A vehicle-mounted audio control method, system, device and medium

Abstract

The application discloses a kind of vehicle-mounted audio control method, system, equipment and medium, by the weight of current each seat passenger passenger information matching current each seat passenger user information by pre-training model is matched first song list for passenger, if there is no current seat passenger information matching, by physiological-behavior joint clustering method, obtain current seat passenger character information and cloud user clustering center match to most similar group, and extract the song list of group and current seat passenger character information highest similarity as second song list, in song list play, listen to pressure vibration sensor and obtain the body movement frequency of current each seat passenger, and the body movement frequency of current each seat passenger is analyzed, to obtain user feedback information using the user feedback information obtains the favorite list of current each seat passenger each other coincident part, generate third song list and play, solve the problem that existing vehicle-mounted audio playing equipment is insufficient in the personalized playing control of multiple occupancy personnel scene.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology and relates to an in-vehicle audio control method, system, device and medium. Background Technology

[0002] With the continuous development of the automotive industry and people's increasing demands for driving experience, in-car audio playback devices have become an indispensable part of modern cars. They not only provide drivers and passengers with a wealth of entertainment content, alleviating the boredom of driving, but also enhance safety and comfort by playing music, news, and other audio information.

[0003] Currently, most in-car audio playback devices on the market control music playback based on the driver's or vehicle's linked account's music preferences. This method can well meet the user's personalized needs when the driver is driving alone or the vehicle is mainly used by a specific person. However, in real-world usage scenarios, a vehicle, as a means of transportation, often carries more than one person. Especially in family trips, gatherings with friends, or carpooling, there may be multiple passengers inside the vehicle, and these passengers' music preferences may vary significantly.

[0004] Existing in-vehicle audio control systems struggle to effectively collect and analyze the music preferences of multiple passengers. Lacking a precise understanding of shared tastes, the system often recommends tracks based solely on the driver's or a single linked account's preferences, resulting in music selections that may not suit the tastes of most passengers. This not only diminishes the audio entertainment experience but also significantly reduces the personalization of in-vehicle audio playback devices.

[0005] Therefore, how to improve the personalized playback control capabilities of in-vehicle audio playback devices in scenarios with multiple passengers, so that they can recommend playback tracks based on the common preferences of multiple passengers, has become an important problem that urgently needs to be solved in the field of in-vehicle audio playback technology. Summary of the Invention

[0006] To address the problems existing in the background technology, this invention proposes an in-vehicle audio control method, system, and device, aiming to solve the problem of insufficient personalized playback control of existing in-vehicle audio playback devices in scenarios with multiple passengers.

[0007] The first aspect of this application provides an in-vehicle audio control method, comprising:

[0008] The system uses the weight of passengers in each seat to match their user information and uses a pre-trained model to match the first playlist for each passenger.

[0009] If there is no matching information for the current seat passenger, the current seat passenger information is obtained through the physiological behavior joint clustering method, matched with the cloud user clustering center to the most similar group, and the playlist with the highest similarity between the group and the current seat passenger information is extracted as the second playlist.

[0010] During the playback of the first playlist or the second playlist, the body movement frequency of each passenger in each seat is obtained by listening to the pressure vibration sensor, and the body movement frequency of each passenger in each seat is analyzed to obtain user feedback information.

[0011] The user feedback information is used to obtain the preference list of each passenger in the current seat. Based on the overlap of the preference lists, a third playlist is generated and played.

[0012] Optionally, construct the passenger information for the current seat, including:

[0013]

[0014] μ v This represents the average vehicle speed. Let μ be the variance of the vehicle's speed. Δv This represents the average value of the vehicle speed change information. Let μt be the variance of vehicle speed variation information, and μt be the average frequency of passenger body movements. Let μ be the variance of the passenger's body movement frequency. Δt This represents the average value of information on changes in the frequency of passenger body movements. Variance of information on changes in passenger body movement frequency. The main frequency for vehicle speed. The dominant frequency of passenger's body movements.

[0015] Optionally, match to the most similar groups, including:

[0016]

[0017] F new For the character model of the passenger in the current seat, F i In the cloud data, the i-th person model, d is the total number of features, and w j F represents the weight of the j-th feature. new，j For the j-th feature of the person model of the passenger in the current seat, F i，j Let j be the j-th feature of the i-th person model in the cloud data.

[0018] Optionally, the process of generating the second playlist includes:

[0019] Read the playlists corresponding to the most similar groups, and find the intersection of the corresponding playlists in the most similar groups;

[0020] A second playlist is generated based on the intersection of the corresponding playlists in the most similar groups.

[0021] Optionally, the monitoring pressure vibration sensor obtains the current body movement frequency of passengers in each seat, including:

[0022] The synchronization rate R between the user's motion frequency F and the song's rhythm B is calculated through motion rhythm coupling analysis:

[0023]

[0024] in, Let F(t) be the phase extraction function, R > 0.7 is marked as liked, R ≤ 0.7 is marked as disliked, T is the total playback time of the first playlist or the second playlist, t is the time period of any song playback, F(t) is the user movement frequency in the time period t, and B(t) is the song rhythm in the time period t.

[0025] Optionally, the process of generating the third playlist is as follows:

[0026] A reinforcement learning strategy is used to update song weights based on the synchronization rate R, generating a third playlist:

[0027]

[0028] in, This is the third playlist. Let α be the first playlist or the second playlist, R be the learning rate, β be the search factor, Entropy be the entropy function, and Q be the search factor. song It can be any song from the first or second playlist.

[0029] Optionally, after generating the third playlist, it may also include:

[0030] The pre-trained model adjusts the third playlist based on the feedback from passengers in each seat, generates multiple target playlists, and generates a fourth playlist based on the intersection of the multiple target playlists.

[0031] A second aspect of this application provides an in-vehicle audio control system, comprising:

[0032] The first module is used to match the current passenger user information with the current weight of each passenger and use a pre-trained model to match the first playlist for each passenger.

[0033] The second module is used to obtain the passenger information of the current seat through a physiological behavior joint clustering method if there is no matching passenger information of the current seat. It then matches the passenger information with the cloud user clustering center to the most similar group and extracts the playlist with the highest similarity between the group and the passenger information of the current seat as the second playlist.

[0034] The third module is used to monitor the body movement frequency of passengers in each seat during the playback of the first and second playlists, and to analyze the body movement frequency of passengers in each seat to obtain user feedback information.

[0035] The fourth module is used to obtain the current preference list of each passenger in each seat using the user feedback information, and to generate a third playlist based on the overlap of each preference list.

[0036] A third aspect of this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described vehicle audio control method.

[0037] A fourth aspect of this application provides a computer-readable medium storing a computer program that, when executed by a processor, implements the above-described vehicle audio control method.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The aforementioned in-vehicle audio control method, system, and device firstly match the weight of passengers in each seat with user information, and then use a pre-trained model to accurately match a first playlist for each passenger. This fully considers the initial characteristics of passengers, giving the music recommendation a high degree of personalization from the initial stage. Then, for passenger information that cannot be directly matched, a physiological behavior joint clustering method is used to match the passenger's information with cloud-based user cluster centers, and the playlist with the highest similarity is extracted as a second playlist. This further broadens the coverage of personalized recommendations, ensuring that even with incomplete information, relatively suitable music choices can be provided to passengers, significantly improving the accuracy of personalized music recommendations. Attached Figure Description

[0040] Figure 1 This is a flowchart of an in-vehicle audio control method according to an embodiment of the present invention;

[0041] Figure 2 This is a schematic diagram of an in-vehicle audio control system according to one embodiment of the present invention;

[0042] Figure 3This is a schematic diagram of a robot production safety supervision device according to one embodiment of the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] In one embodiment, such as Figure 1 As shown, an in-vehicle audio control method is provided, which is applied to... Figure 1 Taking China as an example, the following specific steps will be used:

[0045] S10: Match the current passenger information with the current passenger weight, and match the first playlist for the passenger using a pre-trained model.

[0046] Specifically, in a vehicle, each seat may have a passenger. Pressure and vibration sensors installed in the seats acquire the weight data of each passenger. This weight data is then matched against pre-stored passenger user information, including the passenger's name, age, gender, travel history, and personal preferences. By using weight as a reference factor, the system identifies the corresponding passenger. For example, if a specific weight range corresponds to a particular passenger, the system can determine which passenger is sitting in that seat.

[0047] The pre-trained model is a machine learning or deep learning model trained on a large amount of data. Using this trained model, combined with matched passenger user information such as age, gender, and personal music preferences, a suitable playlist is recommended for each passenger. This playlist is the first playlist, which is composed of a series of songs predicted by the pre-trained model as the most likely favorites of the passenger. This playlist is played during the ride to enhance the passenger's experience.

[0048] For example, after a passenger sits down, the pressure and vibration sensors on the seat detect changes in pressure in real time, thus calculating the passenger's weight. For instance, if the weight of the passenger in the front passenger seat is detected as 60 kg, the vehicle's intelligent system will search its user database. The database records information about passenger Xiaoming (weight range 5862 kg). Through weight matching, the system identifies the passenger currently sitting in the front passenger seat as Xiaoming. Xiaoming's user information also includes age 30, male, preference for pop and rock music, and a history of frequently listening to Jay Chou and Mayday songs while traveling. The vehicle system has a built-in deep learning-based music recommendation pre-trained model. This model selects a series of songs from a massive music library to create a playlist based on Xiaoming's age, gender, and music preferences. For example, the model might select Jay Chou's "Blue and White Porcelain" and "Seven Mile Fragrance," and Mayday's "Contentment" and "Stubborn," generating a playlist of 20 songs as Xiaoming's first playlist. Then, the vehicle's audio system will start playing songs from this playlist, allowing Xiaoming to enjoy his favorite music during the ride.

[0049] The same process will also be applied to passengers in other seats in the car. Based on each passenger's weight, their user information will be matched, and then a personalized first playlist will be matched for each of them through a pre-trained model to improve the overall riding experience.

[0050] S20: If there is no matching information for the current seat passenger, the current seat passenger information is obtained through the physiological behavior joint clustering method, matched with the cloud user clustering center to the most similar group, and the playlist with the highest similarity between the group and the current seat passenger information is extracted as the second playlist.

[0051] Specifically, during vehicle operation, passenger weight and other information are acquired through devices such as pressure and vibration sensors on the seats. However, if a matching passenger cannot be found in the existing local user information database, the process is flawed. For example, a new passenger may be using the vehicle for the first time, resulting in no record of them in the system, or there may be errors in weight measurement. Then, the physiological and behavioral clustering method is used to comprehensively consider the passenger's physiological and behavioral information, classifying and aggregating the passenger information. Since a large amount of user clustering information is stored in the cloud, each cluster center represents a user group with similar characteristics. The physiological and behavioral information of the currently unmatched passenger is compared with the various cluster centers in the cloud to find the most similar user group. Then, the playlist corresponding to the most similar group is read, and the intersection of the corresponding playlists within the most similar group is calculated. A second playlist is generated based on this intersection. The playlist corresponding to the most similar group is a group composed of multiple specific playlists, and the common parts within these groups are used to form the second playlist. For example, if a passenger's physiological and behavioral characteristics are most similar to the "young people" group in the cloud, that passenger will be classified into that group.

[0052] In this embodiment of the application, based on the passenger's physiological characteristics, such as weight, and behavioral characteristics such as the state of body movement and driving status during the playback of the song, the degree of the passenger's liking for the song is determined, and similar songs are searched to form a second playlist.

[0053] The frequency of passenger body movements includes the frequency of body movements and the amplitude of changes in that frequency. By calculating the average passenger body movement frequency, the variance of the passenger body movement frequency, the average value of the passenger body movement frequency variation information, and the variance of the passenger body movement frequency variation information, a precise picture of passenger body movements can be obtained.

[0054] For example, calculating the average body movement frequency of passengers reveals their average hyperactivity and activity level. A higher average body movement frequency indicates a greater preference for lively and upbeat songs, while a lower average indicates a greater preference for quiet and gentle songs. Calculating the variance of body movement frequency reflects the speed of emotional changes. A large variance indicates rapid emotional changes and greater adaptability to different song styles, while a small variance indicates slower emotional changes and less adaptability to different styles. The average variation in body movement frequency reflects the magnitude of emotional fluctuations. A large average variation indicates greater emotional fluctuations and a wider acceptance of various song styles, while a small average variation indicates less emotional fluctuations and a narrower acceptance of different song styles. Furthermore, the variance of passenger body movement frequency changes reflects the frequency of passenger emotional fluctuations. A large variance indicates a high frequency of emotional fluctuations, which in turn indicates a high frequency of variation in different song styles within the passenger's playlist; conversely, a small variance indicates a low frequency of emotional fluctuations, which in turn indicates a low frequency of variation in different song styles within the passenger's playlist. In this application, the magnitude of body movement frequency is correlated with the passenger's weight collected from the passenger seat pressure vibration sensor, but this does not affect the ability to accurately collect the magnitude of body movement frequency using existing technologies.

[0055] Passenger behavior while traveling includes changes in vehicle speed, frequency of acceleration, and magnitude of acceleration. For example, calculating the average vehicle speed reveals whether passengers are more impatient or relaxed during the journey. A higher average speed suggests a preference for fast-paced music, while a lower average speed indicates a preference for slower-paced music. Calculating the variance of vehicle speed reveals the degree of change in the vehicle's speed during travel, representing passenger tolerance for variation. A higher variance indicates greater tolerance for variation and a wider acceptance of various music styles, while a lower variance indicates lower tolerance and a preference for fewer or limited music styles. Calculating the average of vehicle speed variation information reflects passengers' pursuit of stimulation. Greater speed variations suggest a preference for more exciting and intense music, while smaller speed variations suggest a preference for more stable and soothing music. By calculating the variance of vehicle speed change information, we can determine the passenger's acceptability of stimuli. If the variance of vehicle speed change information is large, it indicates that the passenger has a strong acceptability of stimuli, and different types of stimulating and exciting songs can be recommended. If the variance of vehicle speed change information is small, it indicates that the passenger has a weak acceptability of stimuli, and fewer or single types of songs can be recommended.

[0056] It is worth noting that, in this embodiment of the application, the passenger's physiological characteristics are comprehensively judged to indicate their liking for the songs in the first or second playlist by using the average value of the passenger's body movement frequency, the variance of the passenger's body movement frequency change information, the variance of the passenger's body movement frequency, and the variance of the passenger's body movement frequency change information. In addition, driving behavior information is used in conjunction with the passenger's physiological characteristics, such as by calculating the average value of the vehicle's driving speed, the variance of the vehicle's driving speed, the average value of the vehicle's driving speed change information, and the variance of the vehicle's driving speed change information. The physiological and behavioral characteristics are used to comprehensively judge the passenger's liking for the playlist and feed the liking back to the vehicle system to adjust the playlist.

[0057] After finding the most similar group, which may already have multiple different playlists, the intersection of the playlists corresponding to the most similar group is selected and used as a second playlist to recommend to the passenger in the current seat. This playlist is used to play during the journey to meet the passenger's musical needs.

[0058] The specific method for obtaining the passenger information of the current seat is through calculation using the following formula:

[0059]

[0060] Where, μv This represents the average vehicle speed. Let μ be the variance of the vehicle's speed. Δv This represents the average value of the vehicle speed change information. Let μ be the variance of the vehicle speed change information. t This represents the average frequency of passenger body movements. Let μ be the variance of the passenger's body movement frequency. Δt This represents the average value of information on changes in the frequency of passenger body movements. Variance of information on changes in passenger body movement frequency. The dominant frequency of vehicle speed is used, and its frequency domain characteristics are obtained through Fourier transform. This transforms vehicle speed information from the time domain to the frequency domain, providing data support for the average vehicle speed, its variance, the average value of vehicle speed variation, and the variance of vehicle speed variation. This avoids a lack of awareness of changes in passenger data when dealing with large amounts of data. The dominant frequency of passenger body movement is obtained through Fourier transform to obtain the frequency domain characteristics of passenger body movement. The information of passenger body movement is transformed from the time domain to the frequency domain. Data support is provided for the average value of passenger body movement frequency, the variance of passenger body movement frequency, the average value of passenger body movement frequency change information, and the variance of passenger body movement frequency change information, so as to avoid the lack of perception of changes in passenger data when there is a large amount of data.

[0061] The specific method for matching to the most similar group is calculated using the following formula:

[0062]

[0063] F new For the character model of the passenger in the current seat, F i In the cloud data, the i-th person model, d is the total number of features, and w j F represents the weight of the j-th feature. new，j For the j-th feature of the person model of the passenger in the current seat, F i，j Let j be the j-th feature of the i-th person model in the cloud data.

[0064] S30: During the playback of the first playlist or the second playlist, the body movement frequency of each passenger in each seat is obtained by listening to the pressure vibration sensor, and the body movement frequency of each passenger in each seat is analyzed to obtain user feedback information.

[0065] Specifically, when playing a first playlist generated by a pre-trained model after matching known passenger information, or a second playlist generated using a physiological behavior joint clustering method for passengers who cannot be matched, pressure-vibration sensors installed on the vehicle seats can monitor passenger movement in real time. Body movement frequency refers to the number of times or the rhythm of a passenger's body movements within a certain time period. For example, if a passenger gently sways their body to the rhythm of the music, the pressure-vibration sensors can capture this movement and record the corresponding frequency data. The body movement frequency data of each passenger in each seat obtained from the pressure-vibration sensors is then analyzed in depth. By analyzing this data, feedback information such as the passenger's liking for the currently playing playlist and their satisfaction can be inferred. For example, if the passenger's body movement frequency matches the music rhythm and the movement is relatively frequent, it may indicate that the passenger likes the playlist; conversely, if the passenger's body movement is minimal, it may mean that the passenger is not very interested in the playlist.

[0066] For example, when a catchy pop song is played, the pressure vibration sensor detects that the driver's body sways slightly with the music's rhythm. Calculations show that the driver's body movement frequency roughly matches the music's rhythm. Meanwhile, the passenger in the front seat shows virtually no body movement, with a frequency close to zero times per minute. The vehicle's intelligent system analyzes this body movement frequency data. Based on the analysis, it determines that the driver is relatively satisfied with the current playlist because their body movement frequency corresponds to the music's rhythm, demonstrating a positive response; while the passenger may not be very interested in the playlist because they show almost no body movement. Based on this feedback, the vehicle system can further adjust the playlist to include music types that both the driver and passenger enjoy, thereby enhancing the overall riding experience and music enjoyment.

[0067] In this application, the synchronization rate R between the user's motion frequency F and the song's rhythm B is calculated through motion rhythm coupling analysis:

[0068]

[0069] in, Let R be the phase extraction function, R > 0.7 is marked as liked, R ≤ 0.7 is marked as disliked, T is the total playback time of the first playlist or the second playlist, t is the time period of any song playback, F(t) is the user movement frequency in the time period t, and B(t) is the song rhythm in the time period t.

[0070] S40: Using the user feedback information, obtain the current preference list of each passenger in each seat, and generate a third playlist based on the overlap of each preference list for playback.

[0071] Specifically, based on user feedback from passengers in each seat when playing music playlists, a music preference list for each passenger can be generated. This list includes information such as the music genres, artists, and songs that each passenger might like. Then, the preference lists of all passengers are compared to identify overlaps—the music-related elements that all passengers commonly enjoy. For example, multiple passengers' preference lists might all contain the genre "pop music," or they might all have the names of certain artists. Based on the overlaps found in the preference lists of each passenger, corresponding songs are selected from the vehicle's music library to create a new playlist, the third playlist. This third playlist is then played through the vehicle's audio system to satisfy the shared musical preferences of all passengers and enhance the overall in-car music experience.

[0072] It is worth noting that this application employs a reinforcement learning strategy to update song weights based on the synchronization rate R, thereby generating a third playlist:

[0073]

[0074] in, This is the third playlist. This is the second playlist, where α is the learning rate, R is the synchronization rate, β is the search factor, Entropy is the entropy function, and Q is the second playlist. song Any song in the second playlist.

[0075] In one embodiment, step S40, i.e. after generating the third playlist, further includes the following steps:

[0076] The pre-trained model adjusts the third playlist based on the feedback from passengers in each seat, generates multiple target playlists, and generates a fourth playlist based on the intersection of the multiple target playlists.

[0077] Specifically, during the playback of the third playlist, information such as the body movement frequency of passengers in each seat is again obtained through devices such as pressure vibration sensors, serving as passenger feedback on the third playlist. At this time, the pre-trained model uses this feedback information to understand the passengers' level of liking and satisfaction with the songs in the third playlist. Based on the passenger feedback, the pre-trained model modifies and adjusts the third playlist. For example, if some songs are found to be disliked by passengers, they are removed from the playlist, and new songs that might suit passenger preferences are added. Multiple different target playlists are generated through different adjustment methods and strategies. Each target playlist may differ in song selection, song order, etc., but they are all optimized based on passenger feedback on the third playlist. The generated target playlists are compared to identify the songs they commonly contain; the set of these common songs is the intersection of the multiple target playlists. Based on this intersection, a new playlist, the fourth playlist, is generated. This fourth playlist contains songs found in all the target playlists, theoretically better satisfying the common preferences of passengers in each seat.

[0078] For example, the vehicle system played a third playlist for the driver and five passengers. During playback, feedback from each person was collected via pressure and vibration sensors on the seats and other possible monitoring devices. For instance, the driver showed little interest in a slower song, barely moving to the music; Passenger 1 enjoyed a popular song, humming along and swaying frequently; Passenger 2 reacted strongly to a rock song, displaying excitement. The pre-trained model adjusted the third playlist based on this feedback. The slower song the driver disliked was removed, and a moderately paced classic was added; several other popular songs of the same genre and artist were added to Passenger 1's favorite; and several similar rock tracks were added to Passenger 2's favorite. Through these different adjustments, three different target playlists were generated. Target playlist 1 is geared towards the driver's preferences, target playlist 2 focuses on satisfying the tastes of passengers like Passenger 1 who enjoy pop music, and target playlist 3 primarily caters to passengers like Passenger 2 who prefer rock music. Comparing these three target playlists reveals that they all contain a classic pop song, "Friends," and an upbeat rock song, "Nowhere to Hide." Based on these two shared songs, several other songs with similar styles or broad appeal were selected from the music library to create a fourth playlist containing eight songs. The vehicle's audio system then plays this fourth playlist, as it incorporates elements from multiple target playlists, making it more likely to satisfy the musical preferences of all passengers and enhance their riding experience.

[0079] In one embodiment, such as Figure 2 As shown, an in-vehicle audio control system is provided, which corresponds one-to-one with the in-vehicle audio control method in the above embodiments. The in-vehicle audio control system includes: a first module, a second module, a third module, and a fourth module. The functional modules are described in detail below:

[0080] The first module is used to match the current passenger user information with the current weight of each passenger and use a pre-trained model to match the first playlist for each passenger.

[0081] The second module is used to obtain the passenger information of the current seat through a physiological behavior joint clustering method if there is no matching passenger information of the current seat. It then matches the passenger information with the cloud user clustering center to the most similar group and extracts the playlist with the highest similarity between the group and the passenger information of the current seat as the second playlist.

[0082] The third module is used to monitor the body movement frequency of passengers in each seat during the playback of the first and second playlists, and to analyze the body movement frequency of passengers in each seat to obtain user feedback information.

[0083] The fourth module is used to obtain the current preference list of each passenger in each seat using the user feedback information, and to generate a third playlist based on the overlap of each preference list.

[0084] Specific limitations regarding the vehicle audio control system can be found in the limitations of the vehicle audio control method described above, and will not be repeated here. Each module in the aforementioned vehicle audio control system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0085] In one embodiment, such as Figure 3 As shown, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements an in-vehicle audio control method.

[0086] For specific limitations on electronic devices, please refer to the limitations on in-vehicle audio control methods mentioned above, which will not be repeated here.

[0087] In one embodiment, a computer-readable medium is provided, the computer-readable medium storing a computer program, wherein when a processor executes the computer program, it performs the following steps:

[0088] The system uses the weight of passengers in each seat to match their user information and uses a pre-trained model to match the first playlist for each passenger.

[0089] If there is no matching information for the current seat passenger, the current seat passenger information is obtained through the physiological behavior joint clustering method, matched with the cloud user clustering center to the most similar group, and the playlist with the highest similarity between the group and the current seat passenger information is extracted as the second playlist.

[0090] During the playback of the first playlist or the second playlist, the body movement frequency of each passenger in each seat is obtained by listening to the pressure vibration sensor, and the body movement frequency of each passenger in each seat is analyzed to obtain user feedback information.

[0091] The user feedback information is used to obtain the preference list of each passenger in the current seat. Based on the overlap of the preference lists, a third playlist is generated and played.

[0092] For specific limitations regarding computer-readable media, please refer to the limitations on in-vehicle audio control methods mentioned above, which will not be repeated here.

[0093] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0094] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method of controlling audio in a vehicle, the method comprising: Including: The system uses the weight of passengers in each seat to match their user information and uses a pre-trained model to match the first playlist for each passenger. If there is no matching information for the current seat passenger, the passenger's information is obtained through a physiological-behavioral joint clustering method. This information is then matched with the cloud-based user clustering center to find the most similar group. The playlist with the highest similarity between this group and the current seat passenger's information is then extracted as the second playlist. The current seat passenger information includes: ; This represents the average vehicle speed. Let Variance be the vehicle speed. This represents the average value of the vehicle speed change information. The variance of vehicle speed change information. This represents the average frequency of passenger body movements. Let Variance be the frequency of passenger body movements. This represents the average value of information on changes in the frequency of passenger body movements. Variance of information on changes in passenger body movement frequency. The main frequency for vehicle speed. The dominant frequency of passenger body movement; During the playback of the first or second playlist, the pressure vibration sensor detects the current body movement frequency of each passenger and analyzes the body movement frequency to obtain user feedback information. The detection of the current body movement frequency by the pressure vibration sensor includes: calculating the user's movement frequency through motion-rhythm coupling analysis. With the rhythm of the song Synchronization rate : ;in, The phase extraction function is defined as follows: R > 0.7 indicates a favorable value, and R ≤ 0.7 indicates a negative value. The total playback time of the first playlist or the second playlist. For any given song, the time period during which it is played. In order to be in User movement frequency during the time period In order to be in The rhythm of the song within a given time period; The user feedback information is used to obtain the preference lists of each passenger in each seat. Based on the overlap between the preference lists, a third playlist is generated for playback. The generation process of the third playlist is as follows: a reinforcement learning strategy is used, based on the synchronization rate... Update song weights and generate a third playlist: ;in, This is the third playlist. For the first playlist or the second playlist, For learning rate, Synchronization rate, As search factors, It is the entropy function. It can be any song from the first or second playlist.

2. The vehicle audio control method according to claim 1, characterized in that, Match to the most similar groups, including: ； The character model of the passenger in the current seat. For the first in cloud data A character model For the total number of features, For the first The weight of each feature, The first character model of the passenger in the current seat Item features, For the first in cloud data The first character model Feature.

3. The vehicle audio control method according to claim 1, characterized in that, The process of generating the second playlist includes: Read the playlists corresponding to the most similar groups, and find the intersection of the corresponding playlists in the most similar groups; A second playlist is generated based on the intersection of the corresponding playlists in the most similar groups.

4. The vehicle audio control method according to claim 1, characterized in that, After generating the third playlist, it also includes: The pre-trained model adjusts the third playlist based on the feedback from passengers in each seat, generates multiple target playlists, and generates a fourth playlist based on the intersection of the multiple target playlists.

5. A vehicle-mounted audio control system for implementing the method according to any one of claims 1 to 4, characterized in that, Includes the following modules: The first module is used to match the current passenger user information with the current weight of each passenger and use a pre-trained model to match the first playlist for each passenger. The second module is used to obtain the passenger information of the current seat through the physiological-behavioral joint clustering method if there is no matching passenger information of the current seat. It then matches the passenger information with the cloud user clustering center to the most similar group and extracts the playlist with the highest similarity between the group and the passenger information of the current seat as the second playlist. The third module is used to monitor the body movement frequency of passengers in each seat during the playback of the first and second playlists, and to analyze the body movement frequency of passengers in each seat to obtain user feedback information. The fourth module is used to obtain the current preference list of each passenger in each seat using the user feedback information, and to generate a third playlist based on the overlap of each preference list.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the vehicle audio control method according to any one of claims 1 to 4.

7. A computer-readable medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the in-vehicle audio control method according to any one of claims 1 to 4.

Images