WO2021051944A1 - 自动推送助眠乐曲方法、装置、计算机设备及存储介质 - Google Patents

自动推送助眠乐曲方法、装置、计算机设备及存储介质 Download PDF

Info

Publication number
WO2021051944A1
WO2021051944A1 PCT/CN2020/099515 CN2020099515W WO2021051944A1 WO 2021051944 A1 WO2021051944 A1 WO 2021051944A1 CN 2020099515 W CN2020099515 W CN 2020099515W WO 2021051944 A1 WO2021051944 A1 WO 2021051944A1
Authority
WO
WIPO (PCT)
Prior art keywords
sleep
preset
aid music
signal
spectrogram
Prior art date
Application number
PCT/CN2020/099515
Other languages
English (en)
French (fr)
Inventor
王健宗
亢祖衡
彭俊清
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2021051944A1 publication Critical patent/WO2021051944A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M21/00—Other devices or methods to cause a change in the state of consciousness; Devices for producing or ending sleep by mechanical, optical, or acoustical means, e.g. for hypnosis
    • A61M21/02—Other devices or methods to cause a change in the state of consciousness; Devices for producing or ending sleep by mechanical, optical, or acoustical means, e.g. for hypnosis for inducing sleep or relaxation, e.g. by direct nerve stimulation, hypnosis, analgesia
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M21/00—Other devices or methods to cause a change in the state of consciousness; Devices for producing or ending sleep by mechanical, optical, or acoustical means, e.g. for hypnosis
    • A61M2021/0005—Other devices or methods to cause a change in the state of consciousness; Devices for producing or ending sleep by mechanical, optical, or acoustical means, e.g. for hypnosis by the use of a particular sense, or stimulus
    • A61M2021/0027—Other devices or methods to cause a change in the state of consciousness; Devices for producing or ending sleep by mechanical, optical, or acoustical means, e.g. for hypnosis by the use of a particular sense, or stimulus by the hearing sense
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M2230/00—Measuring parameters of the user
    • A61M2230/08—Other bio-electrical signals
    • A61M2230/10—Electroencephalographic signals

Definitions

  • This application relates to the field of artificial intelligence technology, and in particular to a method, device, computer equipment and storage medium for automatically pushing sleep aid music.
  • the embodiments of the present application provide a method, device, computer equipment, and storage medium for automatically pushing sleep aid music, so as to solve the problem of not being able to accurately recommend sleep aid music for people and reducing people's sleep quality.
  • a method for automatically pushing sleep aid music including:
  • a device for automatically pushing sleep aid music including:
  • the first acquisition module is configured to acquire the voltage signal and brain electrical signal collected by the target user during sleep at a preset time interval
  • the vector conversion module is used to perform vector conversion on the voltage signal to obtain a characteristic vector
  • the graph conversion module is used to perform graph conversion processing on the EEG signal to obtain a spectrogram
  • a recognition module configured to import the feature vector and the spectrogram into a pre-trained sleep detection model for recognition, and obtain a sleep state;
  • the second obtaining module is configured to obtain the recommended type of sleep aid music corresponding to the sleep state based on a preset condition
  • the push module is used to push the sleep aid music for the target user according to the recommended type of sleep aid music.
  • a computer device includes a memory, a processor, and computer-readable instructions stored in the memory and capable of running on the processor.
  • the processor executes the computer-readable instructions, the following automatic push assistance is implemented Steps of sleep music method:
  • a non-volatile computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the following automatic push sleep aid is realized The steps of the music method:
  • the above-mentioned method, device, computer equipment and storage medium for automatically pushing sleep aid music can obtain the voltage signal and EEG signal collected by the target user during sleep at preset time intervals, and convert the voltage signal and EEG signal into feature vectors and Spectrogram, and import the feature vector and spectrogram into the pre-trained sleep detection model for recognition, get the sleep state, determine the corresponding sleep aid music recommendation type according to the sleep state, and push the sleep aid music recommendation type corresponding to the target user Sleep aid music.
  • the sleep detection model the sleep state corresponding to the target user can be accurately identified, and appropriate sleep aid music can be accurately pushed to the target user according to the sleep state, and the sleep quality of the target user can be further improved.
  • FIG. 1 is a flowchart of a method for automatically pushing sleep aid music provided by an embodiment of the present application
  • step S2 is a flowchart of step S2 in the method for automatically pushing sleep aid music provided by an embodiment of the present application
  • step S3 is a flowchart of step S3 in the method for automatically pushing sleep aid music provided by an embodiment of the present application
  • FIG. 4 is a flow chart of obtaining a sleep detection model through training using training samples in the method for automatically pushing sleep aid music provided by an embodiment of the present application;
  • step S72 is a flowchart of step S72 in the method for automatically pushing sleep aid music provided by an embodiment of the present application
  • step S5 is a flowchart of step S5 in the method for automatically pushing sleep aid music provided by an embodiment of the present application
  • FIG. 7 is a flowchart of adjusting the playback volume or pausing playback according to the playback time of the sleep-aid music in the method for automatically pushing sleep-aid music provided by an embodiment of the present application;
  • FIG. 8 is a schematic diagram of a device for automatically pushing sleep aid music provided by an embodiment of the present application.
  • Fig. 9 is a basic structural block diagram of a computer device provided by an embodiment of the present application.
  • the method for automatically pushing sleep aid music provided in this application is applied to the server, and the server can be implemented by an independent server or a server cluster composed of multiple servers.
  • a method for automatically pushing sleep aid music is provided, which includes the following steps:
  • S1 Obtain the voltage signal and EEG signal collected by the target user during sleep at a preset time interval.
  • the voltage signal is collected by a sensor dedicated to collecting voltage signals.
  • the voltage signal When the target user sleeps in a place with the sensor, the voltage signal will be saved in the preset database in real time; EEG signal The collection is carried out through a wristband worn by the user specifically for collecting EEG signals. When the target user wears the bracelet, the EEG signals will be saved in the preset database in real time.
  • the voltage signal and the brain electrical signal collected by the target user during sleep are directly obtained from the preset database. If the preset time interval is 1 minute, the voltage signal and EEG signal are obtained from the preset database every 1 minute.
  • the preset database refers to a database specifically used to store voltage signals and brain electrical signals collected by the target user during sleep.
  • the value of the preset time interval can be 1 minute or 5 minutes, and there is no restriction here.
  • the senor is a mattress with a varistor embedded in it. A series of circuits are discarded inside the mattress and the varistors are distributed in different positions of the mattress. In the process of measuring data, the mattress will The voltage value on each varistor will be returned in real time, and its voltage value will change according to the pressure generated by the patient on the bed.
  • S2 Perform vector conversion on the voltage signal to obtain a characteristic vector.
  • the preset vector conversion port refers to a processing port specifically used to convert a voltage signal into a feature vector.
  • the EEG signal obtained in step S1 is imported into the preset map conversion port for map conversion processing, so as to obtain the spectrogram after the map conversion processing.
  • the preset map conversion port refers to a processing port specifically used to convert an EEG signal into a spectrogram.
  • the obtained feature vector and spectrogram are input into the pre-trained sleep monitoring model for identification.
  • the sleep monitoring model will directly determine the corresponding sleep state based on the input feature vector and spectrogram, and perform the sleep state Output.
  • S5 Obtain the recommended type of sleep aid music corresponding to the sleep state based on preset conditions.
  • the sleep-aid music recommendation type corresponding to the sleep state is obtained from the preset recommendation library.
  • the preset condition refers to a rule set according to the actual needs of the user, which specifically may be to determine the type of sleep aid music recommendation based on the same sleep state that is continuously recognized.
  • the preset recommendation library refers to a database specially used to store different description information and the types of sleep aid music recommendations corresponding to the description information.
  • S6 Push sleep-aid music for the target user according to the recommended type of sleep-aid music.
  • any one of the sleep aid music under the recommended type of sleep aid music is randomly selected from the preset music library, and the sleep aid music is output to the preset playback port for playback for the target user .
  • the preset playback port refers to a processing port specifically used to play sleep aid music for the target user.
  • the voltage signals and EEG signals collected by the target user during sleep are acquired at preset time intervals, the voltage signals and EEG signals are converted into feature vectors and spectrograms respectively, and the feature vectors and spectrograms are imported Recognition is performed in the pre-trained sleep detection model, the sleep state is obtained, the corresponding sleep-aid music recommendation type is determined according to the sleep state, and the sleep-aid music corresponding to the sleep-aid music recommendation type is pushed to the target user.
  • the sleep detection model the sleep state corresponding to the target user can be accurately identified, and appropriate sleep aid music can be accurately pushed to the target user according to the sleep state, and the sleep quality of the target user can be further improved.
  • step S2 that is, performing vector conversion on the voltage signal to obtain the characteristic vector includes the following steps:
  • S21 Filter the voltage signal based on the preset filter condition to obtain the target signal.
  • the voltage signal obtained in step S1 is filtered according to a preset filtering condition, and the voltage signal retained after the filtering process is determined as the target signal.
  • the preset filter condition refers to a condition set to filter the target signal according to the actual needs of the user.
  • the voltage signal is mainly expressed in the form of voltage value.
  • the preset filter condition is to filter voltage signals greater than the preset upper limit voltage value and voltage signals less than the preset lower limit voltage value; then each voltage signal obtained by S1 is respectively compared with the preset voltage signal.
  • the predetermined upper limit voltage value is compared with the lower limit voltage value, the voltage signals greater than the upper limit voltage value and less than the lower limit voltage value are filtered, and the voltage signal retained after the filtering process is determined as the target signal.
  • the target signal is imported into a preset processing library for vector conversion processing, and the converted feature vector is output.
  • the preset processing library refers to a database specially used for converting and processing a target signal into a feature vector.
  • the voltage signal is filtered according to the preset filter condition to obtain the target signal, and the target signal is subjected to vector conversion processing to obtain the feature vector.
  • step S3 that is, performing map conversion processing on the EEG signal to obtain a spectrogram includes the following steps:
  • S31 Perform frame and window processing on the EEG signal to obtain a characteristic signal.
  • the optimized EEG signal is divided into many short-term EEG signal segments, and each short-term EEG signal segment is called an analysis frame.
  • frames with a fixed length can be obtained by framing the EEG signal, that is, the total frame length of the EEG signal is divided by the preset frame length. If the last frame of the EEG signal cannot reach the preset frame length , The frame number is 0.
  • the preset frame length may specifically be 200, or it may be set according to the actual needs of the user, which is not limited here.
  • the EEG signal should be divided into frames so that each frame of the brain Electrical signals have short-term stability, that is, the framing signals after framing processing have short-term stability, so that short-term correlation analysis can be performed.
  • a window function is applied to the framing signal after framing. That is to say, the essence of windowing is the process of using a window function to multiply the framing signal after framing.
  • the characteristic signal is obtained by framing and windowing, so that the characteristic signal can better meet the periodicity requirements of the Fourier transform. So as to reduce the influence on the edge of the framed signal after framing.
  • the EEG signal is imported into the preset processing port for frame and window processing, and the characteristic signal after frame and window processing is obtained.
  • the preset processing port refers to a port used for framed and windowed processing of the EEG signal.
  • the framing in the preset processing port can be specifically processed by calling the enframe function in the voicebox tool, and the windowing can be specifically processed by formula (1).
  • the windowing function is as follows:
  • Q n is the characteristic signal
  • T[s(k)] is the framing signal
  • ⁇ (nk) is the window function
  • both n and k are constants.
  • the characteristic signal is intercepted by an analysis frame of length N, and the analysis frame is subjected to short-time Fourier transform to obtain the spectral characteristics of the characteristic signal of each frame.
  • the feature vector can be extracted for the spectral feature, that is, the spectrogram is extracted.
  • the extracted feature vectors include skewness and kurtosis, spectrum center, spectrum flux, spectrum roll-off, spectrum propagation, spectrum flatness, zero-crossing rate, Mel frequency cepstral coefficient (MFCC) and its first and second orders Differential components and so on.
  • MFCC Mel frequency cepstral coefficient
  • the characteristic signal is obtained by performing frame and window processing on the EEG signal, and finally the characteristic signal is transformed by short-time Fourier transform to obtain a spectrogram.
  • the characteristic signal is obtained by performing frame and window processing on the EEG signal, and finally the characteristic signal is transformed by short-time Fourier transform to obtain a spectrogram.
  • the method for automatically pushing sleep aid music further includes the following steps:
  • the training sample refers to sample data specially used for training the convolutional neural network model to obtain the sleep detection model.
  • the preset sample library refers to a database dedicated to storing training samples.
  • the training samples obtained in step S71 are imported into the convolutional neural network model for training, and the model that meets the requirements set by the user after training is determined as the sleep detection model.
  • the sleep detection model is obtained by obtaining training samples and using the training samples to train the convolutional neural network. In this way, accurate training of the sleep detection model is realized, and the accuracy of subsequent recognition using the sleep detection model is ensured.
  • step S72 importing training samples into the convolutional neural network for training, and obtaining a sleep detection model includes the following steps:
  • S721 Initialize the convolutional neural network model to obtain an initial model.
  • the model parameters of the convolutional neural network model are initialized through the server, and an initial parameter is assigned to the weight and bias of each network layer in the convolutional neural network model, so that the convolutional neural network
  • the network model can extract and calculate the characteristics of the training samples according to the initial parameters.
  • the weights and offsets are the model parameters used for the refraction transformation calculation of the input data in the network, so that the calculated output results of the network can be Consistent with the actual situation.
  • the person when a person receives the information, after the judgment and transmission of the human brain neurons, the person will get a certain result or cognition, that is, the process of obtaining cognition from the information, and
  • the training process of the convolutional neural network model is to optimize the weights and biases of the neuron connections in the network, so that the trained convolutional neural network model can achieve the recognition effect consistent with the real situation in the sleep state of the data to be recognized .
  • the server can arbitrarily obtain a weight as an initial parameter in the interval of [-0.30, +0.30], and set the initial parameter in an interval with a mean value of 0 and a smaller interval, which can improve the convergence speed of the model , In order to improve the efficiency of model construction.
  • S722 Import the training samples into the initial model, and calculate the forward output of the initial model.
  • the training samples are sequentially imported as input data into the input layer, convolution layer, pooling layer, fully connected layer, and output layer in the initial model for convolution operation, and finally the output result of the output layer is used as the forward output .
  • the input layer, convolution layer, pooling layer, fully connected layer, and output layer have preset convolution kernels.
  • the convolution kernel can be preset according to Perform convolution operation to get the corresponding output result.
  • S723 Calculate the prediction error between the forward output and the preset target value according to the forward output.
  • the prediction error between the forward output and the preset target value is calculated according to formula (2):
  • Loss is the prediction error
  • K is the preset target value
  • Ki is the forward output.
  • S724 According to the prediction error, use an error back propagation algorithm to adjust the initial parameters of each network layer in the initial model to obtain a sleep detection model.
  • the error back propagation algorithm is used to apportion the prediction error to all the units of each network layer, so as to obtain the error signals of all the units of each network layer, and then adjust the initial parameters of each network layer.
  • the initial parameter is only a parameter preset to facilitate the calculation of the initial model, so that there must be an error between the forward output obtained according to the training sample and the preset target value. This error information needs to be passed back to each layer in the initial model.
  • the layer network structure allows each layer of the network structure to adjust the preset initial parameters to obtain a sleep detection model with better recognition effect.
  • the error back propagation algorithm is used to adjust the initial parameters of each network layer of the initial model, and the error back propagation update is performed on each network layer of the initial model according to the output of each layer, and the updated network layer information is obtained.
  • Weights and biases use the updated weights and biases of each network layer to predict the training samples, and compare the forward output of the training samples with the preset target value to obtain the prediction error less than the preset threshold
  • the training sample is used as the number of accurate samples to predict, and the total number of training samples is counted to obtain the total number of training samples, and calculate the total error of the initial model according to formula (3):
  • W is the total error
  • M is the number of accurate prediction samples
  • N is the total number of training samples
  • the preset accuracy threshold is used to indicate the accuracy of the initial model's prediction of the training sample.
  • the specific accuracy threshold can be based on Actually need to be set, there is no restriction here.
  • step S723 if the total error of the current model is less than the preset accuracy threshold, return to step S723 to continue execution until the total error of the model is greater than the preset accuracy threshold, and use the current model as the sleep detection model.
  • the initial model is obtained by initializing the convolutional neural network model, the forward output of the initial model is calculated according to the training samples, and the prediction error between the forward output and the preset target value is calculated, and finally according to the preset error , Use the error back propagation algorithm to adjust the initial model to obtain the sleep detection model, so as to realize the training and optimization of the initial model, and improve the accuracy of the sleep detection model's recognition of training samples.
  • step S5 that is, based on a preset condition, obtaining the recommended type of sleep aid music corresponding to the sleep state includes the following steps:
  • S51 Obtain all sleep states in a preset time period from a preset recognition library.
  • the sleep state recognized by the sleep detection model since the sleep state recognized by the sleep detection model is based on the voltage signal and EEG signal collected at a preset time interval, the sleep state recognized by the sleep detection model within the preset time period may exist different.
  • the recognition time corresponding to each sleep state is obtained from the preset recognition library, and the obtained recognition time is respectively compared with the preset time period. If the recognition time is within the preset time period, Then extract the sleep state corresponding to the recognition time.
  • the preset recognition library refers to a database specifically used to store the sleep state and the recognition time corresponding to the sleep state.
  • the preset time period refers to the time range set according to the actual needs of the user, for example, it may be 22:00-24:00.
  • state Q1, state Q2, and state Q3 there are 3 sleep states in the preset recognition library: state Q1, state Q2, and state Q3, and their corresponding recognition times are: 22:00, 22:50, and 23:10, if the preset time period is :22:00 ⁇ 23:00, by comparing the recognition time 22:00, 22:50 and 23:10 with the preset time period respectively, it is obtained that the recognition time 22:00 and 22:50 are within the preset time period, Then the state Q1 and the state Q2 are extracted.
  • all sleep states within a preset time period are obtained according to step S51, and the same sleep state is accumulated and counted from all sleep states, that is, the accumulated count result corresponding to the same sleep state is calculated.
  • S53 The accumulated count results of different sleep states are compared, and the sleep state corresponding to the largest accumulated count result is selected as the target sleep state.
  • step S51 according to all sleep states obtained in step S51, the accumulated count results corresponding to different sleep states are compared, and the sleep state with the largest accumulated count result is selected as the target sleep state.
  • the sleep state when there is only one sleep state, the sleep state is directly used as the target sleep state.
  • the sleep state with the highest priority is selected as the target sleep state from the sleep states with the same cumulative count result.
  • the preset priority refers to the priority of setting the sleep state as the target sleep state according to the actual needs of the user.
  • state A is determined as the target sleep state. If there are 3 sleep states in all sleep states: state B, state C, and state D, the corresponding accumulated count results are: 10, 10, and 5, respectively, and the preset priority corresponding to state B, state C, and state D The levels are: the lowest, the next highest, and the highest. Compare the accumulation technical results corresponding to state B, state C, and state D, and get that the accumulated count result of state B and the accumulated count result of state C are the same and both are greater than state D, Since the preset priority of the state C is higher than that of the state B, the state C is determined as the target sleep state.
  • S54 Match the target sleep state with the description information in the preset recommendation library, and select and output the sleep aid music recommendation type corresponding to the successfully matched description information, where the preset recommendation library includes the description information and the sleep aid corresponding to the description information Recommended types of music.
  • the target sleep state obtained in step S53 is matched with the description information in the preset recommendation library. If the target sleep state is matched with the description information, it indicates that the matching is successful, and the sleep aid music corresponding to the description information is output. Recommended type. Among them, each target sleep state has unique description information matching it in the preset recommendation library.
  • description information refers to tag information specifically used for matching with the target sleep state, and different description information corresponds to different types of sleep aid music recommendation.
  • the cumulative count results corresponding to the same sleep state are calculated, the cumulative count results of different sleep states are compared, and the sleep state corresponding to the largest cumulative count result is selected
  • the target sleep state the recommended type of sleep aid music is determined according to the target sleep state.
  • the method for automatically pushing sleep aid music further includes the following steps:
  • S81 Acquire the playing time of the sleep aid music in real time from the preset playback library, where the sleep aid music includes the playback volume.
  • the playing time of the sleep aid music recommended in step S6 is directly obtained from the preset playing library in real time.
  • the preset play library refers to a database specially used to store the play time of sleep aid music
  • the sleep aid music includes the playback volume.
  • S82 Compare the playing time with the preset adjustment time, and if the playing time reaches the preset adjustment time, adjust the playing volume to the preset volume.
  • the playback time obtained in step S81 is compared with the preset adjustment time, and if the playback time reaches the preset adjustment time, the playback volume is adjusted to the preset volume.
  • the preset adjustment time refers to the time set for adjusting the playback volume of the sleep aid music.
  • the preset volume refers to the playback volume set according to the actual needs of the user.
  • the play time obtained in step S81 is compared with the preset stop time, and if the play time reaches the preset stop time, the sleep aid music is paused.
  • the preset stop time refers to the time set for pausing the sleep aid music.
  • the playback time of the sleep aid music is obtained, and the playback time is compared with the preset adjustment time. If the playback time reaches the preset adjustment time, the playback volume corresponding to the sleep aid music is adjusted to the preset volume. When the playing time reaches the preset stop time, the sleep aid music will be paused.
  • a device for automatically pushing sleep-aid music in one embodiment, includes a first acquisition module 81, a vector conversion module 82, a map conversion module 83, an identification module 84, a second acquisition module 85 and a pushing module 86.
  • the detailed description of each functional module is as follows:
  • the first obtaining module 81 is configured to obtain voltage signals and brain electrical signals collected by the target user during sleep at preset time intervals;
  • the vector conversion module 82 is used to perform vector conversion on the voltage signal to obtain a characteristic vector
  • the graph conversion module 83 is used to perform graph conversion processing on the brain electrical signal to obtain a spectrogram
  • the recognition module 84 is used to import the feature vector and the spectrogram into the pre-trained sleep detection model for recognition, and obtain the sleep state;
  • the second obtaining module 85 is configured to obtain the recommended type of sleep aid music corresponding to the sleep state based on a preset condition
  • the push module 86 is used to push the sleep aid music to the target user according to the recommended type of sleep aid music.
  • the vector conversion module 82 includes:
  • the filtering sub-module is used to filter the voltage signal based on the preset filtering conditions to obtain the target signal;
  • the feature vector acquisition sub-module is used to perform vector conversion processing on the target signal to obtain a feature vector.
  • the graph conversion module 83 includes:
  • the frame and window sub-module is used to perform frame and window processing on the EEG signal to obtain the characteristic signal;
  • the spectrogram acquisition sub-module is used to transform the characteristic signal by short-time Fourier transform to obtain a spectrogram.
  • the device for automatically pushing sleep aid music further includes:
  • the third acquisition module is used to acquire training samples from a preset sample library
  • the training module is used to import training samples into the convolutional neural network for training to obtain a sleep detection model.
  • the training module includes:
  • the initialization sub-module is used to initialize the convolutional neural network model to obtain the initial model
  • Import sub-module used to import training samples into the initial model, and calculate the forward output of the initial model
  • the prediction error calculation sub-module is used to calculate the prediction error between the forward output and the preset target value according to the forward output;
  • the sleep detection model determination sub-module is used to adjust the initial parameters of each network layer in the initial model by using the error back propagation algorithm according to the prediction error to obtain the sleep detection model.
  • the second obtaining module 85 includes:
  • the fourth acquiring sub-module is used to acquire all sleep states in a preset time period from the preset recognition library
  • the calculation sub-module is used to calculate the accumulated count result corresponding to the same sleep state based on all sleep states;
  • the target sleep state determination sub-module is used to compare the accumulated count results of different sleep states, and select the sleep state corresponding to the largest accumulated count result as the target sleep state;
  • the matching sub-module is used to match the target sleep state with the description information in the preset recommendation library, and select the type of sleep aid music recommendation corresponding to the successfully matched description information for output.
  • the preset recommendation library contains description information and description information The corresponding recommended type of sleep aid music.
  • the device for automatically pushing sleep aid music further includes:
  • the fifth acquisition module is used to acquire the playing time of the sleep aid music in real time from the preset playback library, where the sleep aid music includes the playback volume;
  • the volume adjustment module is used to compare the playback time with the preset adjustment time, and if the playback time reaches the preset adjustment time, the playback volume is adjusted to the preset volume;
  • the play pause module is used to compare the play time with the preset stop time. If the play time reaches the preset stop time, the sleep aid music will be paused.
  • FIG. 9 is a block diagram of the basic structure of a computer device 90 in an embodiment of this application.
  • the computer device 90 includes a memory 91, a processor 92, and a network interface 93 that are communicatively connected to each other through a system bus. It should be pointed out that FIG. 9 only shows a computer device 90 with components 91-93, but it should be understood that it is not required to implement all the illustrated components, and more or fewer components may be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculation and/or information processing in accordance with pre-set or stored instructions.
  • Its hardware includes, but is not limited to, a microprocessor, a dedicated Integrated Circuit (Application Specific Integrated Circuit, ASIC), Programmable Gate Array (Field-Programmable GateArray, FPGA), Digital Processor (Digital Signal Processor, DSP), embedded equipment, etc.
  • ASIC Application Specific Integrated Circuit
  • ASIC Application Specific Integrated Circuit
  • FPGA Field-Programmable GateArray
  • DSP Digital Processor
  • the computer device may be a computing device such as a desktop computer, a notebook, a palmtop computer, and a cloud server.
  • the computer device can interact with the user through a keyboard, a mouse, a remote control, a touch panel, or a voice control device.
  • the memory 91 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static memory Random access memory (SRAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), magnetic memory, magnetic disks, optical disks, etc.
  • the memory 91 may be an internal storage unit of the computer device 90, such as a hard disk or memory of the computer device 90.
  • the memory 91 may also be an external storage device of the computer device 90, for example, a plug-in hard disk equipped on the computer device 90, a smart memory card (Smart Media Card, SMC), and a secure digital (Secure Digital, SD) card, flash card (Flash Card), etc.
  • the memory 91 may also include both an internal storage unit of the computer device 90 and an external storage device thereof.
  • the memory 91 is generally used to store the operating system and various application software installed in the computer device 90, such as the program code of the method for automatically pushing sleep aid music, etc.
  • the memory 91 can also be used to temporarily store various types of data that have been output or will be output.
  • the processor 92 may be a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, a microprocessor, or other data processing chips.
  • the processor 92 is generally used to control the overall operation of the computer device 90.
  • the processor 92 is configured to run the program code or process data stored in the memory 91, for example, run the program code of the method for automatically pushing sleep aid music.
  • the network interface 93 may include a wireless network interface or a wired network interface, and the network interface 93 is generally used to establish a communication connection between the computer device 90 and other electronic devices.
  • This application also provides another implementation manner, that is, to provide a non-volatile computer-readable storage medium, the non-volatile computer-readable storage medium stores a sleep state information entry process, and the sleep state
  • the information entry process may be executed by at least one processor, so that the at least one processor executes the steps of any one of the aforementioned methods for automatically pushing sleep aid music.
  • the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM/RAM, magnetic disk, The optical disc) includes several instructions to enable a computer device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in the various embodiments of the present application.
  • a computer device which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.

Landscapes

  • Health & Medical Sciences (AREA)
  • Anesthesiology (AREA)
  • Pain & Pain Management (AREA)
  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Psychology (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Hematology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

一种自动推送助眠乐曲方法、装置、计算机设备（90）及存储介质，方法包括：以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号（S1）；对电压信号进行向量转换，得到特征向量（S2）；对脑电信号进行图转换处理，得到频谱图（S3）；将特征向量和频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态（S4）；基于预设条件，获取睡眠状态对应的助眠音乐推荐类型（S5）；根据助眠音乐推荐类型，为目标用户推送助眠乐曲（S6）。从而实现准确为目标用户推送助眠乐曲，进一步提高目标用户的睡眠质量。

Description

自动推送助眠乐曲方法、装置、计算机设备及存储介质
本申请要求于2019年9月18日提交中国专利局、申请号为2019108827195，发明名称为“自动推送助眠乐曲方法、装置、计算机设备及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及人工智能技术领域，尤其涉及一种自动推送助眠乐曲方法、装置、计算机设备及存储介质。
背景技术
在医疗助眠领域，通过检测人的睡眠状态，并检测到的睡眠状态向其推送相应的助眠音乐，能够有效促进人的睡眠质量。传统的检测人的睡眠状态的方法大多通过检测脑电信号来实现，但发明人意识到监测脑电信号的设备太庞大不便于家庭使用，而且需要贴在人脑从而影响睡眠，导致提取到的脑电信号的不准确，进一步影响对人的睡眠状态的检测准确性，从而无法准确为人推荐助眠音乐，降低人的睡眠质量。
发明内容
本申请实施例提供一种自动推送助眠乐曲方法、装置、计算机设备及存储介质，以解决无法准确为人推荐助眠音乐，降低人的睡眠质量的问题。
一种自动推送助眠乐曲方法，包括：
以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号；
对所述电压信号进行向量转换，得到特征向量；
对所述脑电信号进行图转换处理，得到频谱图；
将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态；
基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型；
根据所述助眠音乐推荐类型，为所述目标用户推送助眠乐曲。
一种自动推送助眠乐曲装置，包括：
第一获取模块，用于以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号；
向量转换模块，用于对所述电压信号进行向量转换，得到特征向量；
图转换模块，用于对所述脑电信号进行图转换处理，得到频谱图；
识别模块，用于将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态；
第二获取模块，用于基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型；
推送模块，用于根据所述助眠音乐推荐类型，为所述目标用户推送助眠乐曲。
一种计算机设备，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机可读指令，所述处理器执行所述计算机可读指令时实现下述自动推送助眠乐曲方法的步骤：
以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号；
对所述电压信号进行向量转换，得到特征向量；
对所述脑电信号进行图转换处理，得到频谱图；
将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态；
基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型；
根据所述助眠音乐推荐类型，为所述目标用户推送助眠乐曲。
一种非易失性的计算机可读存储介质，所述非易失性的计算机可读存储介质存储有计算机可读指令，所述计算机可读指令被处理器执行时实现下述自动推送助眠乐曲方法的步骤：
以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号；
对所述电压信号进行向量转换，得到特征向量；
对所述脑电信号进行图转换处理，得到频谱图；
将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态；
基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型；
根据所述助眠音乐推荐类型，为所述目标用户推送助眠乐曲。
上述自动推送助眠乐曲方法、装置、计算机设备及存储介质，以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号，将电压信号和脑电信号分别转换为特征向量和频谱图，并将特征向量和频谱图导入到预先训练好的睡眠检测模型中进行识别，得到睡眠状态，根据睡眠状态确定对应的助眠音乐推荐类型，并为目标用户推送助眠音乐推荐类型对应的助眠乐曲。通过利用睡眠检测模型能够准确识别目标用户对应的睡眠状态，根据睡眠状态能够准确为目标用户推送合适的助眠音乐，进一步提高目标用户的睡眠质量。
附图说明
为了更清楚地说明本申请实施例的技术方案，下面将对本申请实施例的描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1是本申请实施例提供的自动推送助眠乐曲方法的流程图；
图2是本申请实施例提供的自动推送助眠乐曲方法中步骤S2的流程图；
图3是本申请实施例提供的自动推送助眠乐曲方法中步骤S3的流程图；
图4是本申请实施例提供的自动推送助眠乐曲方法中利用训练样本训练得到睡眠检测模型的流程图；
图5是本申请实施例提供的自动推送助眠乐曲方法中步骤S72的流程图；
图6是本申请实施例提供的自动推送助眠乐曲方法中步骤S5的流程图；
图7是本申请实施例提供的自动推送助眠乐曲方法中根据助眠乐曲的播放时间调节播放音量或暂停播放的流程图；
图8是本申请实施例提供的自动推送助眠乐曲装置的示意图；
图9是本申请实施例提供的计算机设备的基本机构框图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本申请提供的自动推送助眠乐曲方法应用于服务端，服务端具体可以用独立的服务器或者多个服务器组成的服务器集群实现。在一实施例中，如图1所示，提供一种自动推送助眠乐曲方法，包括如下步骤：
S1：以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号。
在本申请实施例中，电压信号是通过专门用于采集电压信号的传感器进行采集，当目标用户在带有该传感器的地方睡觉时，电压信号将会实时保存到预设数据库中；脑电信号是通过用户佩戴的专门用于采集脑电信号的手环进行采集，当目标用户佩戴该手环时，脑电信号将会实时保存到预设数据库中。
具体地，根据预设时间间隔，从预设数据库中直接获取目标用户在睡眠时采集到的电压信号和脑电信号。若预设时间间隔为1分钟，则每隔1分钟从预设数据库中获取电压信号和脑电信号。
其中，预设数据库是指专门用于存储目标用户在睡眠时采集到的电压信号和脑电信号的数据库。
预设时间间隔的取值具体可以是1分钟，也可以是5分钟，此处不做限制。
需要说明的是，传感器是一个嵌入了压敏电阻的床垫，其床垫内部舍弃了一系列电路，并将压敏电阻分布在床垫的不同位置，在测量数据的过程中，床垫将会实时返回每个压敏电阻上的电压值，其电压值会根据患者在床上的产生的压力发生变化。
S2：对电压信号进行向量转换，得到特征向量。
在本申请实施例中，通过将步骤S1获取的电压信号导入到预设向量转换端口中进行向量转换处理，得到向量转换处理后的特征向量。其中，预设向量转换端口是指专门用于将电压信号转换成特征向量的处理端口。
S3：对脑电信号进行图转换处理，得到频谱图。
在本申请实施例中，通过将步骤S1获取的脑电信号导入到预设图转换端口中进行图转换处理，得到图转换处理后的频谱图。其中，预设图转换端口是指专门用于将脑电信号转换成频谱图的处理端口。
S4：将特征向量和频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态。
具体地，将获取到的特征向量和频谱图输入到预先训练好的睡眠监测模型中进行识别，睡眠监测模型将根据输入的特征向量和频谱图直接判断出对应的睡眠状态，并将睡眠状态进行输出。
S5：基于预设条件，获取睡眠状态对应的助眠音乐推荐类型。
在本申请实施例中，根据步骤S4得到的睡眠状态与预设条件，从预设推荐库获取睡眠状态对应的助眠音乐推荐类型。其中，预设条件是指根据用户实际需求进行设定的规则，其具体可以是根据连续识别到的相同的睡眠状态确定助眠音乐推荐类型。
预设推荐库是指专门用于存储不同的描述信息及描述信息对应的助眠音乐推荐类型的数据库。
S6：根据助眠音乐推荐类型，为目标用户推送助眠乐曲。
具体地，根据助眠音乐推荐类型，从预设乐曲库中随机选取该助眠音乐推荐类型下的任何一首助眠乐曲，并将该助眠乐曲输出到预设播放端口为目标用户进行播放。其中，预设播放端口是指专门用于为目标用户播放助眠乐曲的处理端口。
本实施例中，以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号，将电压信号和脑电信号分别转换为特征向量和频谱图，并将特征向量和频谱图导入到预先训练好的睡眠检测模型中进行识别，得到睡眠状态，根据睡眠状态确定对应的助眠音乐推荐类型，并为目标用户推送助眠音乐推荐类型对应的助眠乐曲。通过利用睡眠检测模型能够准确识别目标用户对应的睡眠状态，根据睡眠状态能够准确为目标用户推送合适的助眠音乐，进一步提高目标用户的睡眠质量。
在一实施例中，如图2所示，步骤S2中，即对电压信号进行向量转换，得到特征向量包括如下步骤：
S21：基于预设过滤条件，对电压信号进行过滤，得到目标信号。
在本申请实施例中，针对步骤S1获取到的电压信号，按照预设过滤条件对电压信号进行过滤处理，并将过滤处理后保留的电压信号确定为目标信号。其中，预设过滤条件是指根据用户实际需求进行设定用于过滤目标信号的条件。
需要说明的是，电压信号主要是以电压值的形式进行表示。
例如，若预设过滤条件为将大于预先设定的上限电压值的电压信号，以及小于预先设定的下限电压值的电压信号进行过滤；则将S1获取到的每个电压信号分别与预先设定的上限电压值和下限电压值进行比较，将大于上限电压值和小于下限电压值的电压信号进行过滤处理，并将过滤处理后保留的电压信号确定为目标信号。
S22：将目标信号进行向量转换处理，得到特征向量。
具体地，将目标信号导入到预设处理库中进行向量转换处理，输出转换处理后的特征向量。其中，预设处理库是指专门用于将目标信号转换处理成特征向量的数据库。
本实施例中，根据预设过滤条件对电压信号进行过滤，得到目标信号，并对目标信号进行向量转换处理，得到特征向量。通过先对电压信号进行过滤的方式，可以有效排除异常的电压信号，提高特征向量获取的准确性，保证后续利用特征向量导入到睡眠监测模型进行识别的准确性，进一步提高后续助眠乐曲推送的准确性。
在一实施例中，如图3所示，步骤S3中，即对脑电信号进行图转换处理，得到频谱图包括如下步骤：
S31：对脑电信号进行分帧加窗处理，得到特征信号。
在本申请实施例中，将经过优化处理后的脑电信号划分为很多短时的脑电信号段，每个短时的脑电信号段称为一个分析帧。这样，通过对脑电信号进行分帧处理能够得到具有固定长度的帧，也就是将脑电信号的总帧长度除以预设帧长度，若脑电信号的最后一帧无法达到预设帧长度，则该帧数为0。其中，预设帧长度具体可以是200，也可以根据用户的实际需求进行设置，此处不做限制。
需要说明的是，由于音色特征在一个很短的时间段内可以认为具有相对稳定的特征即短时特征，具有短时平稳性特点，因此要对脑电信号进行分帧，使得每一帧脑电信号都具有短时平稳性，即分帧处理后的分帧信号都具有短时平稳性，从而进行短时相关分析。
但对脑电信号进行分帧处理后，存在分帧信号泄露问题，例如，当频谱出现拖尾的情况时，说明分帧信号泄漏严重。为了减少分帧信号泄漏问题，给分帧后的分帧信号施加一个窗函数。也就是说加窗的实质是使用一个窗函数与分帧后的分帧信号作乘积的过程，通过分帧加窗得到特征信号，使特征信号更好地满足傅里叶变换的周期性要求，从而减轻分帧后对分帧信号边缘的影响。
具体地，将脑电信号导入到预设处理端口中进行分帧加窗处理，得到分帧加窗处理后 的特征信号。其中，预设处理端口是指用于对脑电信号进行分帧加窗处理的端口。
进一步地，预设处理端口中分帧具体可以是调用voicebox工具中的enframe函数进行处理，加窗具体可以运用公式(1)进行处理。
其中，加窗函数如下所示：
Figure PCTCN2020099515-appb-000001
其中，Q n为特征信号，T[s(k)]为分帧信号，ω(n-k)为窗函数，n、k都为常数。
S32：利用短时傅里叶变换对特征信号进行变换处理，得到频谱图。
具体地，通过使用海明窗函数(Hamming)h＝hamming(N)对特征信号进行截取长度为N的分析帧，并将分析帧通过短时傅里叶变换得到每帧特征信号的频谱特性，即可对该频谱特征进行提取特征向量，即提取频谱图。
其中，提取的特征向量包括偏度和峰度、谱中心、谱通量、谱滚降、谱传播、谱平坦度、过零率、Mel频率倒谱系数(MFCC)及其一阶和二阶差分量等等。
本实施例中，通过对脑电信号进行分帧加窗处理得到特征信号，最后利用短时傅里叶变换对特征信号进行变换处理，得到频谱图。从而实现准确将脑电信号转换成频谱图，保证后续利用频谱图导入到睡眠监测模型进行识别的准确性，进一步提高后续助眠乐曲推送的准确性。
在一实施例中，如图4所示，步骤S3之后，步骤S4之前，该自动推送助眠乐曲方法还包括如下步骤：
S71：从预设样本库中获取训练样本。
在本申请实施例中，训练样本是指专门用于训练卷积神经网络模型以得到睡眠检测模型的样本数据。通过直接从预设样本库中获取训练样本，其中，预设样本库是指专门用于存储训练样本的数据库。
S72：将训练样本导入到卷积神经网络中进行训练，得到睡眠检测模型。
具体地，将步骤S71得到的训练样本导入到卷积神经网络模型中进行训练，将训练后达到用户设定要求的模型确定为睡眠检测模型。
本实施例中，通过获取训练样本，并利用训练样本对卷积神经网络中进行训练，得到睡眠检测模型。从而实现对睡眠检测模型的准确训练，保证后续利用睡眠检测模型进行识别的准确性。
在一实施例中，如图5所示，步骤S72中，即将训练样本导入到卷积神经网络中进行训练，得到睡眠检测模型包括如下步骤：
S721：初始化卷积神经网络模型，得到初始模型。
在本申请实施例中，通过服务端对卷积神经网络模型的模型参数进行初始化处理，为卷积神经网络模型中的各个网络层的权值和偏置均赋予一个初始参数，使得卷积神经网络模型能够根据初始参数对训练样本进行特征的提取和计算，其中，权值和偏置是用于对输入的数据在网络中进行的折射变换计算的模型参数，使得网络经过计算输出的结果能够与实际情况相符。
可以理解地，以人接收信息为例，当人在接收信息后，经过人脑神经元的判断、传递后，人会得到某个结果或者认知，即从信息中获取认知的过程，而对卷积神经网络模型的训练过程就是优化网络中的神经元连接的权值和偏置，使得训练后的卷积神经网络模型对待识别的数据的睡眠状态，能够达到与真实情况相符的识别效果。
可选地，服务端可以在[-0.30，+0.30]的区间内，任意获取一个权值作为初始参数，将初始参数设置在一个均值为0并且较小的区间内，能够提高模型的收敛速度，以提高模型 的构建效率。
S722：将训练样本导入到初始模型中，计算初始模型的前向输出。
具体地，将训练样本作为输入数据依次导入到初始模型中的输入层、卷积层、池化层、全连接层和输出层中进行卷积操作，最后将输出层的输出结果作为前向输出。其中，输入层、卷积层、池化层、全连接层和输出层都有预先设置好的卷积核，通过将训练样本作为输入数据导入到每一层中能够根据预先设置好卷积核进行卷积操作，得到对应的输出结果。
S723：根据前向输出，计算前向输出与预设目标值之间的预测误差。
在本申请实施例中，根据步骤S722得到的前向输出与预设目标值，按照公式(2)计算前向输出与预设目标值之间的预测误差：
Loss＝K—Ki  公式(2)
其中，Loss为预测误差，K为预设目标值，Ki为前向输出。
S724：根据预测误差，使用误差反向传播算法对初始模型中各个网络层的初始参数进行调整，得到睡眠检测模型。
在本申请实施例中，误差反向传播算法是用于将预测误差分摊给各个网络层的所有单元，从而获得各个网络层的所有单元的误差信号，进而调整各个网络层的初始参数。初始参数只是为了方便初始模型的运算预设的一个参数，使得根据训练样本获得的前向输出与预设目标值之间必然存在误差，需要将这个误差信息逐层回传给初始模型中的各层网络结构，让每一层网络结构对预设的初始参数进行调整，才能获得识别效果更好的睡眠检测模型。
具体地，根据预测误差，使用误差反向传播算法对初始模型各个网络层的初始参数进行调整，根据各层的输出对初始模型各个网络层进行误差反传更新，获取更新后的各个网络层的权值和偏置，使用更新后的各个网络层的权值和偏置，对训练样本进行预测，并将训练样本的前向输出与预设目标值进行对比，获取预测误差小于预设阈值的训练样本作为预测准确样本数，对训练样本的总数进行统计，得到训练样本总数，并按照公式(3)计算初始模型的总误差：
Figure PCTCN2020099515-appb-000002
其中，W为总误差，M为预测准确样本数，N为训练样本总数；
若当前模型的总误差大于预设精度阈值，则将当前调整后的模型作为睡眠检测模型，其中，预设精度阈值是用于表示初始模型对训练样本的预测准确率，具体的精度阈值可以根据实际需要进行设置，此处不做限制。
需要说明的是，若当前模型的总误差小于预设精度阈值，则返回步骤S723继续执行，直到模型的总误差大于预设精度阈值，并将当前模型作为睡眠检测模型。
本实施例中，通过对卷积神经网络模型进行初始化得到初始模型，根据训练样本计算初始模型的前向输出，再计算前向输出与预设目标值之间的预测误差，最后根据预设误差，使用误差反向传播算法对初始模型进行调整得到睡眠检测模型，从而实现对初始模型的训练调优，提高睡眠检测模型对训练样本的识别准确率。
在一实施例中，如图6所示，步骤S5中，即基于预设条件，获取睡眠状态对应的助眠音乐推荐类型包括如下步骤：
S51：从预设识别库中获取预设时间段内的所有睡眠状态。
在本申请实施例中，由于睡眠检测模型识别到的睡眠状态是针对预设时间间隔采集到的电压信号和脑电信号，故在预设时间段内利用睡眠检测模型识别得到的睡眠状态可能存在不同。
具体地，根据预设时间段，从预设识别库中获取每个睡眠状态对应的识别时间，将获取到的识别时间分别与预设时间段进行比较，若识别时间在预设时间段内，则提取该识别时间对应的睡眠状态。
其中，预设识别库是指专门用于存储睡眠状态及睡眠状态对应的识别时间的数据库。
预设时间段是指根据用户实际需求进行设置的时间范围，例如具体可以是22：00～24：00。
例如，预设识别库中存在3个睡眠状态分别为：状态Q1、状态Q2和状态Q3，其对应的识别时间分别为：22：00、22：50和23：10，若预设时间段为：22：00～23：00，通过将识别时间22：00、22：50和23：10分别与预设时间段进行比较，得到识别时间22：00和22：50在预设时间段内，则对状态Q1和状态Q2进行提取。
S52：基于所有睡眠状态，计算相同的睡眠状态对应的累加计数结果。
在本申请实施例中，根据步骤S51得到预设时间段内的所有睡眠状态，从所有睡眠状态中对相同的睡眠状态进行累加计数，即计算相同的睡眠状态对应的累加计数结果。
需要说明的是，当存在所有睡眠状态都不相同时，每种睡眠状态对应的累加计数结果均为1。
S53：将不同的睡眠状态的累加计数结果进行比较，选取最大的累加计数结果对应的睡眠状态作为目标睡眠状态。
在本申请实施例中，根据步骤S51得到的所有睡眠状态，从中对不同的睡眠状态对应的累加计数结果进行比较，并选取累加计数结果最大的睡眠状态作为目标睡眠状态。
需要说明的是，当只有一种睡眠状态时，直接将该睡眠状态作为目标睡眠状态。当出现两种或两种以上的睡眠状态对应的累加计数结果相同时，则按照预设优先级，从累加计数结果相同的睡眠状态种选取优先级最高的睡眠状态作为目标睡眠状态。
其中，预设优先级是指根据用户实际需求对睡眠状态进行设定用于作为目标睡眠状态的优先等级。
例如，当所有睡眠状态都为状态A时，将状态A确定为目标睡眠状态。若所有睡眠状态中存在3种睡眠状态分别为：状态B、状态C和状态D，其对应的累加计数结果分别为：10、10和5，状态B、状态C和状态D对应的预设优先级分别为：最低、次高和最高，将状态B、状态C和状态D分别对应的累加技术结果进行比较，得到状态B的累加计数结果和状态C的累加计数结果相同且均大于状态D，由于状态C的预设优先级高于状态B，故将状态C确定为目标睡眠状态。
S54：将目标睡眠状态与预设推荐库中的描述信息进行匹配，选取匹配成功的描述信息对应的助眠音乐推荐类型进行输出，其中，预设推荐库包含描述信息及描述信息对应的助眠音乐推荐类型。
具体地，将步骤S53获取到的目标睡眠状态与预设推荐库中的描述信息进行匹配，若匹配到目标睡眠状态与描述信息相同，则表示匹配成功，并输出该描述信息对应的助眠音乐推荐类型。其中，每种目标睡眠状态在预设推荐库中都有与其相匹配的唯一的描述信息。
需要说明的是，描述信息是指专门用于与目标睡眠状态进行匹配的标签信息，且不同的描述信息对应不同的助眠音乐推荐类型。
本实施例中，根据获取预设时间段内的所有睡眠状态，计算相同的睡眠状态对应的累加计数结果，将不同的睡眠状态的累加计数结果进行比较，选取最大的累加计数结果对应的睡眠状态作为目标睡眠状态，并根据目标睡眠状态确定助眠音乐推荐类型。通过确定目标睡眠状态的方式能够准确判断目标用户的睡眠情况，保证根据目标睡眠状态确定助眠音乐推荐类型的准确性，进一步提供后续根据助眠音乐推荐类型推送推送助眠乐曲的准确性。
在一实施例中，如图7所示，步骤S6之后，该自动推送助眠乐曲方法还包括如下步骤：
S81：从预设播放库中实时获取助眠乐曲的播放时间，其中，助眠乐曲包含播放音量。
在本申请实施例中，通过直接从预设播放库中实时获取步骤S6推荐的助眠乐曲的播放时间。其中，预设播放库是指专门用于存储助眠乐曲的播放时间的数据库，且助眠乐曲包含播放音量。
需要说明的是，当助眠乐曲停止播放时，将该助眠乐曲的播放时间从预设播放库中进行删除处理。
S82：将播放时间与预设调节时间进行比较，若播放时间达到预设调节时间，则将播放音量调至预设音量。
在本申请实施例中，将步骤S81得到的播放时间与预设调节时间进行比较，若该播放时间达到预设调节时间，则将播放音量调至预设音量。其中，预设调节时间是指设定用于调节助眠乐曲的播放音量的时间。预设音量是指根据用户实际需求设定的播放音量。
S83：将播放时间与预设停止时间进行比较，若播放时间达到预设停止时间，则暂停播放助眠乐曲。
具体地，将步骤S81得到的播放时间与预设停止时间进行比较，若该播放时间达到预设停止时间，则暂停播放助眠乐曲。其中，预设停止时间是指设定用于暂停播放助眠乐曲的时间。
本实施例中，通过获取助眠乐曲的播放时间，将播放时间与预设调节时间进行比较，若播放时间达到预设调节时间，则将助眠乐曲对应的播放音量调至预设音量，若播放时间达到预设停止时间，则暂停播放助眠乐曲。通过对助眠乐曲音量的调节以及播放暂停的方式，能够避免目标用户在深睡的情况下继续听助眠乐曲造成噪声干扰，进而有效提高目标用户的睡眠质量。
应理解，上述实施例中各步骤的序号的大小并不意味着执行顺序的先后，各过程的执行顺序应以其功能和内在逻辑确定，而不应对本申请实施例的实施过程构成任何限定。
在一实施例中，提供一种自动推送助眠乐曲装置，该自动推送助眠乐曲装置与上述实施例中自动推送助眠乐曲方法一一对应。如图8所示，该自动推送助眠乐曲装置包括第一获取模块81，向量转换模块82，图转换模块83，识别模块84，第二获取模块85和推送模块86。各功能模块详细说明如下：
第一获取模块81，用于以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号；
向量转换模块82，用于对电压信号进行向量转换，得到特征向量；
图转换模块83，用于对脑电信号进行图转换处理，得到频谱图；
识别模块84，用于将特征向量和频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态；
第二获取模块85，用于基于预设条件，获取睡眠状态对应的助眠音乐推荐类型；
推送模块86，用于根据助眠音乐推荐类型，为目标用户推送助眠乐曲。
进一步地，向量转换模块82包括：
过滤子模块，用于基于预设过滤条件，对电压信号进行过滤，得到目标信号；
特征向量获取子模块，用于将目标信号进行向量转换处理，得到特征向量。
进一步地，图转换模块83包括：
分帧加窗子模块，用于对脑电信号进行分帧加窗处理，得到特征信号；
频谱图获取子模块，用于利用短时傅里叶变换对特征信号进行变换处理，得到频谱图。
进一步地，该自动推送助眠乐曲装置还包括：
第三获取模块，用于从预设样本库中获取训练样本；
训练模块，用于将训练样本导入到卷积神经网络中进行训练，得到睡眠检测模型。
进一步地，训练模块包括：
初始化子模块，用于初始化卷积神经网络模型，得到初始模型；
导入子模块，用于将训练样本导入到初始模型中，计算初始模型的前向输出；
预测误差计算子模块，用于根据前向输出，计算前向输出与预设目标值之间的预测误差；
睡眠检测模型确定子模块，用于根据预测误差，使用误差反向传播算法对初始模型中各个网络层的初始参数进行调整，得到睡眠检测模型。
进一步地，第二获取模块85包括：
第四获取子模块，用于从预设识别库中获取预设时间段内的所有睡眠状态；
计算子模块，用于基于所有睡眠状态，计算相同的睡眠状态对应的累加计数结果；
目标睡眠状态确定子模块，用于将不同的睡眠状态的累加计数结果进行比较，选取最大的累加计数结果对应的睡眠状态作为目标睡眠状态；
匹配子模块，用于将目标睡眠状态与预设推荐库中的描述信息进行匹配，选取匹配成功的描述信息对应的助眠音乐推荐类型进行输出，其中，预设推荐库包含描述信息及描述信息对应的助眠音乐推荐类型。
进一步地，该自动推送助眠乐曲装置还包括：
第五获取模块，用于从预设播放库中实时获取助眠乐曲的播放时间，其中，助眠乐曲包含播放音量；
音量调节模块，用于将播放时间与预设调节时间进行比较，若播放时间达到预设调节时间，则将播放音量调至预设音量；
播放暂停模块，用于将播放时间与预设停止时间进行比较，若播放时间达到预设停止时间，则暂停播放助眠乐曲。
本申请的一些实施例公开了计算机设备。具体请参阅图9，为本申请的一实施例中计算机设备90基本结构框图。
如图9中所示意的，所述计算机设备90包括通过系统总线相互通信连接存储器91、处理器92、网络接口93。需要指出的是，图9中仅示出了具有组件91-93的计算机设备90，但是应理解的是，并不要求实施所有示出的组件，可以替代的实施更多或者更少的组件。其中，本技术领域技术人员可以理解，这里的计算机设备是一种能够按照事先设定或存储的指令，自动进行数值计算和/或信息处理的设备，其硬件包括但不限于微处理器、专用集成电路(Application Specific Integrated Circuit，ASIC)、可编程门阵列(Field－Programmable GateArray，FPGA)、数字处理器(Digital Signal Processor，DSP)、嵌入式设备等。
所述计算机设备可以是桌上型计算机、笔记本、掌上电脑及云端服务器等计算设备。所述计算机设备可以与用户通过键盘、鼠标、遥控器、触摸板或声控设备等方式进行人机交互。
所述存储器91至少包括一种类型的可读存储介质，所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如，SD或DX存储器等)、随机访问存储器(RAM)、静态随机访问存储器(SRAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、可编程只读存储器(PROM)、磁性存储器、磁盘、光盘等。在一些实施例中，所述存储器91可以是所述计算机设备90的内部存储单元，例如该计算机设备90的硬盘或内存。在另一些实施例中，所述存储器91也可以是所述计算机设备90的外部存储设备，例如该计算机设备90上配备的插接式硬盘，智能存储卡(Smart Media Card,SMC)，安全数字(Secure Digital,SD)卡，闪存卡(Flash Card)等。当然，所述存储器91还可以既包括所述计算机设备90的内部存储单元也包括其外部存储设备。本实施例中，所述存储器91通常用于存储安装于所述计算机设备90的操作系统和各类应用软件，例如所述自动推送助眠乐曲方法的程序代码等。此外，所述存储器91还可以用于暂时地存储已经输出或者将要输出的各类数据。
所述处理器92在一些实施例中可以是中央处理器(Central Processing Unit，CPU)、 控制器、微控制器、微处理器、或其他数据处理芯片。该处理器92通常用于控制所述计算机设备90的总体操作。本实施例中，所述处理器92用于运行所述存储器91中存储的程序代码或者处理数据，例如运行所述自动推送助眠乐曲方法的程序代码。
所述网络接口93可包括无线网络接口或有线网络接口，该网络接口93通常用于在所述计算机设备90与其他电子设备之间建立通信连接。
本申请还提供了另一种实施方式，即提供一种非易失性的计算机可读存储介质，所述非易失性的计算机可读存储介质存储有睡眠状态信息录入流程，所述睡眠状态信息录入流程可被至少一个处理器执行，以使所述至少一个处理器执行上述任意一种自动推送助眠乐曲方法的步骤。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台计算机设备(可以是手机，计算机，服务器，空调器，或者网络设备等)执行本申请各个实施例所述的方法。
最后应说明的是，显然以上所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例，附图中给出了本申请的较佳实施例，但并不限制本申请的专利范围。本申请可以以许多不同的形式来实现，相反地，提供这些实施例的目的是使对本申请的公开内容的理解更加透彻全面。尽管参照前述实施例对本申请进行了详细的说明，对于本领域的技术人员来而言，其依然可以对前述各具体实施方式所记载的技术方案进行修改，或者对其中部分技术特征进行等效替换。凡是利用本申请说明书及附图内容所做的等效结构，直接或间接运用在其他相关的技术领域，均同理在本申请专利保护范围之内。

Claims (20)

  1. 一种自动推送助眠乐曲方法，其中，所述自动推送助眠乐曲方法包括：
    以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号；
    对所述电压信号进行向量转换，得到特征向量；
    对所述脑电信号进行图转换处理，得到频谱图；
    将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态；
    基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型；
    根据所述助眠音乐推荐类型，为所述目标用户推送助眠乐曲。
  2. 如权利要求1所述的自动推送助眠乐曲方法，其中，所述对所述电压信号进行向量转换，得到特征向量的步骤包括：
    基于预设过滤条件，对所述电压信号进行过滤，得到目标信号；
    将所述目标信号进行向量转换处理，得到特征向量。
  3. 如权利要求1所述的自动推送助眠乐曲方法，其中，所述对所述脑电信号进行图转换处理，得到频谱图的步骤包括：
    对所述脑电信号进行分帧加窗处理，得到特征信号；
    利用短时傅里叶变换对所述特征信号进行变换处理，得到所述频谱图。
  4. 如权利要求1所述的自动推送助眠乐曲方法，其中，所述对所述脑电信号进行图转换处理，得到频谱图的步骤之后，所述将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态的步骤之前，所述自动推送助眠乐曲方法还包括：
    从预设样本库中获取训练样本；
    将所述训练样本导入到卷积神经网络中进行训练，得到所述睡眠检测模型。
  5. 如权利要求4所述的自动推送助眠乐曲方法，其中，所述将所述训练样本导入到卷积神经网络中进行训练，得到所述睡眠检测模型的步骤包括：
    初始化所述卷积神经网络模型，得到初始模型；
    将所述训练样本导入到所述初始模型中，计算所述初始模型的前向输出；
    根据所述前向输出，计算所述前向输出与预设目标值之间的预测误差；
    根据所述预测误差，使用误差反向传播算法对所述初始模型中各个网络层的初始参数进行调整，得到所述睡眠检测模型。
  6. 如权利要求1所述的自动推送助眠乐曲方法，其中，所述基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型的步骤包括：
    从预设识别库中获取预设时间段内的所有睡眠状态；
    基于所述所有睡眠状态，计算相同的睡眠状态对应的累加计数结果；
    将不同的所述睡眠状态的累加计数结果进行比较，选取最大的所述累加计数结果对应的睡眠状态作为目标睡眠状态；
    将所述目标睡眠状态与预设推荐库中的描述信息进行匹配，选取匹配成功的描述信息对应的助眠音乐推荐类型进行输出，其中，所述预设推荐库包含描述信息及描述信息对应的助眠音乐推荐类型。
  7. 如权利要求1所述的自动推送助眠乐曲方法，其中，所述根据所述助眠音乐推荐类型，为所述目标用户推送助眠乐曲的步骤之后，所述自动推送助眠乐曲方法还包括：
    从预设播放库中实时获取所述助眠乐曲的播放时间，其中，所述助眠乐曲包含播放音量；
    将所述播放时间与预设调节时间进行比较，若所述播放时间达到预设调节时间，则将所述播放音量调至预设音量；
    将所述播放时间与预设停止时间进行比较，若所述播放时间达到预设停止时间，则暂停播放所述助眠乐曲。
  8. 一种自动推送助眠乐曲装置，其中，所述自动推送助眠乐曲装置包括：
    第一获取模块，用于以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号；
    向量转换模块，用于对所述电压信号进行向量转换，得到特征向量；
    图转换模块，用于对所述脑电信号进行图转换处理，得到频谱图；
    识别模块，用于将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态；
    第二获取模块，用于基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型；
    推送模块，用于根据所述助眠音乐推荐类型，为所述目标用户推送助眠乐曲。
  9. 一种计算机设备，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机可读指令，其中，所述处理器执行所述计算机可读指令时实现如下步骤：
    以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号；
    对所述电压信号进行向量转换，得到特征向量；
    对所述脑电信号进行图转换处理，得到频谱图；
    将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态；
    基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型；
    根据所述助眠音乐推荐类型，为所述目标用户推送助眠乐曲。
  10. 如权利要求9所述的计算机设备，其中，所述对所述电压信号进行向量转换，得到特征向量的步骤包括：
    基于预设过滤条件，对所述电压信号进行过滤，得到目标信号；
    将所述目标信号进行向量转换处理，得到特征向量。
  11. 如权利要求9所述的计算机设备，其中，所述对所述脑电信号进行图转换处理，得到频谱图的步骤包括：
    对所述脑电信号进行分帧加窗处理，得到特征信号；
    利用短时傅里叶变换对所述特征信号进行变换处理，得到所述频谱图。
  12. 如权利要求9所述的计算机设备，其中，所述对所述脑电信号进行图转换处理，得到频谱图的步骤之后，所述将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态的步骤之前，所述处理器执行所述计算机可读指令时还包括实现如下步骤：
    从预设样本库中获取训练样本；
    将所述训练样本导入到卷积神经网络中进行训练，得到所述睡眠检测模型。
  13. 如权利要求12所述的计算机设备，其中，所述将所述训练样本导入到卷积神经网络中进行训练，得到所述睡眠检测模型的步骤包括：
    初始化所述卷积神经网络模型，得到初始模型；
    将所述训练样本导入到所述初始模型中，计算所述初始模型的前向输出；
    根据所述前向输出，计算所述前向输出与预设目标值之间的预测误差；
    根据所述预测误差，使用误差反向传播算法对所述初始模型中各个网络层的初始参数进行调整，得到所述睡眠检测模型。
  14. 如权利要求9所述的计算机设备，其中，所述基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型的步骤包括：
    从预设识别库中获取预设时间段内的所有睡眠状态；
    基于所述所有睡眠状态，计算相同的睡眠状态对应的累加计数结果；
    将不同的所述睡眠状态的累加计数结果进行比较，选取最大的所述累加计数结果对应的睡眠状态作为目标睡眠状态；
    将所述目标睡眠状态与预设推荐库中的描述信息进行匹配，选取匹配成功的描述信息对应的助眠音乐推荐类型进行输出，其中，所述预设推荐库包含描述信息及描述信息对应的助眠音乐推荐类型。
  15. 一种非易失性的计算机可读存储介质，所述非易失性的计算机可读存储介质存储有计算机可读指令，其中，所述计算机可读指令被一种处理器执行时使得所述一种处理器执行如下步骤：
    以预设时间间隔获取目标用户在睡眠时采集到的电压信号和脑电信号；
    对所述电压信号进行向量转换，得到特征向量；
    对所述脑电信号进行图转换处理，得到频谱图；
    将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态；
    基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型；
    根据所述助眠音乐推荐类型，为所述目标用户推送助眠乐曲。
  16. 如权利要求15所述的非易失性的计算机可读存储介质，其中，所述对所述电压信号进行向量转换，得到特征向量的步骤包括：
    基于预设过滤条件，对所述电压信号进行过滤，得到目标信号；
    将所述目标信号进行向量转换处理，得到特征向量。
  17. 如权利要求15所述的非易失性的计算机可读存储介质，其中，所述对所述脑电信号进行图转换处理，得到频谱图的步骤包括：
    对所述脑电信号进行分帧加窗处理，得到特征信号；
    利用短时傅里叶变换对所述特征信号进行变换处理，得到所述频谱图。
  18. 如权利要求15所述的非易失性的计算机可读存储介质，其中，所述对所述脑电信号进行图转换处理，得到频谱图的步骤之后，所述将所述特征向量和所述频谱图导入预先训练好的睡眠检测模型中进行识别，得到睡眠状态的步骤之前，所述计算机可读指令被一种处理器执行时，使得所述一种处理器还执行如下步骤：
    从预设样本库中获取训练样本；
    将所述训练样本导入到卷积神经网络中进行训练，得到所述睡眠检测模型。
  19. 如权利要求18所述的非易失性的计算机可读存储介质，其中，所述将所述训练样本导入到卷积神经网络中进行训练，得到所述睡眠检测模型的步骤包括：
    初始化所述卷积神经网络模型，得到初始模型；
    将所述训练样本导入到所述初始模型中，计算所述初始模型的前向输出；
    根据所述前向输出，计算所述前向输出与预设目标值之间的预测误差；
    根据所述预测误差，使用误差反向传播算法对所述初始模型中各个网络层的初始参数进行调整，得到所述睡眠检测模型。
  20. 如权利要求15所述的非易失性的计算机可读存储介质，其中，所述基于预设条件，获取所述睡眠状态对应的助眠音乐推荐类型的步骤包括：
    从预设识别库中获取预设时间段内的所有睡眠状态；
    基于所述所有睡眠状态，计算相同的睡眠状态对应的累加计数结果；
    将不同的所述睡眠状态的累加计数结果进行比较，选取最大的所述累加计数结果对应的睡眠状态作为目标睡眠状态；
    将所述目标睡眠状态与预设推荐库中的描述信息进行匹配，选取匹配成功的描述信息 对应的助眠音乐推荐类型进行输出，其中，所述预设推荐库包含描述信息及描述信息对应的助眠音乐推荐类型。
PCT/CN2020/099515 2019-09-18 2020-06-30 自动推送助眠乐曲方法、装置、计算机设备及存储介质 WO2021051944A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910882719.5A CN110772700B (zh) 2019-09-18 2019-09-18 自动推送助眠乐曲方法、装置、计算机设备及存储介质
CN201910882719.5 2019-09-18

Publications (1)

Publication Number Publication Date
WO2021051944A1 true WO2021051944A1 (zh) 2021-03-25

Family

ID=69383595

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/099515 WO2021051944A1 (zh) 2019-09-18 2020-06-30 自动推送助眠乐曲方法、装置、计算机设备及存储介质

Country Status (2)

Country Link
CN (1) CN110772700B (zh)
WO (1) WO2021051944A1 (zh)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110772700B (zh) * 2019-09-18 2022-06-03 平安科技（深圳）有限公司 自动推送助眠乐曲方法、装置、计算机设备及存储介质
WO2021258245A1 (zh) * 2020-06-22 2021-12-30 华为技术有限公司 更新助眠音频信号的方法及装置
CN112569448A (zh) * 2020-12-28 2021-03-30 盛铭睿 活动寝具及其所应用的模型构建和助眠方法、终端及介质
CN114431837B (zh) * 2022-04-12 2022-08-16 深圳市心流科技有限公司 一种睡眠状态控制方法、装置、助眠设备及存储介质
CN115120837A (zh) * 2022-06-27 2022-09-30 慕思健康睡眠股份有限公司 基于深度学习的睡眠环境调节方法、系统、装置及介质
CN115904089B (zh) * 2023-01-06 2023-06-06 深圳市心流科技有限公司 App主题场景推荐方法、装置、终端设备及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090143636A1 (en) * 2007-11-30 2009-06-04 Palo Alto Research Center Incorporated Brain-wave aware sleep management
CN103372258A (zh) * 2012-04-12 2013-10-30 孙雪青 失眠治疗仪及失眠治疗方法
CN110075409A (zh) * 2019-04-17 2019-08-02 重庆大学 基于脑电波的个性化音乐辅助睡眠方法
CN110193127A (zh) * 2019-04-23 2019-09-03 平安科技（深圳）有限公司 音乐辅助睡眠的方法、装置、计算机设备及存储介质
CN110772700A (zh) * 2019-09-18 2020-02-11 平安科技（深圳）有限公司 自动推送助眠乐曲方法、装置、计算机设备及存储介质

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN203564213U (zh) * 2013-10-18 2014-04-30 深圳海斯凯医学技术有限公司 睡眠监护仪及其睡眠监护系统
CN205568920U (zh) * 2016-01-12 2016-09-14 杜越新 一种压敏传感式人体监测系统控制的播放设备
CN106419893A (zh) * 2016-09-18 2017-02-22 广州视源电子科技股份有限公司 睡眠状态检测方法和装置
JP7465808B2 (ja) * 2017-12-28 2024-04-11 スリープ ナンバー コーポレイション 睡眠段階検知特徴部を備えたベッド
CN109274757A (zh) * 2018-10-15 2019-01-25 珠海格力电器股份有限公司 一种音乐推送方法、家电设备及计算机存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090143636A1 (en) * 2007-11-30 2009-06-04 Palo Alto Research Center Incorporated Brain-wave aware sleep management
CN103372258A (zh) * 2012-04-12 2013-10-30 孙雪青 失眠治疗仪及失眠治疗方法
CN110075409A (zh) * 2019-04-17 2019-08-02 重庆大学 基于脑电波的个性化音乐辅助睡眠方法
CN110193127A (zh) * 2019-04-23 2019-09-03 平安科技（深圳）有限公司 音乐辅助睡眠的方法、装置、计算机设备及存储介质
CN110772700A (zh) * 2019-09-18 2020-02-11 平安科技（深圳）有限公司 自动推送助眠乐曲方法、装置、计算机设备及存储介质

Also Published As

Publication number Publication date
CN110772700B (zh) 2022-06-03
CN110772700A (zh) 2020-02-11

Similar Documents

Publication Publication Date Title
WO2021051944A1 (zh) 自动推送助眠乐曲方法、装置、计算机设备及存储介质
JP7177167B2 (ja) 混合音声の特定方法、装置及びコンピュータプログラム
US10083397B2 (en) Personalized intelligent wake-up system and method based on multimodal deep neural network
Vafeiadis et al. Audio content analysis for unobtrusive event detection in smart homes
CN110956966B (zh) 声纹认证方法、装置、介质及电子设备
EP3189779A2 (en) Electrocardiogram (ecg) authentication method and apparatus
CN111178548B (zh) 集成学习预测方法与系统
CN108701469A (zh) 咳嗽声音识别方法、设备和存储介质
CN107688790B (zh) 人体行为识别方法、装置、存储介质及电子设备
CN110461215B (zh) 使用便携式设备确定健康标志
CN109065176B (zh) 一种血糖预测方法、装置、终端和存储介质
CN107910020A (zh) 鼾声检测方法、装置、设备及存储介质
CN112690761B (zh) 睡眠状态检测方法、装置、设备及计算机可读介质
WO2017136339A1 (en) Systems and methods for probabilistic pulse rate estimation from photoplethysmographic measurements in the presence of nonstationary and nontrivial signal and noise spectra
US10910000B2 (en) Method and device for audio recognition using a voting matrix
WO2020258507A1 (zh) X光片的分类方法、装置、终端及存储介质
CN113851147A (zh) 音频识别方法、音频识别模型训练方法、装置、电子设备
CN118177766A (zh) 基于个体生理特征的心脑血管健康监测预警方法及系统
CN113793615A (zh) 说话人识别方法、模型训练方法、装置、设备及存储介质
CN110472788B (zh) 一种人员信息预测方法及装置
WO2022049700A1 (ja) 動作評価方法、コンピュータプログラム及び動作評価システム
CN113948089B (zh) 声纹模型训练和声纹识别方法、装置、设备及介质
CN107995181B (zh) 一种基于步态的身份验证方法、装置、设备及存储介质
CN111382712A (zh) 一种手掌图像识别方法、系统及设备
WO2023274894A1 (en) Method for monitoring acoustic events in a patient

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20866130

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 21.07.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 20866130

Country of ref document: EP

Kind code of ref document: A1