A rice bucket intelligent voice control system and method

Abstract

The invention discloses a rice bucket intelligent voice control system and method, which belongs to the technical field of intelligent voice control systems and is used to solve the problem of interference of environmental and positional factors on voice recognition accuracy; the method comprises initial setting, standby conversion, monitoring and analysis, voice quality detection and voice compensation modules; the initial setting module controls the calibration of detection sensors; the standby conversion module realizes the intelligent conversion of the rice bucket standby and working states; the monitoring and analysis module locates the user coordinates, collects the temperature and humidity and comprehensively analyzes to obtain the ring position influence value, and generates a voice signal detection instruction when the threshold is exceeded; the voice quality detection module processes the voice signal to obtain the speech evaluation value, and generates the corresponding instruction according to the preset interval matching; the voice compensation module performs amplification operation on the voice signal to be compensated and then detects it again; the interference of environmental and positional factors on the voice recognition accuracy is effectively reduced, the reliability and stability of the system are improved, the convenience of use is improved and the energy consumption is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent voice control systems, and in particular relates to an intelligent voice control system and method for a rice bucket. Background Art

[0002] With the continuous development of smart home technology, the intelligent voice control system of rice bucket has gradually become popular, providing users with a more convenient operation experience; however, in actual use, the environment in which rice bucket is located is complex and changeable, which brings great challenges to voice recognition;

[0003] The kitchen environment where the rice bucket is located has many problems. From the perspective of temperature factors, during cooking, stoves, ovens and other equipment run for a long time, the kitchen temperature rises sharply, and heat accumulates in a limited space, causing the air density to change. While the speed of sound propagation increases, energy loss increases, voice signal transmission is blocked, and the sound propagation characteristics are interfered, which in turn affects the accuracy of voice recognition.

[0004] Humidity is also a problem. When cooking, a large amount of water vapor evaporates, and the humidity in the kitchen is instantly saturated. The air viscosity increases, and the sound absorption capacity becomes stronger. The voice signal is like being trapped in a "quagmire", and the intensity decays rapidly. After the meal, the humidity fluctuates with the ventilation conditions, and the sound propagation path continues to change, which seriously affects the voice clarity.

[0005] The relative position of the user to the rice bucket is not fixed. The distance from the rice bucket and the angle vary. Due to the directionality of sound propagation and the reflection and absorption characteristics of the rice bucket itself, the change of the user's position will cause significant changes in the transmission path and attenuation of the voice signal, making it difficult to ensure a stable recognition accuracy rate.

[0006] Furthermore, in terms of energy saving, the traditional rice bucket's switching mechanism between working and non-working states is crude and cannot be intelligently adjusted according to real-time kitchen activities. It is common for the rice bucket to be in standby mode for a long time when no one is using it, or to be frequently started up when busy, consuming energy. For this reason, we propose an intelligent voice control system and method for a rice bucket. Summary of the invention

[0007] The purpose of the present invention is to provide a rice bucket intelligent voice control system and method to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions: a rice bucket intelligent voice control system and method, comprising: an initial setting module, a standby conversion module, a monitoring and analysis module, a voice quality detection module, and a voice compensation module;

[0009] Initial setting module: The rice bucket is equipped with a lidar sensor, an image sensor and a detection sensor; the detection sensors include: position, humidity and temperature sensors; when the rice bucket is restarted or receives a detection sensor calibration instruction, the detection sensor is controlled to calibrate;

[0010] Standby conversion module: Use LiDAR to scan the kitchen to obtain the scope of the kitchen area, analyze the user's kitchen area, the number of home appliances, and the activity area in the kitchen to obtain the detection diameter, draw a detection circle based on the detection diameter, and define the detection area based on the kitchen area and the detection circle. According to the activity duration of the user in the detection area, control the rice bucket to switch between standby and working states;

[0011] Monitoring and analysis module: When the rice bucket is working, the position sensor locates the coordinates of the user relative to the rice bucket and calculates the offset value. The temperature and humidity sensors collect the ambient temperature and humidity and analyze them to obtain the temperature shadow value and the humidity shadow value. The user issues a voice command, and the ring position impact value is obtained by comprehensive analysis based on the temperature shadow value, the humidity shadow value and the offset value. If the ring position impact value exceeds the preset threshold, a voice signal detection command is generated and sent to the voice quality detection module. The position sensor, temperature sensor and humidity sensor are regularly calibrated and tested.

[0012] Voice quality detection module: when receiving voice signal detection instructions, it discretizes the voice signal, calculates the speech amplitude value and the optimal frequency ratio value, and comprehensively analyzes and obtains the speech evaluation value, generates voice recognition, compensation or unrecognizable instructions according to the preset interval matching, and sends the compensation instructions to the voice compensation module;

[0013] Voice compensation module: after receiving the voice compensation instruction, it amplifies and compensates the voice signal.

[0014] Preferably, the specific process of controlling the rice bucket to switch between the standby state and the working state is:

[0015] A laser radar sensor is used to perform an all-round scan of the kitchen environment to obtain the area range of the kitchen. A circle is drawn within the kitchen area with the rice bucket as the center and the detection diameter as the radius to obtain a detection circle. The area covered by the detection circle and the overlapping area of ​​the kitchen area are recorded as the detection area. The detection area is detected by an image sensor, and a power-on time is preset. When the image sensor detects that a user appears in the detection area and the stay time is greater than the preset power-on time, the rice bucket is controlled to enter the working state from the standby state; when the image sensor cannot detect the user in the detection area and the duration is greater than the preset power-on time, the rice bucket is controlled to enter the standby state from the working state.

[0016] Preferably, the detection diameter is obtained in the following manner:

[0017] The kitchen area is scanned by a lidar sensor to obtain the user's kitchen area value MZ, the number of home appliances in the kitchen ZS, and the user's activity area value YH in the kitchen. After normalization, they are substituted into the formula: ZP = MZ × a1 + ZS × a2 + YH × a3 to obtain the detection assignment ZP, where a1, a2, and a3 are preset weight coefficients. Home appliances include: refrigerators, microwave ovens, and ovens. Multiple detection assignment intervals are preset, and each interval corresponds to a detection diameter. The detection assignment is matched with multiple detection assignment intervals, and the detection diameter corresponding to the successfully matched detection assignment interval is output as the detection diameter of the detection circle drawn by the user.

[0018] Preferably, the specific process of the position sensor locating the coordinates of the user relative to the rice bucket and calculating the offset value is:

[0019] When the rice bucket enters the working state from the standby state, multiple position sensors distributed in different positions of the rice bucket detect the signals sent by the users entering the detection area; for each position sensor, the time difference between the position sensor sending the signal to the user and the user reflecting the signal at the acquisition time is obtained, which is recorded as the detection time difference. According to the speed of the signal sent by the position sensor in the air and the detection time difference, the distance between the user and the position sensor can be calculated and marked as the detection distance. According to the detection distance between the user and each sensor, the coordinates of the user relative to each sensor with itself as the origin of the coordinate system can be calculated by using the triangulation method (XUi, YUi), where i is the number of the position sensor, i=1, 2..., n; n is the total number of position sensors;

[0020] Establish a two-dimensional rectangular coordinate system with the center of the rice bucket as the origin. Set the coordinates of the center of the rice bucket to (X0, Y0). Use the formula: The coordinates (XU, YU) of the user relative to the rice bucket are obtained and recorded as the detection coordinates, where Wi is the preset weight coefficient assigned to each position sensor;

[0021] When the user issues a voice command, the circular array of microphones on the rice bucket can receive the voice command. When the voice command is received, the coordinates of the user who issued the voice command are immediately detected through the position sensor, and the coordinates of the user relative to the rice bucket when the command is issued are obtained, and recorded as the detection coordinates. The user's detection coordinates are connected with the center of the rice bucket to obtain the detection line; the direction of the user's routine operation in the rice bucket set by default when leaving the factory is recorded as the positive direction of the rice bucket, and the angle between the detection line and the positive direction of the rice bucket is obtained, recorded as the detection angle ZJ, and the distance formula between the two points is used to calculate the distance value between the user's detection coordinates and the center of the rice bucket, which is recorded as the detection distance value ZG.

[0022] Preferably, the specific process of analyzing the ring position impact value and generating a voice signal detection instruction to send to the voice quality detection module is:

[0023] The temperature and humidity values ​​of the surrounding environment are collected by the temperature sensor and the humidity sensor, and several temperature intervals and humidity intervals are set. Each temperature interval corresponds to a preset temperature influence coefficient of sound propagation, and each humidity interval corresponds to a preset humidity influence coefficient of sound propagation. The temperature value and humidity value at the time of collection are matched with the corresponding temperature interval and humidity interval, and the corresponding influence coefficient is output, and multiplied by the corresponding influence coefficient to obtain the temperature shadow value WP and the humidity shadow value SP corresponding to the time of collection;

[0024] The temperature shadow value WP, wetness shadow value SP, detection angle ZJ and detection distance value ZG at the moment of receiving the voice command are obtained, and after normalization, they are substituted into the preset formula: HWY=ZJ×b1+ZG×b2+WP×b3+SP×b4 to obtain the ring position influence value HWY, wherein b1, b2, b3, and b4 are preset weight factors and a preset ring position influence threshold. The ring position influence value is compared with the preset ring position influence threshold. If the ring position influence value is greater than the preset ring position influence threshold, a voice signal detection instruction is generated and sent to the voice quality detection module. At the same time, the preset ring position influence threshold is subtracted from the ring position influence value to obtain the ring position deviation value, and the ring position deviation value is sent to the voice compensation module.

[0025] Preferably, the process of regularly calibrating and testing the position sensor, temperature sensor and humidity sensor is as follows:

[0026] Within the detection area, a fixed piece of furniture is randomly selected as the detection object, and the coordinates of the object detection relative to the rice bucket are obtained when the rice bucket is restarted for three consecutive times to detect the detection object. The horizontal coordinates and the vertical coordinates of the coordinates of the detection object relative to the rice bucket detected three times are averaged respectively to obtain the horizontal coordinate mean and the vertical coordinate mean of the detection object, and the coordinates composed of the horizontal coordinate mean and the vertical coordinate mean of the detection object are marked as calibration coordinates;

[0027] Set the detection sensor update interval and preset the detection time. When the detection sensor is in the update moment, obtain the coordinates of the detection object at each collection moment relative to the rice bucket within the preset detection time, perform Euclidean distance calculation on the coordinates of the detection object at each collection moment relative to the rice bucket and the calibration coordinates of the detection object, and calculate the average value thereof, obtain the average value of the distance between the coordinates of the detection object image relative to the rice bucket and the calibration coordinates at each collection moment, record it as the calibration value, preset the calibration threshold, compare the calibration value with the preset calibration threshold, if the calibration value is greater than the preset calibration threshold, generate a position sensor calibration instruction and send it to the initial setting module, after receiving the instruction, the initial setting module controls the position sensor to perform calibration;

[0028] When the sensor is being updated, the standard deviation of the temperature and humidity values ​​at each collection moment within the preset detection time is calculated and recorded as the temperature wave value and the humidity wave value. The temperature wave threshold and the humidity wave threshold are preset, and the temperature wave value and the humidity wave value are compared with the corresponding thresholds respectively. If the temperature wave value and the humidity wave value are greater than the corresponding thresholds, the corresponding sensor calibration instruction is generated and sent to the initial setting module. The initial setting module controls the corresponding sensor for calibration.

[0029] Preferably, the working mode of the voice quality detection module is:

[0030] When receiving a voice signal detection instruction, the voice signal received by the microphone on the rice bucket is converted into a series of discrete sample value sequences and recorded as Xm, where m represents the number of the sample value, m=1, 2..., g; g is the total number of sample values;

[0031] For each sample value Xm, obtain its instantaneous amplitude |Xm|; use the formula: Get the speech amplitude FD;

[0032] The discretized speech signal is converted from the time domain to the frequency domain through fast Fourier transform, and the main frequency range of human speech 300Hz-3400Hz is set as the optimal identification frequency band. The amplitude values ​​of all frequency points in the optimal identification frequency band and the amplitude values ​​of all frequency points in the entire frequency domain are calculated to obtain the optimal amplitude total value and the full amplitude total value. The optimal frequency ratio YZ is obtained by dividing the optimal amplitude total value by the full amplitude total value.

[0033] After the speech amplitude value FD and the preferred frequency ratio value YZ are normalized, they are inserted into the formula: YPZ = FD × c1 + YZ × c2 to obtain the speech evaluation value YPZ, where c1 and c2 are preset coefficients, and three speech evaluation value intervals are preset, namely, the first-level speech evaluation value interval, the second-level speech evaluation value interval, and the third-level speech evaluation value interval, and the speech evaluation value is matched with the three speech evaluation value intervals;

[0034] If the speech evaluation value is in the first level range, the speech recognition chip is controlled to directly recognize the speech command;

[0035] If the speech evaluation value is in the second level interval of the speech evaluation value, a speech compensation instruction is generated and sent to the speech compensation module;

[0036] If the speech evaluation value is in the third level range, the Mi Tong microphone is controlled to broadcast the voice message "The command is not clear, please resend the voice command".

[0037] Preferably, the voice compensation module works as follows:

[0038] When a voice compensation instruction is received, a voice signal amplitude amplification coefficient is preset, and the voice signal amplitude amplification factor is obtained by multiplying the ring position deviation value corresponding to the voice signal by the preset voice signal amplitude amplification coefficient. A voice signal amplitude amplification factor threshold is preset, and the voice signal amplitude amplification factor is compared with the corresponding threshold. If the voice signal amplitude amplification factor is greater than or equal to the corresponding threshold, the amplitude of each frequency point of the voice signal is multiplied by the voice signal amplitude amplification factor threshold for amplification; if the voice signal amplitude amplification factor is less than the corresponding threshold, the amplitude of each frequency point of the voice signal is multiplied by the voice signal amplitude amplification factor for amplification. The amplified voice signal is sent to the voice quality detection module for re-detection, and the speech evaluation value level interval corresponding to the amplified voice signal is obtained, and the corresponding operation is performed according to the corresponding speech evaluation value level interval. If the voice signal is matched to the second-level speech evaluation value interval again after amplification, the rice bucket microphone is directly controlled to broadcast the voice "The instruction is not clear, please resend the voice instruction".

[0039] Preferably, a control method of a rice bucket intelligent voice control system comprises the following steps:

[0040] Step 1: When the rice bucket is started or receives a detection sensor calibration instruction, the detection sensor is controlled to calibrate;

[0041] Step 2: Use the LiDAR sensor and image sensor to detect the user's activity area in the kitchen, and realize the intelligent conversion of the rice bucket between the standby state and the working state;

[0042] Step 3: When the user issues a voice command, the offset value is calculated based on the user's location, and the temperature shadow value and humidity shadow value are obtained based on the temperature and humidity, and then the ring position impact value is obtained through comprehensive analysis; if the ring position impact value exceeds the preset threshold, a voice signal detection command is generated;

[0043] Step 4: When receiving the voice signal detection command, the voice signal is discretized, the speech amplitude value and the optimal frequency ratio value are calculated, and the speech evaluation value is obtained by comprehensive analysis, and the voice recognition, compensation or unrecognizable command is generated according to the preset interval matching;

[0044] Step 5: After receiving the voice compensation instruction, amplify and compensate the voice signal.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] (1) A rice bucket intelligent voice control system and method, which collects environmental temperature and humidity through a temperature sensor and a humidity sensor, analyzes the temperature shadow value and the humidity shadow value, and calculates the ring position influence value in combination with the detection distance value and the detection angle of the user position; when the ring position influence value exceeds a preset threshold, a voice signal detection instruction is generated and sent to a voice quality detection module. If the detection result needs to be compensated, a voice compensation instruction is generated and sent to a voice compensation module. The voice compensation module obtains a voice signal amplitude magnification factor by multiplying the ring position deviation value corresponding to the voice signal by a preset voice signal amplitude magnification factor. If the voice signal amplitude magnification factor is greater than or equal to the corresponding threshold, each frequency of the voice signal is magnified. The amplitude of each point is multiplied by the threshold of the voice signal amplitude amplification factor for amplification; if the voice signal amplitude amplification factor is less than the corresponding threshold, the amplitude of each frequency point of the voice signal is multiplied by the voice signal amplitude amplification factor for amplification, and the amplified voice signal is sent to the voice quality detection module for re-detection, and the speech evaluation value level interval corresponding to the amplified voice signal is obtained, and the corresponding operation is performed according to the corresponding speech evaluation value level interval. If the voice signal is matched to the second-level speech evaluation value interval again after amplification, the rice bucket microphone is directly controlled to broadcast the voice of "The command is not clear, please resend the voice command"; effectively reduce the interference of environmental factors and location factors on the accuracy of speech recognition.

[0047] (2) The rice bucket intelligent voice control system and method uses a laser radar sensor to scan the kitchen to obtain the area range, and determines the detection diameter by combining the user's kitchen area, the number of home appliances and the activity area analysis to delineate the detection area; based on the image sensor's detection of the user's activity time in the detection area, the rice bucket can be intelligently switched between standby and working states; when no one uses the kitchen and there is no user activity in the detection area for a long time, the rice bucket automatically enters the standby state to avoid long-term standby power consumption; and when the user enters the detection area and stays there for a preset power-on time, the rice bucket quickly enters the working state and responds to voice commands at any time, which improves the convenience of use and effectively reduces energy loss.

[0048] (3) The rice bucket intelligent voice control system and method periodically calibrates and detects the position sensor, temperature sensor and humidity sensor; takes the fixed furniture in the detection area as the detection object, obtains its initial coordinates and calculates the calibration coordinates through multiple restarts, and compares the calibration value of the actual coordinates of the detection object with the calibration coordinates when the detection sensor is updated, as well as the standard deviation of the temperature and humidity at each acquisition time and the corresponding threshold; once the calibration value or the temperature wave value or the humidity wave value exceeds the threshold, a calibration instruction is immediately generated to ensure the accuracy of the sensor data, thereby improving the reliability and stability of the entire intelligent voice control system; during long-term use, even if the sensor is affected by environmental factors or other factors and causes data deviation, the system can detect and correct it in time to ensure that the system is always in good working condition, further improving the accuracy of rice bucket in voice recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] Embodiment 1

[0052] See also Figure 1 ,The present invention provides a rice bucket intelligent voice control system and method, including: an initial setting module, a standby conversion module, a monitoring and analysis module, a voice quality detection module, and a voice compensation module;

[0053] Initial setting module: The rice bucket is equipped with a lidar sensor, an image sensor and a detection sensor; the detection sensor includes a position sensor, a humidity sensor and a temperature sensor; when the rice bucket is restarted, the detection sensor is controlled to be calibrated;

[0054] When the position sensor, temperature sensor and humidity sensor are calibrated, the infrared technology positioning device in the position sensor will perform self-test; the temperature sensor and humidity sensor will perform self-test on their own measurement accuracy to ensure the accuracy of the collected data;

[0055] The standby conversion module detects the user's activity area in the kitchen through a lidar sensor and an image sensor to realize the intelligent conversion of the rice bucket between the standby state and the working state. The specific process is as follows:

[0056] The kitchen environment is scanned in all directions by using a laser radar sensor to obtain the area range of the kitchen. A circle is drawn within the area range of the kitchen with the rice bucket as the center and the detection diameter as the radius to obtain a detection circle. The area covered by the detection circle and the area range of the kitchen that overlap is recorded as the detection area. The detection area is detected by an image sensor, and a power-on time is preset. When the image sensor detects that a user appears in the detection area and the duration of the stay is greater than the preset power-on time, the rice bucket is controlled to enter the working state from the standby state. When the image sensor cannot detect a user in the detection area and the duration of the stay is greater than the preset power-on time, the rice bucket is controlled to enter the standby state from the working state.

[0057] The detection diameter is obtained as follows:

[0058] The kitchen area is scanned by a lidar sensor to obtain the user's kitchen area value MZ, the number of home appliances in the kitchen ZS, and the user's activity area value YH in the kitchen. After normalization, they are substituted into the formula: ZP=MZ×a1+ZS×a2+YH×a3 to obtain the detection assignment ZP, where a1, a2, and a3 are preset weight coefficients, and home appliances include but are not limited to: refrigerators, microwave ovens, and ovens; multiple detection assignment intervals are preset, and each interval corresponds to a detection diameter. The detection assignment ZP is matched with multiple detection assignment intervals, and the detection diameter corresponding to the successfully matched detection assignment interval is output as the detection diameter of the detection circle drawn by the user.

[0059] The detection and analysis module uses position sensors, temperature sensors, humidity sensors and a circular array of microphones distributed at different positions of the rice bucket. When the rice bucket is in working state, it can realize the positioning of the user's position and real-time monitoring of the ambient temperature and humidity. When the user issues a voice command, the offset value is calculated according to the user's position, and the temperature shadow value and humidity shadow value are obtained according to the temperature and humidity, and then the ring position impact value is obtained through comprehensive analysis; if the ring position impact value exceeds the preset threshold, a voice signal detection command is generated and sent to the voice quality detection module. The specific process is as follows:

[0060] When the rice bucket enters the working state from the standby state, multiple position sensors distributed in different positions of the rice bucket detect the signals sent by the users entering the detection area; for each position sensor, the time difference between the position sensor sending the signal to the user and the user reflecting the signal at the acquisition time is obtained, which is recorded as the detection time difference. According to the speed of the signal sent by the position sensor in the air and the detection time difference, the distance between the user and the position sensor can be calculated and marked as the detection distance. According to the detection distance between the user and each sensor, the coordinates of the user relative to each sensor with itself as the origin of the coordinate system can be calculated by using the triangulation method (XUi, YUi), where i is the number of the position sensor, i=1, 2..., n; n is the total number of position sensors;

[0061] Establish a two-dimensional rectangular coordinate system with the center of the rice bucket as the origin. Set the coordinates of the center of the rice bucket to (X0, Y0). Use the formula: The coordinates (XU, YU) of the user relative to the rice bucket are obtained and recorded as the detection coordinates, where Wi is the preset weight coefficient assigned to each position sensor;

[0062] When the user issues a voice command, the circular array of microphones on the rice bucket can receive the voice command. When the voice command is received, the position sensor is used to immediately detect the coordinates of the user who issued the voice command, and the coordinates of the user relative to the rice bucket when the command is issued are obtained and recorded as the detection coordinates. The user's detection coordinates are connected with the center of the rice bucket to obtain the detection line; the direction of the rice bucket facing the user's normal operation when it leaves the factory is recorded as the positive direction of the rice bucket, and the angle between the detection line and the positive direction of the rice bucket is obtained, recorded as the detection angle ZJ. At the same time, the distance formula between the two points is used to calculate the distance value between the user's detection coordinates and the center of the rice bucket, recorded as the detection distance value ZG;

[0063] Within the detection area, a fixed piece of furniture is randomly selected as the detection object, and the coordinates of the object detection relative to the rice bucket are obtained when the rice bucket is restarted for three consecutive times to detect the detection object. The horizontal coordinates and the vertical coordinates of the coordinates of the detection object relative to the rice bucket detected three times are averaged respectively to obtain the horizontal coordinate mean and the vertical coordinate mean of the detection object, and the coordinates composed of the horizontal coordinate mean and the vertical coordinate mean of the detection object are marked as calibration coordinates;

[0064] Set the detection sensor update interval and preset the detection time. When the detection sensor is in the update moment, obtain the coordinates of the detection object at each collection moment relative to the rice bucket within the preset detection time, perform Euclidean distance calculation on the coordinates of the detection object at each collection moment relative to the rice bucket and the calibration coordinates of the detection object, and calculate the average value thereof, obtain the average value of the distance between the coordinates of the detection object image relative to the rice bucket and the calibration coordinates at each collection moment, record it as the calibration value, preset the calibration threshold, compare the calibration value with the preset calibration threshold, if the calibration value is greater than the preset calibration threshold, generate a position sensor calibration instruction and send it to the initial setting module, after receiving the instruction, the initial setting module controls the position sensor to perform calibration;

[0065] The temperature and humidity values ​​of the surrounding environment can be collected by the temperature sensor and the humidity sensor, and several temperature intervals and humidity intervals are set. Each temperature interval corresponds to a preset temperature influence coefficient of sound propagation, and each humidity interval corresponds to a preset humidity influence coefficient of sound propagation. The temperature value and humidity value at the time of collection are matched with the corresponding temperature interval and humidity interval, and the corresponding influence coefficient is output, and multiplied by the corresponding influence coefficient to obtain the temperature shadow value WP and the humidity shadow value SP corresponding to the time of collection;

[0066] When the sensor is being updated, the standard deviation of the temperature and humidity values ​​at each collection time within the preset detection time is calculated and recorded as the temperature wave value and the humidity wave value. The temperature wave threshold and the humidity wave threshold are preset, and the temperature wave value and the humidity wave value are compared with the corresponding thresholds respectively. If the temperature wave value and the humidity wave value are greater than the corresponding threshold, a corresponding sensor calibration instruction is generated and sent to the initial setting module, and the initial setting module controls the corresponding sensor for calibration;

[0067] The temperature shadow value WP, wet shadow value SP, detection angle ZJ and offset value ZG at the moment of receiving the voice command are obtained, and after normalization, they are substituted into the preset formula: HWY=ZJ×b1+ZG×b2+WP×b3+SP×b4 to obtain the ring position influence value HWY, where b1, b2, b3, and b4 are preset weight factors. The larger the ring position influence value, the greater the degree to which the signal strength and signal quality of the rice bucket are affected by the position, temperature, and humidity when the user sends the voice command. A preset ring position influence threshold is set, and the ring position influence value is compared with the preset ring position influence threshold. If the ring position influence value is greater than the preset ring position influence threshold, a voice signal detection instruction is generated and sent to the voice quality detection module. At the same time, the preset ring position influence threshold is subtracted from the ring position influence value to obtain the ring position deviation value, and the ring position deviation value is sent to the voice compensation module.

[0068] When the voice quality detection module receives the voice signal detection instruction, it discretizes the voice signal to obtain the sample value, calculates the speech amplitude value and the optimal frequency ratio value, and then comprehensively obtains the speech evaluation value, matches it according to the preset speech evaluation value interval, and generates corresponding instructions and executes corresponding operations according to the matching results, including speech recognition instructions, speech compensation instructions or unrecognizable instructions, thereby realizing intelligent detection and classification processing of the voice instruction quality. The specific process is as follows:

[0069] When receiving a voice signal detection instruction, the voice signal received by the microphone on the rice bucket is converted into a series of discrete sample value sequences and recorded as Xm, where m represents the number of the sample value, m=1, 2..., g; g is the total number of sample values;

[0070] For each sample value Xm, obtain its instantaneous amplitude |Xm|; for example, if a sample value is (-0.05)V, then its instantaneous amplitude is (0.05)V; using the formula: The speech amplitude value FD is obtained, wherein the larger the speech amplitude value is, the stronger the signal of the voice command received by the rice bucket is;

[0071] The discretized speech signal is converted from the time domain to the frequency domain through fast Fourier transform, and the main frequency range of human speech 300Hz-3400Hz is set as the optimal frequency band. The amplitude values ​​of all frequency points in the optimal frequency band and the amplitude values ​​of all frequency points in the entire frequency domain are calculated to obtain the total optimal amplitude value and the total full amplitude value. The optimal frequency ratio YZ is obtained by dividing the total optimal amplitude value by the total full amplitude value; the larger the optimal frequency ratio, the higher the relative concentration of the energy of the speech signal in the main frequency range of human speech, which means that the possibility of the speech signal being interfered by other frequency bands is smaller, and accordingly, the quality of the speech signal may be relatively better;

[0072] After normalizing the speech amplitude value FD and the optimal frequency ratio value YZ, substitute them into the formula: YPZ=FD×c1+YZ×c2 to obtain the speech evaluation value YPZ, where c1 and c2 are preset coefficients. The larger the speech evaluation value, the better the signal of the voice command and the higher the quality; three speech evaluation value intervals are preset, namely, the first-level speech evaluation value interval, the second-level speech evaluation value interval and the third-level speech evaluation value interval, and the speech evaluation value is matched with the three speech evaluation value intervals;

[0073] If the speech evaluation value is in the first level range, the speech recognition chip is controlled to directly recognize the speech command;

[0074] If the speech evaluation value is in the second level interval of the speech evaluation value, a speech compensation instruction is generated and sent to the speech compensation module;

[0075] If the speech evaluation value is in the third level range, the Mi Tong microphone is controlled to broadcast the voice message "The command is not clear, please resend the voice command".

[0076] After receiving the voice compensation instruction, the voice compensation module amplifies and compensates the voice signal. The specific process is as follows:

[0077] When a voice compensation instruction is received, a voice signal amplitude amplification coefficient is preset, and the voice signal amplitude amplification factor is obtained by multiplying the ring position deviation value corresponding to the voice signal by the preset voice signal amplitude amplification coefficient. A voice signal amplitude amplification factor threshold is preset, and the voice signal amplitude amplification factor is compared with the corresponding threshold. If the voice signal amplitude amplification factor is greater than or equal to the corresponding threshold, the amplitude of each frequency point of the voice signal is multiplied by the voice signal amplitude amplification factor threshold for amplification; if the voice signal amplitude amplification factor is less than the corresponding threshold, the amplitude of each frequency point of the voice signal is multiplied by the voice signal amplitude amplification factor for amplification. The amplified voice signal is sent to the voice quality detection module for re-detection, and the speech evaluation value level interval corresponding to the amplified voice signal is obtained, and the corresponding operation is performed according to the corresponding speech evaluation value level interval. If the voice signal is matched to the second-level speech evaluation value interval again after amplification, the rice bucket microphone is directly controlled to broadcast the voice "The instruction is not clear, please resend the voice instruction".

[0078] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A rice bucket intelligent voice control system, comprising: The initial setting module, the standby conversion module, the monitoring and analysis module, the voice quality detection module and the voice compensation module are characterized by: Initial setting module: The rice bucket is equipped with a lidar sensor, an image sensor and a detection sensor; the detection sensors include: position, humidity and temperature sensors; when the rice bucket is restarted or receives a detection sensor calibration instruction, the detection sensor is controlled to calibrate; Standby conversion module: Use LiDAR to scan the kitchen to obtain the scope of the kitchen area, analyze the user's kitchen area, the number of home appliances, and the activity area in the kitchen to obtain the detection diameter, draw a detection circle based on the detection diameter, and define the detection area based on the kitchen area and the detection circle. According to the activity duration of the user in the detection area, control the rice bucket to switch between standby and working states; Monitoring and analysis module: When the rice bucket is working, the position sensor locates the coordinates of the user relative to the rice bucket and calculates the offset value. The temperature and humidity sensors collect the ambient temperature and humidity and analyze them to obtain the temperature shadow value and the humidity shadow value. The user issues a voice command, and the ring position impact value is obtained by comprehensive analysis based on the temperature shadow value, the humidity shadow value and the offset value. If the ring position impact value exceeds the preset threshold, a voice signal detection command is generated and sent to the voice quality detection module. The position sensor, temperature sensor and humidity sensor are regularly calibrated and tested. Voice quality detection module: when receiving voice signal detection instructions, it discretizes the voice signal, calculates the speech amplitude value and the optimal frequency ratio value, and comprehensively analyzes and obtains the speech evaluation value, generates voice recognition, compensation or unrecognizable instructions according to the preset interval matching, and sends the compensation instructions to the voice compensation module; Voice compensation module: after receiving the voice compensation instruction, it amplifies and compensates the voice signal.

2. The intelligent voice control system for rice bucket according to claim 1 is characterized by: The specific process of controlling the rice bucket to switch between standby and working states is as follows: A laser radar sensor is used to perform an all-round scan of the kitchen environment to obtain the area range of the kitchen. A circle is drawn within the kitchen area with the rice bucket as the center and the detection diameter as the radius to obtain a detection circle. The area covered by the detection circle and the overlapping area of ​​the kitchen area are recorded as the detection area. The detection area is detected by an image sensor, and a power-on time is preset. When the image sensor detects that a user appears in the detection area and the stay time is greater than the preset power-on time, the rice bucket is controlled to enter the working state from the standby state; when the image sensor cannot detect the user in the detection area and the duration is greater than the preset power-on time, the rice bucket is controlled to enter the standby state from the working state.

3. The intelligent voice control system for rice bucket according to claim 2 is characterized by: The detection diameter is obtained as follows: The kitchen area is scanned by a lidar sensor to obtain the user's kitchen area value MZ, the number of home appliances in the kitchen ZS, and the user's activity area value YH in the kitchen. After normalization, they are substituted into the formula: ZP=MZ×a1+ZS×a2+YH×a3 to obtain the detection assignment ZP, where a1, a2, and a3 are preset weight coefficients. Home appliances include: refrigerators, microwave ovens, and ovens. Multiple detection assignment intervals are preset, and each interval corresponds to a detection diameter. The detection assignment is matched with multiple detection assignment intervals, and the detection diameter corresponding to the successfully matched detection assignment interval is output as the detection diameter of the detection circle drawn by the user.

4. The intelligent voice control system for rice bucket according to claim 1 is characterized by: The specific process of the position sensor locating the user's coordinates relative to the rice bucket and calculating the offset value is as follows: When the rice bucket enters the working state from the standby state, multiple position sensors distributed in different positions of the rice bucket detect the signals sent by users entering the detection area; for each position sensor, the time difference between the position sensor sending the signal to the user and the user reflecting the signal at the time of collection is obtained, which is recorded as the detection time difference. According to the speed of the signal sent by the position sensor in the air and the detection time difference, the distance between the user and the position sensor can be calculated and marked as the detection distance. According to the detection distance between the user and each sensor, the coordinates of the user relative to each sensor with itself as the origin of the coordinate system can be calculated by using the triangulation method (XUi, YUi), where i is the number of the position sensor, i=1, 2..., n; n is the total number of position sensors; Establish a two-dimensional rectangular coordinate system with the center of the rice bucket as the origin. Set the coordinates of the center of the rice bucket to (X0, Y0), and use the formula: , the coordinates (XU, YU) of the user relative to the rice bucket are obtained and recorded as the detection coordinates, where Wi is the preset weight coefficient assigned to each position sensor; When the user issues a voice command, the circular array of microphones on the rice bucket can receive the voice command. When the voice command is received, the coordinates of the user who issued the voice command are immediately detected through the position sensor, and the coordinates of the user relative to the rice bucket when the command is issued are obtained, and recorded as the detection coordinates. The user's detection coordinates are connected with the center of the rice bucket to obtain the detection line; the direction of the user's routine operation in the rice bucket set by default when leaving the factory is recorded as the positive direction of the rice bucket, and the angle between the detection line and the positive direction of the rice bucket is obtained, recorded as the detection angle ZJ, and the distance formula between the two points is used to calculate the distance value between the user's detection coordinates and the center of the rice bucket, which is recorded as the detection distance value ZG.

5. The intelligent voice control system for rice bucket according to claim 4 is characterized by: The specific process of analyzing the ring position impact value and generating a voice signal detection instruction to send to the voice quality detection module is as follows: The temperature and humidity values ​​of the surrounding environment are collected by the temperature sensor and the humidity sensor, and several temperature intervals and humidity intervals are set. Each temperature interval corresponds to a preset temperature influence coefficient of sound propagation, and each humidity interval corresponds to a preset humidity influence coefficient of sound propagation. The temperature value and humidity value at the time of collection are matched with the corresponding temperature interval and humidity interval, and the corresponding influence coefficient is output, and multiplied by the corresponding influence coefficient to obtain the temperature shadow value WP and the humidity shadow value SP corresponding to the time of collection; The temperature shadow value WP, wet shadow value SP, detection angle ZJ and offset value ZG at the moment the voice command is received are obtained, and after normalization, they are substituted into the preset formula: HWY=ZJ×b1+ZG×b2+WP×b3+SP×b4 to obtain the ring position influence value HWY, where b1, b2, b3, and b4 are preset weight factors and a preset ring position influence threshold. The ring position influence value is compared with the preset ring position influence threshold. If the ring position influence value is greater than the preset ring position influence threshold, a voice signal detection instruction is generated and sent to the voice quality detection module. At the same time, the preset ring position influence threshold is subtracted from the ring position influence value to obtain the ring position deviation value, and the ring position deviation value is sent to the voice compensation module.

6. The intelligent voice control system for rice bucket according to claim 4 is characterized by: The process of regularly calibrating and testing the position sensor, temperature sensor, and humidity sensor is as follows: Within the detection area, a fixed piece of furniture is randomly selected as the detection object, and the coordinates of the object detection relative to the rice bucket are obtained when the rice bucket is restarted for three consecutive times to detect the detection object. The horizontal coordinates and the vertical coordinates of the coordinates of the detection object relative to the rice bucket detected three times are averaged respectively to obtain the horizontal coordinate mean and the vertical coordinate mean of the detection object, and the coordinates composed of the horizontal coordinate mean and the vertical coordinate mean of the detection object are marked as calibration coordinates; Set the detection sensor update interval and preset the detection time. When the detection sensor is in the update moment, obtain the coordinates of the detection object at each collection moment relative to the rice bucket within the preset detection time, perform Euclidean distance calculation on the coordinates of the detection object at each collection moment relative to the rice bucket and the calibration coordinates of the detection object, and calculate the average value thereof, obtain the average value of the distance between the coordinates of the detection object image relative to the rice bucket and the calibration coordinates at each collection moment, record it as the calibration value, preset the calibration threshold, compare the calibration value with the preset calibration threshold, if the calibration value is greater than the preset calibration threshold, generate a position sensor calibration instruction and send it to the initial setting module, after receiving the instruction, the initial setting module controls the position sensor to perform calibration; When the sensor is being updated, the standard deviation of the temperature and humidity values ​​at each collection moment within the preset detection time is calculated and recorded as the temperature wave value and the humidity wave value. The temperature wave threshold and the humidity wave threshold are preset, and the temperature wave value and the humidity wave value are compared with the corresponding thresholds respectively. If the temperature wave value and the humidity wave value are greater than the corresponding thresholds, the corresponding sensor calibration instruction is generated and sent to the initial setting module. The initial setting module controls the corresponding sensor for calibration.

7. The intelligent voice control system for rice bucket according to claim 5 is characterized by: The voice quality detection module works as follows: When receiving a voice signal detection instruction, the voice signal received by the microphone on the rice bucket is converted into a series of discrete sample value sequences and recorded as Xm, where m represents the number of the sample value, m=1, 2..., g; g is the total number of sample values; For each sample value Xm, obtain its instantaneous amplitude |Xm|; use the formula: to obtain the amplitude FD; The discretized speech signal is converted from the time domain to the frequency domain through fast Fourier transform, and the main frequency range of human speech 300Hz-3400Hz is set as the optimal identification frequency band. The amplitude values ​​of all frequency points in the optimal identification frequency band and the amplitude values ​​of all frequency points in the entire frequency domain are calculated to obtain the optimal amplitude total value and the full amplitude total value. The optimal frequency ratio YZ is obtained by dividing the optimal amplitude total value by the full amplitude total value. After the speech amplitude value FD and the preferred frequency ratio value YZ are normalized, they are inserted into the formula: YPZ=FD×c1+YZ×c2 to obtain the speech evaluation value YPZ, where c1 and c2 are preset coefficients, and three speech evaluation value intervals are preset, namely, the first-level speech evaluation value interval, the second-level speech evaluation value interval, and the third-level speech evaluation value interval, and the speech evaluation value is matched with the three speech evaluation value intervals; If the speech evaluation value is in the first level range, the speech recognition chip is controlled to directly recognize the speech command; If the speech evaluation value is in the second level interval of the speech evaluation value, a speech compensation instruction is generated and sent to the speech compensation module; If the speech evaluation value is in the third level range, the Mi Tong microphone is controlled to broadcast the voice "The command is not clear, please resend the voice command".

8. The intelligent voice control system for rice bucket according to claim 7 is characterized by: The voice compensation module works as follows: When a voice compensation instruction is received, a voice signal amplitude amplification coefficient is preset, and the voice signal amplitude amplification factor is obtained by multiplying the ring position deviation value corresponding to the voice signal by the preset voice signal amplitude amplification coefficient. A voice signal amplitude amplification factor threshold is preset, and the voice signal amplitude amplification factor is compared with the corresponding threshold. If the voice signal amplitude amplification factor is greater than or equal to the corresponding threshold, the amplitude of each frequency point of the voice signal is multiplied by the voice signal amplitude amplification factor threshold for amplification; if the voice signal amplitude amplification factor is less than the corresponding threshold, the amplitude of each frequency point of the voice signal is multiplied by the voice signal amplitude amplification factor for amplification. The amplified voice signal is sent to the voice quality detection module for re-detection, and the speech evaluation value level interval corresponding to the amplified voice signal is obtained, and the corresponding operation is performed according to the corresponding speech evaluation value level interval. If the voice signal is matched to the second-level speech evaluation value interval again after amplification, the rice bucket microphone is directly controlled to broadcast the voice "The instruction is not clear, please resend the voice instruction".

9. The control method of a rice bucket intelligent voice control system according to claims 1-8, characterized in that: The following steps are involved: Step 1: When the rice bucket is started or receives a detection sensor calibration instruction, the detection sensor is controlled to perform calibration; Step 2: Use the LiDAR sensor and image sensor to detect the user's activity area in the kitchen, and realize the intelligent conversion of the rice bucket between the standby state and the working state; Step 3: When the user issues a voice command, the offset value is calculated based on the user's location, and the temperature shadow value and humidity shadow value are obtained based on the temperature and humidity, and then the ring position impact value is obtained through comprehensive analysis; if the ring position impact value exceeds the preset threshold, a voice signal detection command is generated; Step 4: When receiving the voice signal detection command, the voice signal is discretized, the speech amplitude value and the optimal frequency ratio value are calculated, and the speech evaluation value is obtained by comprehensive analysis, and the voice recognition, compensation or unrecognizable command is generated according to the preset interval matching; Step 5: After receiving the voice compensation instruction, the voice signal is amplified and compensated.

Images