing empty swallowing action of the patient. Through the timing closed-loop control of pressure, speech and swallowing, the application ensures that the patient can perform the swallowing training only after the tongue muscle strength and pronunciation clarity meet the standards, effectively reduces the risk of aspiration, and realizes the integrated safe cooperation of tongue muscle resistance, speech rehabilitation and swallowing training.