Intelligent interaction system for old people

Abstract

The invention discloses an intelligent interaction system for old people, and relates to the technical field of intelligent interaction systems for old people. The system architecture of the intelligent interaction system comprises a user interface module, a voice recognition module, a personalized vocabulary setting module, a dialect recognition module and an output module. According to the system, an administrator is allowed to flexibly add, delete and modify personalized vocabularies according to specific conditions and requirements of old people, so that the usability and the pertinence of the system are improved; the system adopts a neural network model based on deep learning to perform dialect recognition, can automatically learn and adapt to features of different dialects, ensures accurate recognition of the dialects through the steps of data preprocessing, feature extraction and model training, collects the response state of the old after playing related intelligence development test content through related intelligent equipment, and performs intelligent intelligence development test on the old. And the physical condition of the user is judged and uploaded to other intelligent receiving terminals, so that the use experience is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent interaction systems for the elderly, and in particular to an intelligent interaction system for the elderly. Background Art

[0002] As the population ages, the elderly population has a growing demand for intelligent interactive systems. Existing interactive systems often have the following problems:

[0003] Complex operation: The operation interface of many systems is complex and difficult for the elderly to understand and use;

[0004] Lack of personalization: General interactive systems cannot meet the personalized needs of the elderly, especially in terms of dialect recognition and specific vocabulary understanding.

[0005] After searching, the Chinese patent application number CN201810460399.X discloses a voice interaction method and system for wearable electronic devices for the elderly, which includes querying the user's identity and age on the cloud server after obtaining the user's fingerprint; if the user's age is within the preset elderly age range, the wearable electronic device enters the elderly interaction mode; in the elderly interaction mode, the wearable electronic device slows down the speed of playing the voice by a preset multiple; the wearable electronic device increases the volume of the played voice by a preset multiple; and when the wearable electronic device obtains the user's control voice, the user's control voice is corrected. The voice interaction method in the above patent has the following shortcomings: it cannot recognize dialects well and cannot meet the personalized needs of the elderly well. Summary of the invention

[0006] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent interactive system for the elderly.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] An intelligent interactive system for the elderly, the system architecture of the intelligent interactive system includes: a user interface module, a speech recognition module, a personalized vocabulary setting module, a dialect recognition module and an output module;

[0009] Among them, the user interface module is used to receive user input information; the speech recognition module is used to convert the user input information into text information; the personalized vocabulary setting module is used to set specific personalized vocabulary for the user according to the settings of the administrator; the dialect recognition module is used to recognize the user's dialect; the output module is used to display the recognition results to the user in an appropriate manner;

[0010] The personalized vocabulary setting of the intelligent interactive system is set based on the identity of the administrator; the administrator enters the personalized vocabulary setting interface through the user interface module to set a specific personalized vocabulary for the user;

[0011] The dialect recognition module of the intelligent interactive system adopts a neural network model based on deep learning, and realizes automatic recognition of dialects by learning and training a large amount of dialect data; when the dialect recognition module performs dialect recognition, it specifically includes the following steps:

[0012] S1: Data preprocessing: preprocess the input speech signal, including denoising, framing, and feature extraction operations;

[0013] S2: Feature extraction: extracting Mel frequency cepstral coefficients from the preprocessed speech signal as feature vectors;

[0014] S3: Model training: Use a large amount of dialect data to train the neural network model so that the model can learn the characteristics of different dialects;

[0015] S4: Model prediction: Input the speech signal to be recognized into the trained neural network model and output the corresponding dialect category.

[0016] Preferably: the output module of the intelligent interactive system adopts different display methods according to different recognition results; specifically: for ordinary text information, it is directly displayed on the screen; for personalized vocabulary, it is highlighted using special fonts or colors so that users can recognize it more easily; in addition, the output module supports the voice broadcast function, and plays the recognition results in the form of voice.

[0017] Preferably, the personalized vocabulary setting of the intelligent interactive system specifically includes:

[0018] ① Administrator identity verification: including username and password login and biometric recognition; after passing the verification, the administrator obtains the authority to access and modify the personalized vocabulary library;

[0019] ② User interface design: Administrators can add, delete or modify personalized vocabulary through the user interface. The user interface includes a search function for quickly locating specific vocabulary. In addition, batch import and export functions are provided to facilitate administrators to update vocabulary on a large scale.

[0020] ③Data storage: Personalized vocabulary and related information are stored in the database; the database is backed up regularly.

[0021] Preferably, the dialect recognition of the intelligent interactive system specifically includes:

[0022] ① Speech signal preprocessing: Before the dialect is recognized, the input speech signal is preprocessed;

[0023] ② Feature extraction: Extract feature vectors from preprocessed speech signals; extract feature vectors based on Mel frequency cepstral coefficients, linear prediction coding coefficients, and cepstral mean subtraction;

[0024] ③Model training and prediction: Use deep learning framework to build neural network model.

[0025] Preferably: the speech signal preprocessing includes reducing background noise by noise reduction, improving speech signal quality by speech enhancement, segmenting continuous speech signals into short time frames based on framing, and emphasizing high frequency parts by pre-emphasis to compensate for energy loss during sound transmission.

[0026] Preferably: the model training and prediction specifically adopts TensorFlow or PyTorch; it comprises convolutional layers, pooling layers and fully connected layers for automatically learning and extracting high-level features in speech signals; during the training phase, the model adjusts internal parameters by minimizing the difference between the predicted output and the actual label; after the training is completed, the model accurately classifies the speech signal into dialects.

[0027] Preferably, the intelligent interactive system performs learning and optimization based on a personalized vocabulary dynamic learning mechanism of user behavior data, and the specific steps are as follows:

[0028] S11: Data collection: The system records daily conversations through the speech recognition module and stores them in the cloud server;

[0029] S12: Data analysis: Use natural language processing technology to analyze the conversation content and extract high-frequency words, phrases and their context information;

[0030] S13: Model training: The extracted data is input into the personalized vocabulary dynamic learning model for training, so that the model can automatically identify and update the personalized vocabulary library.

[0031] Preferably: after the model is trained, the personalized vocabulary library is regularly updated according to the training results of the model.

[0032] Preferably: the intelligent interactive system is based on a multimodal interactive function and interacts with the user through a variety of perception methods; the specific implementation methods include:

[0033] ① Gesture recognition: Using computer vision technology, the camera captures the user's gestures and converts them into corresponding instructions or operations;

[0034] ② Expression recognition: Analyze the user’s emotional state through facial expression recognition technology and adjust the system’s feedback method based on the emotion.

[0035] The beneficial effects of the present invention are:

[0036] 1. The system of the present invention allows administrators to flexibly add, delete and modify personalized vocabulary according to the specific circumstances and needs of the elderly, thereby improving the usability and pertinence of the system; the system uses a neural network model based on deep learning for dialect recognition, which can automatically learn and adapt to the characteristics of different dialects; through data preprocessing, feature extraction and model training steps, it ensures accurate recognition of dialects, thereby improving the universality and practicality of the system.

[0037] 2. The present invention supports multiple perception modes for interaction, uses computer vision technology to capture gestures and convert them into commands, and uses facial expression recognition technology to analyze emotional states and adjust system feedback methods, thereby enhancing the richness and interactivity of the user experience.

[0038] 3. The system of the present invention has a personalized vocabulary dynamic learning mechanism based on user behavior data, which can automatically record the content of daily conversations and analyze high-frequency words, phrases and their contextual information, and regularly update the personalized vocabulary library to reflect the user's latest language habits and needs, thereby maintaining the timeliness and adaptability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 The present invention provides a flow chart for dialect recognition in an intelligent interactive system for the elderly. DETAILED DESCRIPTION

[0040] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0041] Embodiment 1:

[0042] An intelligent interactive system for the elderly, the system architecture of the intelligent interactive system includes: a user interface module, a speech recognition module, a personalized vocabulary setting module, a dialect recognition module and an output module;

[0043] Among them, the user interface module is used to receive user input information; the speech recognition module is used to convert the user input information into text information; the personalized vocabulary setting module is used to set specific personalized vocabulary for the elderly according to the settings of the administrator; the dialect recognition module is used to recognize the dialect of the elderly; the output module is used to display the recognition results to the user in an appropriate manner;

[0044] The personalized vocabulary setting of the intelligent interactive system is set based on the identity of the administrator, which requires certain permissions to avoid mistakes by the elderly; the administrator enters the personalized vocabulary setting interface through the user interface module to set specific personalized vocabulary for the elderly;

[0045] These personalized words can be words, phrases or sentences that are closely related to the lives of the elderly, such as "take medicine", "measure blood pressure", etc.; administrators can flexibly add, delete and modify personalized words according to the actual situation and needs of the elderly.

[0046] The dialect recognition module of the intelligent interactive system adopts a neural network model based on deep learning, and realizes automatic recognition of dialects by learning and training a large amount of dialect data; when the dialect recognition module performs dialect recognition, it specifically includes the following steps:

[0047] S1: Data preprocessing: preprocess the input speech signal, including denoising, framing, and feature extraction operations;

[0048] S2: Feature extraction: Mel frequency cepstral coefficients (MFCC) are extracted from the preprocessed speech signal as feature vectors;

[0049] S3: Model training: Use a large amount of dialect data to train the neural network model so that the model can learn the characteristics of different dialects;

[0050] S4: Model prediction: Input the speech signal to be recognized into the trained neural network model and output the corresponding dialect category.

[0051] The output module of the intelligent interactive system adopts different display methods according to different recognition results; for ordinary text information, it is directly displayed on the screen; for personalized vocabulary, special fonts or colors are used to highlight it so that users can recognize it more easily; in addition, the output module supports the voice broadcast function, and plays the recognition results in the form of voice.

[0052] The personalized vocabulary setting of the intelligent interactive system specifically includes:

[0053] ① Administrator identity verification: including username and password login and biometric recognition; after passing the verification, the administrator obtains the authority to access and modify the personalized vocabulary library;

[0054] ② User interface design: Administrators can add, delete or modify personalized vocabulary through the user interface. The user interface includes a search function for quickly locating specific vocabulary. In addition, batch import and export functions are provided to facilitate administrators to update vocabulary on a large scale.

[0055] ③Data storage: Personalized vocabulary and related information (such as pronunciation, meaning, usage scenarios, etc.) are stored in the database; the database is backed up regularly.

[0056] The dialect recognition of the intelligent interactive system specifically includes:

[0057] ① Speech signal preprocessing: Before the dialect is recognized, the input speech signal is first preprocessed, including reducing background noise through noise reduction, improving speech signal quality through speech enhancement, dividing continuous speech signals into short time frames based on framing, and emphasizing high-frequency parts through pre-emphasis to compensate for energy loss during sound transmission;

[0058] ② Feature extraction: Extract feature vectors from preprocessed speech signals; extract feature vectors based on Mel-frequency cepstral coefficients (MFCC), linear predictive coding coefficients (LPC), and cepstral mean subtraction (CMS);

[0059] ③Model training and prediction: Use a deep learning framework to build a neural network model, specifically TensorFlow or PyTorch; it contains convolutional layers, pooling layers, and fully connected layers for automatic learning and extraction of high-level features in speech signals; during the training phase, the model adjusts internal parameters by minimizing the difference between the predicted output and the actual label; after training, the model accurately classifies the speech signal into dialects.

[0060] The intelligent interactive system extracts feature vectors based on Mel-frequency cepstral coefficients, specifically:

[0061] The preprocessed speech signal is assumed to be x(n), where n represents a discrete time index;

[0062] Framing and windowing function: Segment the speech signal into overlapping frames, each with a length of N and a frame shift of R. Apply the window function w(n) to each frame, based on the Hamming window:

[0063]

[0064] The windowed signal is:

[0065] x w (m) = w(n) x((m-1)R+n)

[0066] 0≤n≤N-1

[0067] Where m is the frame index;

[0068] Fast Fourier transform, perform FFT on each frame to get the spectrum X(k):

[0069]

[0070] Mel filter bank, compute the energy of a Mel-frequency triangular filter bank:

[0071]

[0072] Among them, h(m,k) is the boundary function of the Mel filter:

[0073]

[0074] B(m,k)=f mel (k)-f mel (k-1)

[0075]

[0076] Mel energies are converted to MFCCs via discrete cosine transform:

[0077]

[0078] Here, K is the number of filters.

[0079] Embodiment 2:

[0080] An intelligent interactive system for the elderly. Based on Example 1, this embodiment uses a personalized vocabulary dynamic learning mechanism based on user behavior data to perform learning and optimization. The specific steps are as follows:

[0081] S11: Data collection: The system records daily conversations through the speech recognition module and stores them in the cloud server;

[0082] S12: Data analysis: Use natural language processing technology to analyze the conversation content and extract high-frequency words, phrases and their context information;

[0083] S13: Model training: input the extracted data into the personalized vocabulary dynamic learning model for training, so that the model can automatically identify and update the personalized vocabulary library;

[0084] S14: Automatic update: Based on the training results of the model, the personalized vocabulary library is updated regularly to ensure that it always reflects the user's latest language habits and needs.

[0085] The intelligent interactive system is based on multimodal interactive functions and interacts with the elderly through multiple perception methods (such as voice, text, gestures, expressions, etc.); the specific implementation methods include:

[0086] ① Gesture recognition: Using computer vision technology, the camera captures the gestures of the elderly and converts them into corresponding instructions or operations; for example, the elderly can answer or hang up the phone by waving their hands, or make selections by gesturing at numbers, etc.

[0087] ② Expression recognition: Through facial expression recognition technology, the emotional state of the elderly can be analyzed, and the system’s feedback method can be adjusted according to the emotions; for example, when the elderly show confusion, the system can automatically switch to a more concise and clear explanation mode.

[0088] The present invention uses relevant intelligent devices to play relevant intelligence test content, and then collects the reaction status of the elderly to determine their physical condition and upload it to other intelligent receiving terminals, thereby improving the user experience.

[0089] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An intelligent interactive system for the elderly, characterized in that: The system architecture of the intelligent interactive system includes: a user interface module, a speech recognition module, a personalized vocabulary setting module, a dialect recognition module and an output module; The user interface module is used to receive user input information; the speech recognition module is used to convert the user input information into text information; the personalized vocabulary setting module is used to set specific personalized vocabulary for the user according to the settings of the administrator; the dialect recognition module is used to recognize the user's dialect; the output module is used to display the recognition results to the user; The personalized vocabulary setting of the intelligent interactive system is set based on the identity of the administrator; the administrator enters the personalized vocabulary setting interface through the user interface module to set a specific personalized vocabulary for the user; The dialect recognition module of the intelligent interactive system adopts a neural network model based on deep learning, and realizes automatic recognition of dialects through learning and training of a large amount of dialect data.

2. The intelligent interactive system for the elderly according to claim 1, characterized in that: When the dialect recognition module of the intelligent interactive system performs dialect recognition, the following steps are specifically included: S1: Data preprocessing: preprocess the input speech signal, including denoising, framing, and feature extraction operations; S2: Feature extraction: extracting Mel frequency cepstral coefficients from the preprocessed speech signal as feature vectors; S3: Model training: Use dialect data to train the neural network model; S4: Model prediction: Input the speech signal to be recognized into the trained neural network model and output the corresponding dialect category.

3. The intelligent interactive system for the elderly according to claim 2, characterized in that: The output module of the intelligent interactive system highlights the personalized vocabulary; the output module plays the recognition result in the form of voice.

4. The intelligent interactive system for the elderly according to claim 3, characterized in that: The personalized vocabulary setting of the intelligent interactive system specifically includes: Administrator identity verification: including username and password login and biometric recognition; after passing the verification, the administrator obtains the authority to access and modify the personalized vocabulary library; User interface design: Administrators can add, delete or modify personalized vocabulary through the user interface. The user interface includes a search function for quickly locating specific vocabulary; it also provides batch import and export functions; Data storage: Personalized vocabulary and related information are stored in the database; the database is backed up regularly.

5. The intelligent interactive system for the elderly according to claim 4, characterized in that: The dialect recognition of the intelligent interactive system specifically includes: Speech signal preprocessing: Before the dialect is recognized, the input speech signal is preprocessed; Feature extraction: Extract feature vectors from preprocessed speech signals; extract feature vectors based on Mel frequency cepstral coefficients, linear prediction coding coefficients, and cepstral mean subtraction; Model training and prediction: Use deep learning framework to build neural network models.

6. The intelligent interactive system for the elderly according to claim 5, characterized in that: The speech signal preprocessing includes reducing background noise by noise reduction, improving speech signal quality by speech enhancement, dividing continuous speech signals into short time frames based on framing, and emphasizing high frequency parts by pre-emphasis to compensate for energy loss during sound transmission; The model training and prediction specifically adopts TensorFlow or PyTorch; it includes convolutional layers, pooling layers and fully connected layers, which are used to automatically learn and extract high-level features in speech signals.

7. The intelligent interactive system for the elderly according to claim 6, characterized in that: The model is trained and predicted. During the training phase, the model adjusts internal parameters by minimizing the difference between the predicted output and the actual label. After the training is completed, the model accurately classifies the dialect of the speech signal.

8. The intelligent interactive system for the elderly according to claim 7, characterized in that: The intelligent interactive system performs learning and optimization based on the personalized vocabulary dynamic learning mechanism of user behavior data. The specific steps are as follows: S11: Data collection: The system records daily conversations through the speech recognition module and stores them in the cloud server; S12: Data analysis: Use natural language processing technology to analyze the conversation content and extract high-frequency words, phrases and their context information; S13: Model training: The extracted data is input into the personalized vocabulary dynamic learning model for training, so that the model can automatically identify and update the personalized vocabulary library.

9. The intelligent interactive system for the elderly according to claim 8, characterized in that: After the model is trained, the personalized vocabulary library is regularly updated according to the training results of the model.

10. An intelligent interactive system for the elderly according to claim 9, characterized in that: The intelligent interactive system is based on a multimodal interactive function and interacts with the user through a variety of perception methods; the specific implementation methods include gesture recognition and expression recognition, wherein: Gesture recognition is implemented by using computer vision technology to capture the user's gestures through a camera and convert them into corresponding commands or operations; Expression recognition is achieved by analyzing the user's emotional state through facial expression recognition technology, and adjusting the system's feedback method based on the emotion.

Images