Field recording device for music teaching

Abstract

The invention relates to the field of music recording equipment, and discloses a field recording device for music teaching, which comprises a cylinder body, an upper cover is fixedly mounted at the top end of the cylinder body, an opening is formed in the middle of the upper cover, and a rotating ring is movably mounted in the upper cover. A plurality of opening and closing blades are evenly and movably installed at the positions, close to the opening, in the upper cover, connecting rods are movably installed at the positions, corresponding to the opening and closing blades, of the top of the rotating ring, and the tail ends of the connecting rods are movably installed on one sides of the tops of the opening and closing blades on the corresponding sides; a supporting plate is movably mounted at the inner top of the barrel, and a recorder is fixedly mounted in the middle of the top end of the supporting plate. According to the structure, the unique design enables the recorder to be easy to operate and prevent damage, the recording system is complete in function, all modules from audio input to storage cooperate, signal processing, accurate recognition, sound output and storage management are achieved, the recording quality is guaranteed, and stable operation of the device is ensured.

Description

Technical Field

[0001] The present invention relates to the field of music recording equipment, and particularly to a live recording device for music teaching. Background Art

[0002] In the field of music teaching, effective teaching recording is of great significance for improving teaching quality, improving teaching methods, and evaluating students' learning outcomes. With the continuous development of educational concepts and the increasing diversification of teaching methods, the requirements for live recording in music teaching are getting higher and higher. Traditional recording methods have gradually revealed many limitations, and there is an urgent need for an innovative live recording device to meet the teaching needs.

[0003] Traditional live recording in music teaching mainly relies on simple voice recorders or cameras. Although voice recorders are portable, they have a single function and can only perform basic sound recording, lacking effective processing of sound signals. In a complex music teaching environment, such as a classroom with multiple instruments playing simultaneously and frequent teacher-student interactions, ambient noise is likely to interfere with the recording, resulting in a large amount of noise in the recorded sound, seriously affecting the recording quality and making it difficult to clearly distinguish key teaching content during subsequent teaching review and analysis. Cameras can record video images, but they mainly focus on visual information, have limited sound processing capabilities, and store a large amount of data, which is not conducive to convenient storage and quick retrieval of specific audio content.

[0004] Music teaching has unique characteristics different from other subject teachings. Music teaching emphasizes live performances, singing demonstrations, and real-time interactive communication between teachers and students. These sound information contains rich teaching content, such as details of playing techniques, pitch and rhythm control of singing, and real-time guidance and feedback from teachers. Accurately recording these sound information is crucial for students' after-class review and consolidation, teachers' reflection on the teaching process, and the accumulation and inheritance of teaching resources. However, traditional recording methods are difficult to meet the professional requirements of music teaching sound recording and cannot accurately identify, classify, and process the sound at the music teaching site with high quality.

[0005] In recent years, with the rapid development of technologies such as artificial intelligence and signal processing, it provides technical support for the innovation of live recording devices in music teaching. In terms of signal processing, advanced filtering and noise reduction algorithms have emerged continuously, which can more effectively remove interference noise in sound signals and improve the purity of sound. Machine learning algorithms in artificial intelligence technology make the feature extraction and classification recognition of sound signals more accurate and efficient. The progress of these technologies creates conditions for the research and development of a recording device that can record, accurately process, and intelligently analyze the sound at the music teaching site with high quality. Summary of the Invention

[0006] In view of the deficiencies of the prior art, the present invention provides a live recording device for music teaching, which solves the problems that traditional music teaching recording devices have single functions, poor recording quality, weak sound processing capabilities, inability to accurately identify and classify sound signals, difficulty in meeting the special needs of music teaching, and inability to effectively assist in teaching review, analysis and improvement.

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions: A live recording device for music teaching includes a cylinder body, the top end of the cylinder body is fixedly installed with an upper cover, an opening is provided in the middle of the upper cover, a rotating ring is movably installed inside the upper cover, and a plurality of opening and closing blades are evenly and movably installed at positions near the opening inside the upper cover. At positions corresponding to each opening and closing blade on the top of the rotating ring, connecting rods are movably installed, and the ends of the connecting rods are movably installed on one side of the top of the corresponding opening and closing blade. A support plate is movably installed at the inner top of the cylinder body, and a recorder is fixedly installed in the middle of the top end of the support plate.

[0008] Preferably, it further includes a recording system. The recording system includes an audio input module, a signal processing module, a feature extraction module, a feature processing module, a sound output module, and a storage control module. The audio input module is used to receive the sound at the live music teaching and convert it into a sound signal. The signal processing module filters and denoises the sound signal. The feature extraction module extracts features from the processed sound signal and constructs a joint feature matrix. The feature processing module first performs dimensionality reduction processing on the extracted sound signal, and then trains and tests the extracted sound signal respectively to obtain an optimal classifier model to accurately identify the required sound signal. The sound output module is used to output the sound signal after feature processing, and the storage control module is used to store the processed sound.

[0009] Preferably, the audio input module includes a preamplifier, a DSP module, and a filter and noise reduction module. The preamplifier is used to amplify the received sound signal and increase the anti-interference ability to ensure the recording quality. The output end of the filter and noise reduction module is connected to the data receiving end of the DSP module through an analog-to-digital converter, and the output end of the DSP module is connected to the data receiving end of the filter and noise reduction module through an MCBSP serial port. The filter and noise reduction module includes a through filter, an adder, and a gradient calculation and coefficient correction module.

[0010] Preferably, the through filter is used to receive the sound signal, take the sound signal as the input signal X(n), and perform algorithm processing and compensation on X(n) to obtain the filter output signal y(n) and input it into the adder. At the same time, the desired output signal Y(n) after being inverted is also input into the adder.

[0011] Preferably, the gradient calculation and coefficient correction module is used to receive the difference between Y(n) and y(n) as the error signal e(n), calculate the input signal X(n) and the error signal e(n) each time, obtain the coefficients used for the next round of filtering, continuously update the coefficients of the filter through an adder, use the error between the feedback amount simulated by the filter and the actual feedback amount, and finally output the obtained final output signal L(n);

[0012] Preferably, the feature processing module includes a training module and a testing module, which are used to train and test the sound signal to obtain result data. The feature processing module also includes a classifier model acquisition module and a classification result determination module, which are used to obtain the optimal classifier model during the training and testing processes and confirm the classification results.

[0013] Preferably, a plurality of arc-shaped grooves are formed between the inner wall and the outer wall of the rotating ring. A limiting column is fixedly installed at the bottom wall of the upper cover near each arc-shaped groove, and the outer walls of the limiting columns are movably arranged inside the corresponding arc-shaped grooves.

[0014] Preferably, a straight notch is formed on one side of the upper cover. A knob is fixedly installed on one side of the rotating ring, and the end of the knob passes through the inside of the straight notch and extends to the outside.

[0015] Preferably, a panel is fixedly installed in the middle of the inner bottom of the cylinder body. An inverted heart-shaped groove is formed on the front side of the panel. A hook rod is movably installed in the middle of the bottom end of the support plate, and the end of the hook rod is movably arranged inside the inverted heart-shaped groove.

[0016] Preferably, both sides of the bottom end of the support plate are connected to both sides of the inner bottom of the cylinder body through jacking springs. A torsion spring is fixedly installed at the top of the hook rod. A pressing cover is fixedly installed at the top end of the recorder, and a control panel is fixedly installed at the front end of the cylinder body.

[0017] Preferably, the feature processing module performs dimensionality reduction processing on the sound signal, and realizes the dimensionality reduction processing of the joint feature matrix by multiplying the constructed joint feature matrix by a random matrix, including the following steps:

[0018] (1) Denote the feature vector matrix describing the sound signal as D∈I Rd×p, where d is the length of the feature vector of the key point module and p is the number of key point modules;

[0019] (2) Project the vector with dimension d into a subspace with dimension n, where n << d;

[0020] (3) Realize the dimensionality reduction processing of the feature vector matrix by multiplying the matrix D by a random matrix R, and reduce the dimensionality of the feature vector matrix to n×p:

[0021] Y = RD

[0022] Where Y ∈ IRn×p is the feature vector matrix after dimensionality reduction, R is the random projection matrix, R ∈ IRn×d, with a mean of 0 and a variance of 1.

[0023] Preferably, the method for the feature processing module to obtain the optimal classifier model through training and testing is specifically to divide the processed sound signal into a training group and a testing group. The training group performs format conversion, feature point extraction, feature vector calculation, dimensionality reduction, and category histogram calculation on the sound signal as training data to obtain the optimal parameters of the classifier model for the training data. The testing group performs format conversion, feature point extraction, feature vector calculation, dimensionality reduction, and category histogram calculation on the sound signal as training data, and then performs sparse representation using the optimal parameters of the classifier model in the training group to obtain the optimal classifier model. Subsequently, the data of the testing group is input into the trained classifier model for classification to complete the recognition and classification of the sound signal.

[0024] Preferably, the storage control module includes a volume control unit, a storage unit, a PC, a JTAG interface, and a power module.

[0025] The present invention provides a on-site recording device for music teaching. It has the following beneficial effects:

[0026] 1. Through the unique design of the barrel and upper cover structure of the device of the present invention, the opening and closing blades can be controlled to open and close through the rotating ring and connecting rod. When opened, the recorder can be exposed, and when not in use, the opening can be sealed to effectively prevent dust from falling and damage caused by dropping. At the same time, the cooperation of the support plate with the hook rod and the inverted heart-shaped groove, as well as the setting of the jacking spring, enables the recorder to be conveniently ejected and retracted, facilitating the recording operation.

[0027] 2. The recording system of the present invention has comprehensive functions to ensure the recording quality. The preamplifier enhances the sound signal and improves the anti-interference ability. The filter and noise reduction module performs preliminary filtering and noise reduction processing on the sound signal through the coordinated work of the through filter, adder, and gradient calculation and coefficient correction module to ensure the recording quality.

[0028] 3. The present invention performs secondary filtering and noise reduction on the sound signal processed by the audio input module to further remove noise, making the sound signal purer. The feature extraction module constructs a joint feature matrix to integrate the key features of the sound signal. The feature processing module first performs dimensionality reduction processing on the sound signal to reduce the data volume, and then obtains the optimal classifier model through the operations of the training group and the testing group to achieve accurate recognition and classification of the sound signal, which helps to improve the pertinence and effectiveness of recording. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a three-dimensional view of the present invention;

[0030] Figure 2 It is a schematic internal structure diagram of the upper cover in the present invention;

[0031] Figure 3 It is a schematic internal structure diagram of the cylinder body in the present invention;

[0032] Figure 4 It is a system architecture diagram of the present invention;

[0033] Figure 5 It is a system structure diagram of the storage control module in the present invention;

[0034] Figure 6 It is a system structure diagram of the feature processing module in the present invention;

[0035] Figure 7 It is a system structure diagram of the filtering and noise reduction module in the present invention.

[0036] Among them, 1. Cylinder body; 2. Upper cover; 3. Opening; 4. Rotating ring; 5. Arc-shaped groove; 6. Limit post; 7. Opening and closing blade; 8. Connecting rod; 9. Straight slot; 10. Dial button; 11. Support plate; 12. Recorder; 13. Lifting spring; 14. Panel; 15. Inverted heart-shaped groove; 16. Hook rod; 17. Torsion spring; 18. Pressure cover; 19. Control panel. Specific embodiments

[0037] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0038] Embodiment:

[0039] Please refer to the attached Figure 1 - Attached Figure 7, an embodiment of the present invention provides a on-site recording device for music teaching, including a cylinder 1. The cylinder 1 is cylindrical as a whole, and its material is selected as a strong and durable high-strength engineering plastic. This material is not only light in texture, convenient to carry and move, but also has good impact resistance, which can effectively resist minor collisions during daily use and ensure the safety of the internal components of the device. The top of the cylinder 1 is fixedly installed with an upper cover 2 through tight screws. The upper cover 2 is also made of the same material as the cylinder 1, and its surface is finely sandblasted, which not only increases the friction force, facilitates the operation of the user, but also improves the overall aesthetics. An opening 3 is precisely opened in the middle of the upper cover 2. The shape of the opening 3 is circular, and its size is just large enough to accommodate the top of the recorder 12 to protrude, ensuring that the recorder 12 can fully receive external sound signals during operation. A rotating ring 4 is movably installed inside the upper cover 2. A number of arc-shaped grooves 5 are evenly opened between the inner wall and the outer wall of the rotating ring 4. These arc-shaped grooves 5 are concentrically distributed, and the radian is carefully designed to perfectly cooperate with the limit posts 6 to achieve precise limit rotation. Limit posts 6 are firmly fixedly installed at positions close to each arc-shaped groove 5 on the bottom wall of the upper cover 2. The outer walls of the limit posts 6 are all movably arranged inside the corresponding arc-shaped grooves 5 on the corresponding side. Under the combined action of the arc-shaped grooves 5 and the limit posts 6, the rotating ring 4 can only rotate smoothly in a circle inside the upper cover 2, without shaking or detaching from the upper cover 2.

[0040] A number of opening and closing blades 7 are evenly and movably installed inside the upper cover 2 near the opening 3. These opening and closing blades 7 are radially distributed with the opening 3 as the center, and one end of each opening and closing blade 7 is movably connected to the inside of the upper cover 2 through a precise rotating shaft, so that the opening and closing blades 7 can rotate flexibly around the rotating shaft. Connecting rods 8 are movably installed at positions corresponding to each opening and closing blade 7 on the top of the rotating ring 4. Both ends of the connecting rods 8 are respectively connected to the rotating ring 4 and the opening and closing blades 7 through movable pins. This connection method can not only ensure the flexibility of the connecting rods 8 during rotation, but also ensure their stability when transmitting power. The ends of the connecting rods 8 are all movably installed on one side of the top of the corresponding opening and closing blades 7. When the rotating ring 4 rotates, it will drive one end of all the connecting rods 8 to move accordingly, so that the other end of the connecting rods 8 pushes all the opening and closing blades 7 to rotate around the shaft on which they are movably installed, making an outward opening movement, so that the opening 3 in the center of the upper cover 2 is opened, exposing the internal recorder 12; when the rotating ring 4 is rotated in the reverse direction, the opening and closing blades 7 will rotate in the reverse direction and close the opening 3 again.

[0041] A straight slot 9 is provided on one side of the upper cover 2. The length and width of the straight slot 9 are precisely calculated to ensure that the dial button 10 can move smoothly therein without generating excessive gaps to cause shaking. A dial button 10 is fixedly installed on one side of the rotating ring 4. The shape of the dial button 10 conforms to the principles of ergonomics, which is convenient for the user to pinch and turn it with his fingers. The end thereof passes through the interior of the straight slot 9 and extends to the outside. The user can easily control the rotation of the rotating ring 4 by turning the dial button 10, thereby realizing the opening and closing operation of the opening and closing blades 7.

[0042] A support plate 11 is movably installed on the inner top of the cylinder 1. The support plate 11 is made of lightweight aluminum alloy, which is not only high in strength but also light in weight. It can reduce the overall weight of the device while ensuring the support stability. A recorder 12 is firmly fixed to the middle of the top of the support plate 11 by strong glue. The recorder 12 uses highly sensitive professional recording equipment and can clearly and accurately capture various sound signals at the music teaching scene.

[0043] In the present embodiment, a recording system is also included. The recording system includes an audio input module, a signal processing module, a feature extraction module, a feature processing module, a sound output module and a storage control module. The audio input module is used to receive the sound of the music teaching site and convert it into a sound signal. The audio input module includes a preamplifier, a DSP module and a filter noise reduction module. The preamplifier is used to amplify the received sound signal and increase the anti-interference ability to ensure the quality of the recording. It can effectively amplify the weak sound signal to a suitable amplitude. At the same time, through the internal anti-interference circuit, it can reduce the impact of external electromagnetic interference on the sound signal. The output end of the filter noise reduction module is connected to the data receiving end of the DSP module through a digital-to-analog converter. The output end of the DSP module is connected to the data receiving end of the filter noise reduction module through an MCBSP serial port. The filter noise reduction module includes a transparent filter, an adder and a gradient calculation and coefficient correction module. The transparent filter is used The sound signal is received and used as the input signal X(n). The filter output signal y(n) is obtained after algorithm processing and compensation of X(n) and input into the adder. At the same time, the expected output signal Y(n) is also input into the adder after being inverted. The transparent filter can accurately process and compensate according to the characteristics of the sound signal, effectively improving the quality of the sound signal. The gradient calculation and coefficient correction module is used to receive the difference between Y(n) and y(n) as the error signal e(n), and calculate the input signal X(n) and the error signal e(n) each time to obtain the coefficient used for the next round of filtering. The filter coefficient is continuously updated through the adder, and the error between the feedback amount obtained by the filter simulation and the actual feedback amount is calculated, and finally the final output signal L(n) is output. The gradient calculation and coefficient correction module realizes dynamic adjustment of the filter coefficient through complex mathematical operations and logical control to ensure the optimization of the filtering effect.

[0044] The signal processing module filters and denoises the sound signal, and can perform secondary in-depth processing on the sound signal processed by the audio input module to further remove noise, making the sound signal purer and providing a high-quality data basis for subsequent feature extraction and analysis.

[0045] The feature extraction module extracts features from the processed sound signal and constructs a joint feature matrix. It can accurately extract key features from complex sound signals and integrate these features into a joint feature matrix for subsequent in-depth analysis by the feature processing module.

[0046] The feature processing module first performs dimensionality reduction on the extracted sound signal, and then trains and tests the extracted sound signal respectively to obtain an optimal classifier model to accurately identify the required sound signal. The feature processing module includes a training module and a testing module for training and testing the sound signal to obtain result data. The feature processing module also includes a classifier model acquisition module and a classification result determination module for obtaining the optimal classifier model during the training and testing processes and for confirming the classification result. The feature processing module performs dimensionality reduction on the sound signal by multiplying the constructed joint feature matrix by a random matrix. The dimensionality reduction process of the joint feature matrix includes the following steps:

[0047] (1) Denote the feature vector matrix describing the sound signal as D ∈ I Rd×p, where d is the length of the feature vector of the key point module, and p is the number of key point modules. These key point modules are determined according to the characteristics and analysis requirements of the music teaching sound signal. Each key point module contains important information about the sound signal in a specific aspect.

[0048] (2) Project the vector with dimension d into a subspace with dimension n, where n << d. Through this dimensionality reduction operation, while retaining the key information of the sound signal, the data volume can be reduced and the efficiency of subsequent processing can be improved.

[0049] (3) Realize the dimensionality reduction processing of the feature vector matrix by multiplying the matrix D by a random matrix R, and reduce the dimensionality of the feature vector matrix to n × p:

[0050] Y = RD

[0051] where Y ∈ I Rn×p is the dimensionality-reduced feature vector matrix, and R is the random projection matrix, R ∈ I Rn×d, with a mean of 0 and a variance of 1. This dimensionality reduction method based on a random matrix has good stability and generality while ensuring the dimensionality reduction effect.

[0052] The method for the feature processing module to obtain the optimal classifier model through training and testing is specifically as follows: The processed voice signals are divided into a training group and a testing group. For the training group, after performing format conversion, feature point extraction, feature vector calculation, dimensionality reduction, and category histogram calculation on the voice signals as training data, the optimal parameters of the classifier model for the training data are obtained. For the testing group, after performing format conversion, feature point extraction, feature vector calculation, dimensionality reduction, and category histogram calculation on the voice signals as training data, sparse representation is performed using the optimal parameters of the classifier model in the training group to obtain the optimal classifier model. Subsequently, the data in the testing group is input into the trained classifier model for classification, completing the recognition and classification of the voice signals. In this process, through a large amount of training data and a strict testing process, the classifier model is continuously optimized to ensure that it can accurately identify various voice signals at the music teaching site.

[0053] The voice output module is used to output the processed voice signals, which can be used for real-time monitoring or subsequent playback. It can output the processed voice signals in a clear and stable manner, facilitating the user to monitor and playback.

[0054] The storage control module is used to store the processed voice. The storage control module includes a volume control unit, a storage unit, a PC, a JTAG interface, and a power supply module. The volume control unit can adjust the volume-related parameters during storage. By adjusting the volume control unit, the user can adjust the volume of the stored voice according to actual needs to ensure that the stored voice has an appropriate volume during playback. The storage unit is used to store voice data. The storage unit has characteristics such as large capacity and high-speed reading and writing, and can stably store a large amount of music teaching recording data. The PC can perform operations such as managing and analyzing the stored data. Through the specialized software installed on the PC, the user can conveniently classify, retrieve, and edit the stored recording data, and can also conduct in-depth analysis of the recording data, such as audio spectrum analysis and rhythm analysis, providing strong support for music teaching research. The JTAG interface facilitates the debugging and program update of the device. Through the JTAG interface, technicians can conveniently debug the hardware and software of the device, timely discover and solve potential problems, and can also conveniently update the program of the device to improve the performance and functions of the device. The power supply module supplies power to the entire storage control module and other parts of the device. The power supply module can provide reliable power support for each module of the device to ensure that the device can work normally in various environments.

[0055] A panel 14 is fixedly installed on the middle part of the inner bottom of the cylinder 1 by screws. The panel 14 is made of high-strength insulating material and its surface is smoothed, which is not only beautiful but also easy to clean. An inverted heart-shaped groove 15 is opened on the front side of the panel 14. The inverted heart-shaped groove 15 consists of two upper and lower hook grooves and arc grooves on both sides. Its shape and size are carefully designed to perfectly cooperate with the hook rod 16 to realize the pop-up and fixation functions of the recorder 12. A hook rod 16 is movably installed in the middle part of the bottom end of the support plate 11, and the end of the hook rod 16 is movably set inside the inverted heart-shaped groove 15. The hook rod 16 is connected to the support plate 11 through a precise rotating shaft, so that the hook rod 16 can move flexibly in the inverted heart-shaped groove 15.

[0056] Both sides of the bottom end of the support plate 11 are connected to both sides of the inner bottom of the cylinder 1 through a lifting spring 13. The lifting spring 13 is made of high-strength spring steel. Its elastic coefficient has been accurately calculated and debugged, and it can quickly return to its original state after being compressed by external force, providing a stable lifting force for the support plate 11. A torsion spring 17 is fixedly installed on the top of the hook rod 16. The function of the torsion spring 17 is to provide a certain torsional force during the movement of the hook rod 16 to ensure that the hook rod 16 can accurately fall into the corresponding position of the inverted heart-shaped groove 15 to achieve reliable fixation of the recorder 12. A pressure cover 18 is fixedly installed on the top of the recorder 12. The material of the pressure cover 18 is the same as that of the upper cover 2, and its surface is also frosted to increase friction and facilitate the user to press and operate. A control panel 19 is fixedly installed at the front end of the cylinder 1. Various operation buttons and indicator lights are integrated on the control panel 19, such as a power switch button, a recording start / stop button, a volume adjustment button, a working status indicator light, etc. The user can conveniently perform various operations on the device and monitor the working status of the device through the control panel 19.

[0057] During recording, the sound of on-site music teaching is received by the audio input module. The preamplifier first amplifies the sound signal and simultaneously enhances its anti-interference ability to ensure the recording quality. The filter and noise reduction module performs preliminary processing on the sound signal. The transparent filter receives the sound signal X(n), and after algorithm processing and compensation, y(n) is input to the adder. At the same time, the inverted desired output signal Y(n) is also input to the adder. The gradient calculation and coefficient correction module receives the difference between Y(n) and y(n) as the error signal e(n), calculates the next-round filter coefficient according to the input signal X(n) and e(n) each time, updates the filter coefficient through the adder, and finally outputs the signal L(n). L(n) is transmitted to the DSP module through the digital-to-analog converter. The DSP module interacts with the filter and noise reduction module through the MCBSP serial port. The signal processing module receives the sound signal processed by the audio input module and performs secondary processing of filtering and noise reduction to further remove noise and make the sound signal purer. The feature extraction module conducts feature extraction work on the sound signal processed by the signal processing module. Based on the characteristics of the sound signal, a joint feature matrix is constructed to integrate the key features of the sound signal for subsequent analysis. The feature processing module first performs dimensionality reduction processing on the extracted sound signal. Denote the feature vector matrix describing the sound signal as D ∈ IRd×p. Project the vector with dimension d onto a subspace with dimension n (n << d). Dimensionality reduction is achieved by multiplying the matrix D by a random matrix R (R ∈ IRn×d, with a mean of 0 and a variance of 1) to obtain the dimensionality-reduced feature vector matrix Y ∈ IRn×p. Then, the processed sound signal is divided into a training group and a test group. The training group performs format conversion, feature point extraction, feature vector calculation, dimensionality reduction, and category histogram calculation on the sound signal to obtain the optimal parameters of the classifier model for the training data. After the test group performs the same operations, sparse representation is carried out using the optimal parameters obtained by the training group to obtain the optimal classifier model. Finally, the test group data is input into the trained classifier model for classification to complete the recognition and classification of the sound signal. The sound output module outputs the sound signal after feature processing, which can be used for real-time monitoring or subsequent playback. The storage control module stores the processed sound. The volume control unit can adjust the volume-related parameters during storage. The storage unit is used to store the sound data. The PC can manage, analyze, and other operations on the stored data. The JTAG interface facilitates the debugging and program update of the device. The power supply module supplies power to the entire storage control module and other parts of the device.

[0058] Working principle:

[0059] When in use, the device is placed in an unobstructed position at the scene of music teaching. First, pinch the dial button 10 with your fingers and gently turn it. Under the limiting action of the arc groove 5 and the limiting column 6, the rotating ring 4 will be driven to rotate smoothly inside the upper cover 2. When the rotating ring 4 rotates, it will drive one end of all the connecting rods 8 to move with it, so that the other end of the connecting rod 8 pushes all the opening and closing blades 7 to rotate around the axis of their movable installation, making an outward opening action, so that the opening 3 in the center of the upper cover 2 is opened, exposing the internal recorder 12, and then press down the pressure cover 18 on the top of the recorder 12 with your fingers. The pressure cover 18 drives the recorder 12 and the support plate 11 to descend, thereby driving the hook rod 16 to follow The inverted heart-shaped groove 15 is composed of two upper and lower hook grooves and arc grooves on both sides. When the hook rod 16 is pressed down, it will be disengaged from the hook groove at the bottom of the inverted heart-shaped groove 15. After releasing the finger, the lifting spring 13 will quickly push up the support plate 11, and at the same time, the hook rod 16 will move along the arc groove on one side of the inverted heart-shaped groove 15 to the hook groove at the top. At this time, the recorder 12 is pushed out of the opening 3 by the pushed-up support plate 11, which is convenient for subsequent recording work. When not in use, only the pressure cover 18 needs to be pressed down to press the recorder 12 back into the cylinder body 1, and the dial button 10 is turned in the opposite direction to make all the opening and closing blades 7 re-seal the opening 3, which can prevent dust and falling damage when not in use.

[0060] During recording, the sound of on-site music teaching is received by the audio input module. The preamplifier first amplifies the sound signal and simultaneously enhances its anti-interference ability to ensure the recording quality. The filter and noise reduction module conducts preliminary processing on the sound signal. The transparent filter receives the sound signal X(n), and through algorithm processing and compensation, y(n) is obtained and input into the adder. At the same time, the inverted desired output signal Y(n) is also input into the adder. The gradient calculation and coefficient correction module receives the difference between Y(n) and y(n) as the error signal e(n), calculates the filter coefficient for the next round based on the input signal X(n) and e(n) each time, and continuously updates the filter coefficient through the adder. Finally, the output signal L(n) is transmitted to the DSP module through the digital-to-analog converter. The DSP module interacts with the filter and noise reduction module through the MCBSP serial port. The signal processing module receives the sound signal processed by the audio input module and conducts secondary processing of filtering and noise reduction to further remove background noise and make the sound signal purer. The feature extraction module conducts feature extraction on the sound signal processed by the signal processing module. Based on the characteristics of the sound signal, a joint feature matrix is constructed to integrate the key features of the sound signal for subsequent analysis. The feature processing module first conducts dimensionality reduction processing on the extracted sound signal. Denote the feature vector matrix describing the sound signal as D ∈ IRd×p. Project the vector with dimension d onto the subspace with dimension n (n << d). Dimensionality reduction is achieved by multiplying the matrix D by the random matrix R (R ∈ IRn×d, with a mean of 0 and a variance of 1), obtaining the dimensionality-reduced feature vector matrix Y ∈ IRn×p. Then, the processed sound signal is divided into a training group and a test group. The training group conducts format conversion, feature point extraction, feature vector calculation, dimensionality reduction, and class histogram calculation on the sound signal to obtain the optimal parameters of the classifier model for the training data. After the test group conducts the same operations, sparse representation is performed using the optimal parameters obtained from the training group to obtain the optimal classifier model. Finally, the test group data is input into the trained classifier model for classification to complete the recognition and classification of the sound signal. The sound output module outputs the sound signal after feature processing, which can be used for real-time monitoring or subsequent playback. The storage control module stores the processed sound. The volume control unit can adjust the volume-related parameters during storage. The storage unit is used to store the sound data. The PC can manage, analyze, and perform other operations on the stored data. The JTAG interface facilitates the debugging and program update of the device. The power supply module supplies power to the entire storage control module and other parts of the device.

[0061] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A field recording device for music teaching, comprising a cylinder (1), characterized in that: An upper cover (2) is fixedly mounted on the top of the cylinder (1), an opening (3) is provided in the middle of the upper cover (2), a rotating ring (4) is movably mounted inside the upper cover (2), a plurality of opening and closing blades (7) are evenly and movably mounted at a position near the opening (3) inside the upper cover (2), a connecting rod (8) is movably mounted at a position corresponding to each opening and closing blade (7) on the top of the rotating ring (4), the ends of the connecting rods (8) are movably mounted on one side of the top of the opening and closing blade (7) on the corresponding side, a supporting plate (11) is movably mounted on the inner top of the cylinder (1), and a recorder (12) is fixedly mounted at the middle of the top of the supporting plate (11); It also includes a recording system, which includes an audio input module, a signal processing module, a feature extraction module, a feature processing module, a sound output module and a storage control module. The audio input module is used to receive the sound of the music teaching site and convert it into a sound signal. The signal processing module filters and reduces noise on the sound signal. The feature extraction module extracts features from the processed sound signal and constructs a joint feature matrix. The feature processing module first performs dimensionality reduction on the extracted sound signal, and then trains and tests the extracted sound signal respectively to obtain an optimal classifier model to accurately identify the required sound signal. The sound output module is used to output the sound signal after feature processing, and the storage control module is used to store the processed sound. The audio input module includes a preamplifier, a DSP module and a filter noise reduction module. The preamplifier is used to amplify the received sound signal and increase the anti-interference ability to ensure the quality of the recording. The output end of the filter noise reduction module is connected to the data receiving end of the DSP module through a digital-to-analog converter. The output end of the DSP module is connected to the data receiving end of the filter noise reduction module through an MCBSP serial port. The filter noise reduction module includes a transparent filter, an adder and a gradient calculation and coefficient correction module. The transparent filter is used to receive a sound signal, take the sound signal as an input signal X(n), perform algorithm processing and compensation on X(n) to obtain a filter output signal y(n) which is input into an adder, and at the same time, the expected output signal Y(n) is also input into the adder after being inverted; The gradient calculation and coefficient correction module is used to receive the difference between Y(n) and y(n) as the error signal e(n), and calculate the input signal X(n) and the error signal e(n) each time to obtain the coefficient used for the next round of filtering, and continuously update the coefficient of the filter through the adder, using the error between the feedback amount obtained by the filter simulation and the actual feedback amount, and finally output the final output signal L(n); The feature processing module includes a training module and a testing module, which are used to train and test the sound signal and obtain result data. The feature processing module also includes a classifier model acquisition module and a classification result determination module, which are used to obtain the optimal classifier model in the training and testing process and confirm the classification result.

2. A field recording device for music teaching according to claim 1, characterized in that: A plurality of arc-shaped grooves (5) are provided between the inner wall and the outer wall of the rotating ring (4); a limiting column (6) is fixedly installed on the bottom wall of the upper cover (2) near each arc-shaped groove (5); and the outer wall of the limiting column (6) is movably arranged inside the arc-shaped groove (5) on the corresponding side.

3. A field recording device for music teaching according to claim 1, characterized in that: A straight slot (9) is provided on one side of the upper cover (2), and a dial button (10) is fixedly mounted on one side of the rotating ring (4), with the end of the dial button (10) passing through the interior of the straight slot (9) and extending to the outside.

4. A field recording device for music teaching according to claim 1, characterized in that: A panel (14) is fixedly mounted in the middle of the inner bottom of the cylinder (1), an inverted heart-shaped groove (15) is provided on the front side of the panel (14), a hook rod (16) is movably mounted in the middle of the bottom end of the support plate (11), and the end of the hook rod (16) is movably arranged inside the inverted heart-shaped groove (15).

5. The on-site recording device for music teaching according to claim 1, characterized in that: Both sides of the bottom end of the support plate (11) are connected to both sides of the inner bottom of the cylinder (1) through lifting springs (13), a torsion spring (17) is fixedly installed on the top of the hook rod (16), a pressure cover (18) is fixedly installed on the top of the recorder (12), and a control panel (19) is fixedly installed on the front end of the cylinder (1).

6. The on-site recording device for music teaching according to claim 1, characterized in that: The feature processing module performs dimensionality reduction processing on the sound signal, and implements dimensionality reduction processing of the joint feature matrix by multiplying the constructed joint feature matrix by a random matrix, including the following steps: (1) The feature vector matrix describing the sound signal is denoted as D∈IRd×p, where d is the feature vector length of the key point module and p is the number of key point modules; (2) Project the vector of dimension d into a subspace of dimension n, where n < < d; (3) The eigenvector matrix is ​​reduced in dimension by multiplying the matrix D by a random matrix R, reducing the dimension of the eigenvector matrix to n×p: Y=RD Among them, Y∈IRn×p is the eigenvector matrix after dimensionality reduction, R is the random projection matrix, R∈IRn×d, with a mean of 0 and a variance of 1.

7. The on-site recording device for music teaching according to claim 1, characterized in that: The method for the feature processing module to obtain the optimal classifier model through training and testing is specifically as follows: the processed sound signal is divided into a training group and a test group, the training group uses the sound signal as training data to perform format conversion, feature point extraction, feature vector calculation, dimensionality reduction and category histogram calculation, and then obtains the optimal parameters of the classifier model of the training data; the test group uses the sound signal as training data to perform format conversion, feature point extraction, feature vector calculation, dimensionality reduction and category histogram calculation, and then uses the optimal parameters of the classifier model in the training group for sparse expression to obtain the optimal classifier model, and then inputs the test group data into the trained classifier model for classification, thereby completing the recognition and classification of the sound signal.

8. The on-site recording device for music teaching according to claim 1, characterized in that: The storage control module comprises a volume control unit, a storage unit, a PC, a JTAG interface and a power supply module.

Images