Intelligent device interaction system and method based on voice recognition

Abstract

The invention belongs to the technical field of equipment interaction, and particularly relates to an intelligent equipment interaction system and method based on voice recognition, and the system comprises a voice signal collection and processing module, an information configuration module, a voice recognition module, a control module and an information interaction module. According to the technical scheme, whether the user has the authority or not is judged by recognizing the first voiceprint feature # imgabs0 #, if the user is recognized not to have the authority, the control request is not responded, and therefore the situation that students use intelligent equipment for entertainment, and the teaching environment is affected can be avoided; and if the first voiceprint feature # imgabs1 # is matched with the second voiceprint feature # imgabs2 #, a control request is generated based on the first voice signal # imgabs3 #, so that the corresponding intelligent equipment responds and acts, otherwise, if the first voiceprint feature # imgabs4 # is not matched with the second voiceprint feature # imgabs5 #, an alarm prompt is generated, and the control request is not generated.

Description

Technical Field

[0001] The present invention belongs to the technical field of device interaction, and in particular to an intelligent device interaction system and method based on speech recognition. Background Art

[0002] As the core place for school education activities, classrooms play a vital role in the construction of smart campuses. At present, in most schools in my country, the use and management of classrooms still depends on the school's management center. If teachers need to use classrooms, they need to apply to the management center and obtain keys, and the use of equipment such as projection requires the assistance of professional managers.

[0003] This management model not only increases the workload of the management center, but also affects the overall operational efficiency of the school. After an in-depth analysis of the school's teaching needs, combined with the current trend of information-based teaching in colleges and universities and the school's development plan, it is necessary to use mobile technology, sensor technology, artificial intelligence technology, network technology and multimedia technology to upgrade classroom facilities and optimize the learning environment. With the increasing maturity and widespread application of voice recognition technology, such as identity authentication and smart home, a complete solution for applying voice recognition technology to smart classroom control has not yet been formed.

[0004] The prior art optimizes the process of teachers using smart devices in the classroom through keys and other means, that is, teachers can use the keys of smart devices in the classroom to turn on and use smart devices. However, the security of using smart devices through keys and other means cannot be guaranteed. After obtaining the keys, students can privately turn on smart devices and entertain themselves, which may cause damage to the smart devices in the classroom. At the same time, considering that the classroom is a place for educational activities, using smart devices for entertainment will also affect the teaching environment and is not conducive to students improving their learning level.

[0005] To this end, the present invention provides a smart device interaction system and method based on voice recognition. Summary of the invention

[0006] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0007] The technical solution adopted by the present invention to solve the technical problem is: the intelligent device interaction system based on speech recognition described in the present invention comprises:

[0008] Voice signal acquisition and processing module, used to acquire the user's first voice signal , and process to obtain the first voiceprint feature ;

[0009] Information configuration module, used to store the second voiceprint feature of the authorized user ;

[0010] A voice recognition module is used to convert the first voiceprint feature With the second voiceprint feature To compare:

[0011] When the first voiceprint feature Second voiceprint feature If the first similarity is greater than the first threshold, a control request is generated and issued; otherwise, an alarm is generated and no control request is generated;

[0012] A control module, used to receive a control request and instruct the smart device to perform a control action according to the control request;

[0013] The information interaction module is used to generate analysis results for unauthorized users.

[0014] Preferably, the processing obtains the first voiceprint feature The method is:

[0015] Get the first voice signal , , is the number of sample points;

[0016] For the first speech signal Sampling and quantification;

[0017] The first speech signal after sampling and quantization Perform noise suppression;

[0018] Extract the first speech signal Corresponding first voiceprint feature , as a real-time voiceprint feature;

[0019] The first voiceprint feature The method to obtain is:

[0020] For the first speech signal Divide the frame and set the number of sample points contained in each frame , get the frame ;

[0021]

[0022] in ;

[0023] For each frame Applying window functions Windowing is performed, which can be expressed as:

[0024]

[0025] Then the windowed frame Perform fast Fourier transform to get the spectrum , expressed as:

[0026]

[0027] From the spectrum Calculate Mel-frequency cepstral coefficients in , expressed as:

[0028]

[0029] Using Mel-frequency cepstral coefficients Characterizing the first speech signal The first voiceprint feature .

[0030] Preferably, the method for generating an alarm prompt is:

[0031] Get the recognition results of the speech recognition module:

[0032] When the first voiceprint feature Any second voiceprint feature No match, get the first voice signal Recording time;

[0033] Based on the first speech signal And record time, generate alarm prompts.

[0034] Preferably, the speech recognition module includes a voiceprint feature recognition model; the voiceprint feature recognition model is used to compare the first voiceprint feature With the second voiceprint feature The first similarity of

[0035] When the first similarity is greater than the first threshold, the first voiceprint feature is considered With the second voiceprint feature match;

[0036] Otherwise, the first voiceprint feature is considered With the second voiceprint feature No match;

[0037] The first voiceprint feature is compared With the second voiceprint feature The first similarity method is:

[0038] Get the first voiceprint feature With the second voiceprint feature ;

[0039] The first voiceprint feature With the second voiceprint feature Calculate the average eigenvector;

[0040] Calculate the first voiceprint feature according to the formula The average eigenvector of , and calculate the second voiceprint feature The average eigenvector of ;

[0041]

[0042]

[0043] in, The first voiceprint feature The number of eigenvectors of The second voiceprint feature The number of eigenvectors of

[0044] Calculate the first voiceprint feature according to the cosine first similarity formula With the second voiceprint feature The first similarity ;

[0045]

[0046] The first voiceprint feature is calculated With the second voiceprint feature The first similarity .

[0047] Preferably, the comparison of the first voiceprint feature With the second voiceprint feature The first similarity method also includes:

[0048] When the first similarity When the value is less than the first threshold and greater than the second threshold, analyze the first voiceprint feature The multi-dimensional characteristics of the

[0049] Analyzing the first voiceprint feature The multi-dimensional characteristics and comprehensive judgment method are as follows:

[0050] Get the first voice signal The rhythmic and intonation characteristics of

[0051] Synchronously processing to obtain rhythm features and intonation features of the second speech signal;

[0052] Based on the first speech signal The rhythm features and intonation features of the first speech signal are calculated by The second similarity with the second speech signal ;

[0053] If the second similarity If it is greater than the first threshold, then the first voiceprint feature is considered With the second voiceprint feature Match; otherwise, do not match;

[0054] When the second similarity When the value is greater than the first threshold, the first speech signal is obtained. The recording time of the first voice signal Generate exception prompts;

[0055] The second similarity To calculate the rhythm similarity , intonation similarity Weighted calculation of comprehensive similarity;

[0056] According to the formula:

[0057]

[0058] in, Rhythm similarity The weight of is the intonation similarity The weight of ;

[0059] The second similarity is calculated .

[0060] Preferably, the information configuration module further includes a voiceprint feature database and a scene database; the voiceprint feature database is used to store the second voiceprint feature ; The scene database is used to store scene modes corresponding to different users;

[0061] The scene mode is used to automatically complete the self-response status of multiple smart devices after the authorized user is successfully identified, and the self-response status includes on and off;

[0062] The response method of the scene mode is:

[0063] The user outputs the first voice signal ;

[0064] For the first speech signal Perform voiceprint feature recognition to determine whether the user is an authorized user:

[0065] If the user is an authorized user, the scene mode is turned on;

[0066] If the user is an unauthorized user, the scene mode will not be enabled, and an alarm will be generated and displayed synchronously in the scene mode of the authorized user.

[0067] Preferably, it also includes an information interaction module, the information interaction module is used to generate analysis results for unauthorized users; the information interaction module completes the data collection function based on the multimedia array, and the multimedia array includes an electronic touch screen;

[0068] The information interaction module includes a data collection and processing unit; the data collection and processing unit is used to collect the answer records of unauthorized users and generate feedback records based on the answer records:

[0069] The method for generating feedback records based on answer records is:

[0070] Obtain operation records generated by unauthorized users operating electronic touch screens;

[0071] Generate answer records based on unauthorized user names and corresponding operation records;

[0072] Collect all answer records and build feedback records based on dates;

[0073] The information interaction module also includes a data analysis unit, which is used to analyze the feedback records and generate analysis results; the analysis results are synchronously displayed in the scene mode of the authorized user.

[0074] Preferably, the multimedia array further includes a microphone and a loudspeaker, wherein the microphone is used for receiving sound and the loudspeaker is used for amplifying sound;

[0075] The information configuration module also includes an audio control unit, which is used to control the connection status of any microphone and the loudspeaker. When any microphone and the loudspeaker are in a connected state, the voice signal collected by any microphone will be amplified. When any microphone and the loudspeaker are not connected, any microphone is in a sound-only state.

[0076] Preferably, when any of the microphones and loudspeaker are in an unconnected state and any of the microphones picks up a voice signal of an unauthorized user, a voice prompt is generated based on the voice signal of the unauthorized user, the name of the unauthorized user and the time; the voice prompt is synchronously displayed in the scene mode of the authorized user.

[0077] A method for intelligent device interaction based on voice recognition, comprising the following steps:

[0078] Collect the user's first voice signal , and process to obtain the first voiceprint feature ;

[0079] The first voiceprint feature With the second voiceprint feature Make a comparison;

[0080] When the first voiceprint feature Second voiceprint feature If the first similarity is greater than the first threshold, a control request is generated and issued; otherwise, an alarm is generated and no control request is generated;

[0081] Receive control requests and make smart devices perform control actions according to the control requests.

[0082] The beneficial effects of the present invention are as follows:

[0083] The intelligent device interaction system and method based on voice recognition described in the present invention can detect the first voice signal sent by the user. Process and extract the first voiceprint feature And through the first voiceprint feature to determine whether the user has the authority. If the user does not have the authority, the control request will not be responded to, thereby preventing students from using smart devices for entertainment and affecting the teaching environment. With the second voiceprint feature Match, then based on the first voice signal Generate a control request to make the corresponding smart device respond and take action. Conversely, if the first voiceprint feature With the second voiceprint feature If they do not match, an alarm is generated but no control request is generated. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] The present invention will be further described below in conjunction with the accompanying drawings.

[0085] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0086] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0087] like Figure 1 As shown, an intelligent device interaction system based on voice recognition according to an embodiment of the present invention comprises a voice signal acquisition and processing module, an information configuration module, a voice recognition module, a control module and an information interaction module; the voice signal acquisition and processing module is used to acquire a first voice signal of a user , and process to obtain the first voiceprint feature ; The information configuration module is used to store the second voiceprint feature of the authorized user The speech recognition module is used to convert the first voiceprint feature With the second voiceprint feature Compare: When the first voiceprint feature Second voiceprint feature If the first similarity is greater than a first threshold, a control request is generated and issued; otherwise, an alarm is generated and no control request is generated; the control module is used to receive the control request and instruct the smart device to perform a control action according to the control request; the information interaction module is used to generate analysis results for unauthorized users.

[0088] Based on the above, in one embodiment of the present invention, in order to optimize the use process and method of classroom smart devices, voice recognition is used to control the use of smart devices. In the application stage, teachers can configure permissions in advance and control the opening and closing of corresponding smart devices only by speaking. Compared with the key method, the process is simplified and the teaching efficiency is improved. In addition, in order to prevent students from using smart devices for entertainment without the supervision of teachers, the first voice signal sent by the user can be recognized by voice recognition technology when the user speaks. Process and extract the first voiceprint feature And through the first voiceprint feature to determine whether the user has the authority. If the user does not have the authority, the control request will not be responded to, thereby preventing students from using smart devices for entertainment and affecting the teaching environment. When the user sends a first voice signal After that, the system will respond to the user's first voice signal Collect and process to extract the first voiceprint feature , and then the first voiceprint feature With the second voiceprint feature Compare, if the first voiceprint feature With the second voiceprint feature Match, then based on the first voice signal Generate a control request to make the corresponding smart device respond and take action. Conversely, if the first voiceprint feature With the second voiceprint feature If the first voiceprint feature does not match, an alarm is generated and no control request is generated. With the second voiceprint feature If they do not match, it is highly likely that an unauthorized user is making a control request to the smart device, so an alarm is generated to inform the sub-administrator or administrator of the phenomenon and take appropriate action;

[0089] Among them, the first voice signal is sent The users include authorized users and unauthorized users, and the first voice signal is a real-time voice signal. Similarly, the first voiceprint feature obtained by processing the first voice signal X is the real-time voiceprint feature, and the second voiceprint feature The pre-stored voiceprint feature is obtained after processing the second voice signal, where the second voice signal is a voice signal of an authorized user and the pre-stored voiceprint feature can be obtained after processing;

[0090] Exemplary method for obtaining pre-stored voiceprint features is:

[0091] recording a second voice signal, and processing the second voice signal;

[0092] Extracting the second voiceprint feature of the second speech signal , as a pre-stored voiceprint feature;

[0093] Match the pre-stored voiceprint features with the newly created authorized user to complete the creation of the new authorized user.

[0094] In one embodiment, the processing obtains the first voiceprint feature The method is:

[0095] Get the first voice signal , , is the number of sample points;

[0096] For the first speech signal Sampling and quantification;

[0097] The first speech signal after sampling and quantization Perform noise suppression;

[0098] Extract the first speech signal Corresponding first voiceprint feature , as a real-time voiceprint feature;

[0099] The first voiceprint feature The method to obtain is:

[0100] For the first speech signal Divide the frame and set the number of sample points contained in each frame , get the frame ;

[0101]

[0102] in ;

[0103] For each frame Applying window functions Windowing is performed, which can be expressed as:

[0104]

[0105] Then the windowed frame Perform fast Fourier transform to get the spectrum , expressed as:

[0106]

[0107] From the spectrum Calculate Mel-frequency cepstral coefficients in , expressed as:

[0108]

[0109] Using Mel-frequency cepstral coefficients Characterizing the first speech signal The first voiceprint feature .

[0110] Exemplarily, the first voice signal sent by the user , contains 9 sample points, expressed as follows:

[0111]

[0112]

[0113] Select frame length , that is, each frame contains 3 sample points, the first speech signal It is divided into the following frames:

[0114]

[0115]

[0116]

[0117] Then use the Hamming window For each frame After windowing, we get:

[0118]

[0119]

[0120]

[0121] Then for each frame after windowing Perform fast Fourier transform, simplified as:

[0122]

[0123]

[0124]

[0125] From the spectrum Calculate Mel-frequency cepstral coefficients in , expressed as:

[0126]

[0127]

[0128]

[0129] Based on the above, the first speech signal is obtained Mel-frequency cepstral coefficients .

[0130] In one embodiment, the method for generating an alarm prompt is:

[0131] Get the recognition results of the speech recognition module:

[0132] When the first voiceprint feature Any second voiceprint feature No match, get the first voice signal Recording time;

[0133] Based on the first speech signal And record time, generate alarm prompts.

[0134] Based on the above, since unauthorized users operating smart devices privately in the classroom may accelerate the damage of smart devices and affect the teaching environment, it is necessary to prohibit such behavior, and when such behavior occurs, it is also necessary to monitor such behavior and provide feedback to authorized users. In one embodiment of the present invention, when the voice recognition module outputs the first voiceprint feature With the second voiceprint feature After the unmatched recognition result, the unmatched first voiceprint feature The corresponding first speech signal The alarm prompt is generated together with the recording time, and is displayed synchronously in the scene mode opened after the authorized user recognizes it. The authorized user can view it, and the viewing content includes the first voice signal And record time. It is worth noting that if there is a camera in the classroom, the recognition result can also be the first voiceprint feature With the second voiceprint feature If there is no match, a short video will be automatically taken and the first voice signal , record the time and generate an alarm prompt for the administrator or sub-administrator to view. Through the above method, if an unauthorized user operates the smart device privately, the sub-administrator or administrator will identify it and push the alarm prompt in time, so as to notify the sub-administrator or administrator of such behavior.

[0135] In one embodiment, the speech recognition module includes a voiceprint feature recognition model; the voiceprint feature recognition model is used to compare the first voiceprint feature With the second voiceprint feature The first similarity of

[0136] When the first similarity is greater than the first threshold, the first voiceprint feature is considered With the second voiceprint feature match;

[0137] Otherwise, the first voiceprint feature is considered With the second voiceprint feature No match;

[0138] The first voiceprint feature is compared With the second voiceprint feature The first similarity method is:

[0139] Get the first voiceprint feature With the second voiceprint feature ;

[0140] The first voiceprint feature With the second voiceprint feature Calculate the average eigenvector;

[0141] Calculate the first voiceprint feature according to the formula The average eigenvector of , and calculate the second voiceprint feature The average eigenvector of ;

[0142]

[0143]

[0144] in, The first voiceprint feature The number of eigenvectors of The second voiceprint feature The number of eigenvectors of

[0145] Calculate the first voiceprint feature according to the cosine first similarity formula With the second voiceprint feature The first similarity ;

[0146]

[0147] The first voiceprint feature is calculated With the second voiceprint feature The first similarity .

[0148] Since the voiceprint recognition technology is used to control the smart device in the previous embodiment, how to quickly and accurately identify the voiceprint features is the basis for preventing unauthorized users from operating the smart device privately. In one embodiment of the present invention, the voice recognition module compares the first voiceprint feature with the second voiceprint feature based on the voiceprint feature recognition model. Specifically, after the voice recognition module obtains the first voiceprint feature, the first similarity can be calculated for the first voiceprint feature and the second voiceprint feature, and the first similarity is used to measure the similarity between the first voiceprint feature and the second voiceprint feature. When the first similarity is greater than the first threshold, it is considered that the first voiceprint feature is similar to the second voiceprint feature, and it can be determined that the first voiceprint feature matches the second voiceprint feature. Otherwise, it is determined that the first voiceprint feature does not match the second voiceprint feature. If the recognition result is a match, a control request can be generated according to the first voice signal, and any smart device can be controlled through the communication module to generate a corresponding control action. Otherwise, an alarm prompt is generated. When the Mel frequency cepstrum coefficient of the first voice signal is obtained Then, the Mel-frequency cepstrum coefficients of the first speech signal can be used to calculate Calculate the average feature vector of the first speech signal; exemplary,

[0149]

[0150]

[0151] For ease of understanding, set and are equal, then:

[0152]

[0153] According to the calculation, we get , indicating the first voiceprint feature With the second voiceprint feature Exactly the same, so the first voiceprint feature can be determined With the second voiceprint feature If the first voiceprint feature matches, a corresponding control request is generated according to the first voice signal corresponding to the first voiceprint feature; based on the above, by extracting the first voiceprint feature and the second voiceprint feature By comparison, it can be determined whether the user who issues the control request is an authorized user. If the first similarity is greater than the first threshold, it means that the user who issues the control request is an authorized user, otherwise it is an unauthorized user. Therefore, through the voiceprint feature recognition model, unauthorized users can be prevented from operating smart devices privately, which is conducive to protecting smart devices and creating a good teaching environment.

[0154] In one embodiment, the comparison of the first voiceprint feature With the second voiceprint feature The first similarity method also includes:

[0155] When the first similarity When the value is less than the first threshold and greater than the second threshold, analyze the first voiceprint feature The multi-dimensional characteristics of the

[0156] Analyzing the first voiceprint feature The multi-dimensional characteristics and comprehensive judgment method are as follows:

[0157] Get the first voice signal The rhythmic and intonation characteristics of

[0158] Synchronously processing to obtain rhythm features and intonation features of the second speech signal;

[0159] Based on the first speech signal The rhythm features and intonation features of the first speech signal are calculated by The second similarity with the second speech signal ;

[0160] If the second similarity If it is greater than the first threshold, then the first voiceprint feature is considered With the second voiceprint feature Match; otherwise, do not match;

[0161] When the second similarity When the value is greater than the first threshold, the first speech signal is obtained. The recording time of the first voice signal Generate exception prompts;

[0162] The second similarity To calculate the rhythm similarity , intonation similarity Weighted calculation of comprehensive similarity;

[0163] According to the formula:

[0164]

[0165] in, Rhythm similarity The weight of is the intonation similarity The weight of ;

[0166] The second similarity is calculated .

[0167] When the user's voice changes or he / she suffers from an illness, the first voiceprint feature corresponding to the first voice signal emitted by the user may be different, which may cause the first voiceprint feature to be unable to be accurately recognized, that is, the similarity with the second voiceprint feature is less than the first threshold, resulting in the inability to be recognized;

[0168] Therefore, in one embodiment of the present invention, a second threshold is also introduced. After calculating the similarity between the first voiceprint feature and the second voiceprint feature, if the similarity is greater than the second threshold and less than the first threshold, the first voiceprint feature needs to be further analyzed. The rhythm features and intonation features of the first speech signal are obtained, and the rhythm features and intonation features of the first speech signal are used to calculate the similarity with the rhythm features and intonation features of the second speech signal to obtain the rhythm similarity. and intonation similarity , and then use the formula to calculate the rhythm similarity , intonation similarity Weighted second similarity , through the second similarity If the second similarity is greater than the first threshold, it means that the first voiceprint feature matches the second voiceprint feature; in addition, after combining the multi-dimensional analysis of the first voiceprint feature and the second voiceprint feature, and based on this, it is judged that the first voiceprint feature matches the second voiceprint feature, in order to prevent imitation, a new log is also created based on the first voice signal and the recording time, and an abnormal prompt is generated.

[0169] In one embodiment, the information configuration module further includes a voiceprint feature database and a scene database; the voiceprint feature database is used to store the second voiceprint feature ; The scene database is used to store scene modes corresponding to different users;

[0170] The scene mode is used to automatically complete the self-response status of multiple smart devices after the authorized user is successfully identified, and the self-response status includes on and off;

[0171] The response method of the scene mode is:

[0172] The user outputs the first voice signal ;

[0173] For the first speech signal Perform voiceprint feature recognition to determine whether the user is an authorized user:

[0174] If the user is an authorized user, the scene mode is turned on;

[0175] If the user is an unauthorized user, the scene mode will not be enabled, and an alarm will be generated and displayed synchronously in the scene mode of the authorized user.

[0176] Based on the above, not all smart devices in the classroom need to be used during the teaching process, and due to differences in the types of teachers, the smart devices used are also different. For example, if teacher A is a mathematics teacher, the smart devices used by teacher A include interactive whiteboards, answer machines and other devices, while the smart devices used by teacher B include projectors, computers and sound systems. Therefore, the frequency of smart devices used by each teacher is combined to improve the feedback of each teacher on the use effect of the smart devices. In one embodiment of the present invention, the authorized users to operate the smart devices in the classroom are further optimized, and the user's authority editing can be performed through the information configuration module, specifically the editing authority of the scene mode. The scene mode is to set the self-response actions of multiple smart devices, that is, the teacher can realize the application of the scene mode through identification, which is exemplary. In the configuration module for the login information of the authorized user, the multiple smart devices included in the scene mode set by the authorized user can respond automatically, avoiding repeated output of control requests to make the smart devices respond. For example, if teacher A is a mathematics teacher, the smart devices automatically adjusted in the scene mode of teacher A include interactive whiteboards, answer machines and other devices, and if teacher B is a music teacher, the smart devices automatically adjusted in the scene mode of teacher B include projectors, computers and sound systems. When teacher A is successfully identified, multiple smart devices can respond automatically and start up, and when teacher B is successfully identified, multiple smart devices can also respond automatically and start up. Through the above method, before the start of future teaching, only successful identification is required to start the corresponding scene mode, thereby improving the teaching quality and improving the authorized user's feedback on the use effect of smart devices;

[0177] In addition, the method for building a scene mode is as follows: creating a new scene mode, setting the self-response status of multiple smart devices in the new scene mode, wherein the self-response status includes on and off; during the application of the scene mode, when the user sends a first voice signal After that, the speech recognition module determines the first speech signal Whether the user is an authorized user, if it is an authorized user, the scene mode of the authorized user is turned on; if the user is an unauthorized user, the scene mode is not turned on, and an alarm prompt is generated at the same time. Through the above method, after the authorized user logs in to the information configuration module, the scene mode can be set accordingly. Before subsequent teaching, the authorized user can be identified on the one hand, and the scene mode can be turned on synchronously on the other hand, so as to improve the authorized user's feedback on the use effect of the smart device. At the same time, in the scene mode turned on after the authorized user is identified, the alarm prompt and abnormal prompt will be displayed synchronously, so as to help the authorized user to accurately and timely grasp the situation in the classroom and optimize the teaching environment.

[0178] In one embodiment, it further includes an information interaction module, the information interaction module is used to generate analysis results of unauthorized users; the information interaction module completes the data collection function based on the multimedia array, and the multimedia array includes an electronic touch screen;

[0179] The information interaction module includes a data collection and processing unit; the data collection and processing unit is used to collect the answer records of unauthorized users and generate feedback records based on the answer records:

[0180] The method for generating feedback records based on answer records is:

[0181] Obtain operation records generated by unauthorized users operating electronic touch screens;

[0182] Generate answer records based on unauthorized user names and corresponding operation records;

[0183] Collect all answer records and build feedback records based on dates;

[0184] The information interaction module also includes a data analysis unit, which is used to analyze the feedback records and generate analysis results; the analysis results are synchronously displayed in the scene mode of the authorized user.

[0185] After the popularization of intelligent classrooms, the process of classroom tests for students, i.e., unauthorized users, can also be optimized through electronic touch screens, i.e., students can directly complete classroom tests on the electronic touch screens, and collect classroom tests to the information interaction module in the form of electronic data, and analyze the results of classroom tests, so as to facilitate teachers, i.e., authorized users, to understand the students' knowledge mastery. As an embodiment of the present invention, small tests conducted by teachers in class are collected using electronic touch screens, i.e., operation records corresponding to different users are obtained using the electronic touch screens, and answer records are generated based on user names and operation records, and the answer records can be traced back. At the same time, feedback records are generated based on the answer records of all users, and statistics can be used to quickly browse, analyze and count all answer records in the feedback records, and all questions are arranged in order of error rate, and analysis results are generated based on the error rate arrangement of the questions, and are displayed synchronously with the scene mode opened after the authorized user is identified, i.e., the teacher can obtain the analysis results before the next teaching starts, quickly understand the types of questions with high error rates of students, and based on this, consolidate and review the knowledge points corresponding to the questions with high error rates before the next class;

[0186] It is worth noting that the analysis results include not only the error rate of the same question for all unauthorized users, but also the error rate of the answer record for all unauthorized users. It can be understood that the error rate of the same question is the ratio of the number of errors of the same question for all unauthorized users to the total number, while the error rate of the answer record is the ratio of the number of wrong questions in the answer record of any unauthorized user to the total number. The summary of the error rate of the question can help teachers understand which knowledge points of the questions are difficult to master among all unauthorized users, and the summary of the error rate of the answer record can help teachers understand the degree of mastery of the knowledge points by any unauthorized user. By arranging the error rates of the answer records, it is possible to quickly understand which unauthorized users did not listen to the class seriously.

[0187] In addition, when the authorized user is successfully identified, the scene mode corresponding to the authorized user is automatically turned on due to the association of the scene mode. At the same time, as the scene mode is turned on, the analysis results of the test in the previous class can also be displayed synchronously for the authorized user to view. For example, the mathematics teacher did a quiz in the previous class. After the students answered the questions using the electronic touch screen, they uploaded the answers to the data collection and processing unit in the form of electronic data. After being analyzed by the data analysis unit, the analysis results of the quiz in the previous class are formed, and the analysis results can be automatically displayed when the scene mode is turned on in the next class of the mathematics teacher. Synchronous display enables mathematics teachers to understand the students' knowledge mastery in the previous class. The display of analysis results can also help mathematics teachers adjust the teaching content before the next class. For example, for questions with a high error rate shown in the analysis results, the calculation logic behind them can be explained in detail to help students quickly master and skillfully apply knowledge points. At the same time, according to the teacher's class content, the scene mode can automatically push exercises for students to take in-class tests. The pushed exercises come from the exercise repository, which is included in the information configuration module together with the voiceprint feature database and the scene database.

[0188] In one embodiment, the multimedia array further includes a microphone and a loudspeaker, wherein the microphone is used for receiving sound and the loudspeaker is used for amplifying sound;

[0189] The information configuration module also includes an audio control unit, which is used to control the connection status of any microphone and the loudspeaker. When any microphone and the loudspeaker are in a connected state, the voice signal collected by any microphone will be amplified. When any microphone and the loudspeaker are not connected, any microphone is in a sound-only state.

[0190] As an implementation mode of the present invention, a teacher needs to use a microphone to teach while imparting knowledge. By connecting the microphone to the loudspeaker, the teacher's voice can be amplified, so that students in any corner of the classroom can hear it. Therefore, during the teacher's class, the teacher's microphone, i.e., the authorized user, is always connected to the loudspeaker, while the student's microphone, i.e., the unauthorized user, is not connected to the loudspeaker. However, it is worth noting that during the course of teaching, the teacher may randomly ask any student to answer a question, and if the student's voice is too small, it may not be heard by all students in the classroom. Therefore, when asking any student to answer a question, a connection state can be established between the microphone and the loudspeaker based on the teacher's authorized user identity, thereby producing a sound amplification effect.

[0191] At the same time, based on the above, due to the setting of the microphone, the microphone positions of the authorized user and the unauthorized user are different, and the voice signal recorded by the microphone of the authorized user can be used as the first voice signal That is to say, only the voice signal recorded by the microphone at the authorized user's location can be used as the first voice signal to control the opening and closing of the smart device. .

[0192] In one embodiment, when any of the microphones and the loudspeaker are not connected and any of the microphones picks up a voice signal of an unauthorized user, a voice prompt is generated based on the voice signal of the unauthorized user, the name of the unauthorized user and the time; the voice prompt is synchronously displayed in the scene mode of the authorized user.

[0193] In addition, if some students chat with each other during the course of the teacher's lecture, it may affect the teaching environment and affect the students' grasp of knowledge points. In one embodiment of the present invention, since the microphone can only be used for sound reception when the microphone and the loudspeaker are not connected, if the student, i.e., the unauthorized user, chats during the class, the microphone can record the voice signal generated by the student's chat, and generate a voice prompt based on the name of the unauthorized user combined with the time, and display it synchronously with the scene mode opened after the authorized user is identified, so that the authorized user can view it;

[0194] A method for intelligent device interaction based on voice recognition, comprising the following steps:

[0195] Collect the user's first voice signal , and process to obtain the first voiceprint feature ;

[0196] The first voiceprint feature With the second voiceprint feature Make a comparison;

[0197] When the first voiceprint feature Second voiceprint feature If the first similarity is greater than the first threshold, a control request is generated and issued; otherwise, an alarm is generated and no control request is generated;

[0198] Receive control requests and make smart devices perform control actions according to the control requests.

[0199] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A smart device interaction system based on voice recognition, characterized in that: include: Voice signal acquisition and processing module, used to acquire the user's first voice signal , and process to obtain the first voiceprint feature ; Information configuration module, used to store the second voiceprint feature of the authorized user ; A speech recognition module is used to convert the first voiceprint feature With the second voiceprint feature Compare: When the first voiceprint feature Second voiceprint feature If the first similarity is greater than the first threshold, a control request is generated and issued; otherwise, an alarm is generated and no control request is generated; The control module is used to receive control requests and instruct the smart device to perform control actions according to the control requests.

2. The intelligent device interaction system based on voice recognition according to claim 1, characterized in that: The processing obtains the first voiceprint feature The method is: Get the first voice signal , , is the number of sample points; For the first speech signal Sampling and quantification; The first speech signal after sampling and quantization Perform noise suppression; Extract the first speech signal Corresponding first voiceprint feature , as a real-time voiceprint feature; The first voiceprint feature The method to obtain is: For the first speech signal Divide the frame and set the number of sample points contained in each frame , get the frame ; in ; For each frame Applying window functions Windowing is performed, which can be expressed as: Then the windowed frame Perform fast Fourier transform to get the spectrum , expressed as: From the spectrum Calculate Mel-frequency cepstral coefficients in , expressed as: Using Mel-frequency cepstral coefficients Characterizing the first speech signal The first voiceprint feature .

3. The intelligent device interaction system based on voice recognition according to claim 1, characterized in that: The method for generating an alarm prompt is: Get the recognition results of the speech recognition module: When the first voiceprint feature Any second voiceprint feature No match, get the first voice signal Recording time; Based on the first speech signal And record time, generate alarm prompts.

4. The intelligent device interaction system based on voice recognition according to claim 1, characterized in that: The speech recognition module includes a voiceprint feature recognition model; the voiceprint feature recognition model is used to compare the first voiceprint feature With the second voiceprint feature The first similarity of When the first similarity is greater than the first threshold, the first voiceprint feature is considered With the second voiceprint feature match; Otherwise, the first voiceprint feature is considered With the second voiceprint feature No match; The first voiceprint feature is compared With the second voiceprint feature The first similarity method is: Get the first voiceprint feature With the second voiceprint feature ; The first voiceprint feature With the second voiceprint feature Calculate the average eigenvector; Calculate the first voiceprint feature according to the formula The average eigenvector of , and calculate the second voiceprint feature The average eigenvector of ; in, The first voiceprint feature The number of eigenvectors of The second voiceprint feature The number of eigenvectors of Calculate the first voiceprint feature according to the cosine first similarity formula With the second voiceprint feature The first similarity ; The first voiceprint feature is calculated With the second voiceprint feature The first similarity .

5. The intelligent device interaction system based on voice recognition according to claim 4, characterized in that: The first voiceprint feature is compared With the second voiceprint feature The first similarity method also includes: When the first similarity When the value is less than the first threshold and greater than the second threshold, analyze the first voiceprint feature The multi-dimensional characteristics of the Analyzing the first voiceprint feature The multi-dimensional characteristics and comprehensive judgment method are as follows: Get the first voice signal The rhythmic and intonation characteristics of Synchronously processing to obtain rhythm features and intonation features of the second speech signal; Based on the first speech signal The rhythm features and intonation features of the first speech signal are calculated by The second similarity with the second speech signal ; If the second similarity If it is greater than the first threshold, then the first voiceprint feature is considered With the second voiceprint feature Match; otherwise, do not match; When the second similarity When the value is greater than the first threshold, the first speech signal is obtained. The recording time of the first voice signal Generate exception prompts; The second similarity To calculate the rhythm similarity , intonation similarity Weighted calculation of comprehensive similarity; According to the formula: in, Rhythm similarity The weight of is the intonation similarity The weight of ; The second similarity is calculated .

6. The intelligent device interaction system based on voice recognition according to claim 1, characterized in that: The information configuration module also includes a voiceprint feature database and a scene database; the voiceprint feature database is used to store the second voiceprint feature ; The scene database is used to store scene modes corresponding to different users; The scene mode is used to automatically complete the self-response status of multiple smart devices after the authorized user is successfully identified, and the self-response status includes on and off; The response method of the scene mode is: The user outputs the first voice signal ; For the first speech signal Perform voiceprint feature recognition to determine whether the user is an authorized user: If the user is an authorized user, the scene mode is turned on; If the user is an unauthorized user, the scene mode will not be enabled, and an alarm will be generated and displayed synchronously in the scene mode of the authorized user.

7. The intelligent device interaction system based on voice recognition according to claim 1, characterized in that: It also includes an information interaction module, the information interaction module is used to generate analysis results of unauthorized users. The information interaction module completes the data collection function based on the multimedia array, and the multimedia array includes an electronic touch screen; The information interaction module includes a data collection and processing unit; the data collection and processing unit is used to collect the answer records of unauthorized users and generate feedback records based on the answer records: The method for generating feedback records based on answer records is: Obtain operation records generated by unauthorized users operating electronic touch screens; Generate answer records based on unauthorized user names and corresponding operation records; Collect all answer records and build feedback records based on dates; The information interaction module also includes a data analysis unit, which is used to analyze the feedback records and generate analysis results; the analysis results are synchronously displayed in the scene mode of the authorized user.

8. The intelligent device interaction system based on voice recognition according to claim 7, characterized in that: The multimedia array further includes a microphone and a loudspeaker, wherein the microphone is used for receiving sound and the loudspeaker is used for amplifying sound; The information configuration module also includes an audio control unit, which is used to control the connection status of any microphone and the loudspeaker. When any microphone and the loudspeaker are in a connected state, the voice signal collected by any microphone will be amplified. When any microphone and the loudspeaker are not connected, any microphone is in a sound-only state.

9. The intelligent device interaction system based on voice recognition according to claim 8, characterized in that: When any of the microphones and the loudspeaker is not connected, and any of the microphones receives a voice signal of an unauthorized user, a voice prompt is generated based on the voice signal of the unauthorized user, the name of the unauthorized user and the time; The voice prompt is displayed synchronously in the scene mode of the authorized user.

10. A method for intelligent device interaction based on voice recognition, characterized in that: The following steps are involved: Collect the user's first voice signal , and process to obtain the first voiceprint feature ; The first voiceprint feature With the second voiceprint feature Make a comparison; When the first voiceprint feature Second voiceprint feature If the first similarity is greater than the first threshold, a control request is generated and issued; otherwise, an alarm is generated and no control request is generated; Receive control requests and make smart devices perform control actions according to the control requests.

Images