Music recommendation method, vehicle and computer readable storage medium

Abstract

The invention provides a music recommendation method, a vehicle and a computer readable storage medium, and relates to the technical field of vehicles, the music recommendation method comprises the following steps: determining a current road scene type; wherein the road scene type is used for representing the driving road condition of the vehicle; based on the current road scene type, target music matched with the current road scene type is determined from the target mapping relation, and the preference degree value of the target music is larger than or equal to a target threshold value; wherein the target mapping relation comprises the media information of the first music played in the historical time period of the vehicle, the preference degree value of the user for the first music and the corresponding relation of the road scene type for playing the first music, and the media information at least comprises the name information of the first music; and recommending the target music to the user. According to the technical scheme of the invention, based on the liking degree of the music played for the first time in the previous time period and the road scene, the music which the user likes in the current road scene can be accurately obtained, and the music recommendation accuracy is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicles, in particular to a music recommendation method, a vehicle and a computer readable storage medium. BACKGROUND

[0002] A general music recommendation method is to construct a music feature vector by audio feature extraction of music, and to obtain music to be recommended by matching a user portrait with the aforementioned music feature vector. The audio feature extraction is a process of extracting numerical information representing some auditory properties (such as music style) of music from the original music audio signal; the music feature vector is a multidimensional numerical vector combining all audio features of a song, used to mathematically represent the characteristics of the song; the user portrait refers to the music feature vector of the type of music (such as rock) that the user likes to listen to.

[0003] The general music recommendation method only recommends music of a certain music type to the user, and not all music of a certain music type is liked by the user, so the accuracy of music recommendation is low. SUMMARY

[0004] The embodiments of the present application provide a music recommendation method, a vehicle and a computer readable storage medium, which can accurately obtain music that a user likes in a current road scene based on the like degree of each played music in a past time period and the road scene, and improve the accuracy of recommended music.

[0005] To achieve the above object, the embodiments of the present application adopt the following technical solutions: In a first aspect, a music recommendation method is provided, the method being applied to a vehicle, and the method comprising: determining a current road scene type; wherein the road scene type is used to represent the driving conditions of the vehicle; determining, based on the current road scene type, a target music that matches the current road scene type and has a like degree value greater than or equal to a target threshold from a target mapping relationship; wherein the target mapping relationship comprises a corresponding relationship among media information of a first music played in a historical time period of the vehicle, a like degree value of the user for the first music, and a road scene type in which the first music is played; the like degree value is determined based on the operation behavior of the user for the first music during the period from the start of playing the first music to the end of playing the first music; the media information at least comprises name information of the first music, and the higher the like degree value, the higher the like degree of the user for the music; and recommending the target music to the user.

[0006] In this application, the embodiment is based on the user's various operational behaviors towards a certain music during the period from the start to the end of playing the music, which determines the degree of liking for a certain music. The determination of the music liked by the user is more accurate. Therefore, the vehicle can more accurately obtain the music that the user likes in the current road scenario based on the degree of liking for each music played in the past time period and the road scenario, thus improving the accuracy of music recommendation.

[0007] In conjunction with the first aspect, in one possible design approach, before determining the current road scene type, the method further includes: determining the road scene type when a vehicle is playing the first music if the first music is detected during a historical time period; determining the user's liking value for the first music after the first music finishes playing; and storing the correspondence between the media information of the first music, the user's liking value for the first music, and the road scene type when the first music is played in the target mapping relationship.

[0008] In this application, the songs played in the vehicle can accurately reflect the user's listening habits. After the first song finishes playing, the vehicle can store the media information of the first song, the user's level of liking for the first song, and the corresponding road scene type when the first song was played in the target mapping relationship. In this way, the vehicle can accurately store the user's listening habits, making it easier for the vehicle to accurately find the user's favorite music from the target mapping relationship.

[0009] In conjunction with the first aspect, in one possible design approach, determining the user's level of liking for the first piece of music includes: acquiring user action data related to the first piece of music; wherein the action data includes any one or more of initiation actions, state transition actions, and termination actions, where initiation actions include user actions to trigger the start of playback of the first piece of music, state transition actions include user actions to change the playback state and / or preference state of the first piece of music during playback, and termination actions include user actions to end playback of the first piece of music; determining the intent type of the action actions in the action data; wherein the intent type includes positive intent and / or negative intent, where positive intent represents that the user likes the first piece of music, and negative intent represents that the user dislikes the first piece of music; and determining the user's level of liking for the first piece of music based on the intent type of the action actions in the action data.

[0010] In this application, the cycle of a piece of music includes a start-up phase, a playback phase, and an end-of-play phase. Based on the user's operation behavior in each of the aforementioned phases, the degree of the user's liking for a piece of music can be comprehensively considered, avoiding the judgment of the user's liking for a piece of music based on a single user operation behavior, reducing the misjudgment rate of the user's liking for a song, and improving the accuracy of determining the user's liking for a song.

[0011] In conjunction with the first aspect, in one possible design approach, the user's level of liking for the first music is determined based on the intent type of the operational behavior in the operational behavior data. This includes: determining the target operational behavior based on the intent type of the operational behavior in the operational behavior data; wherein the target operational behavior includes operational behaviors with an intent type of positive intent or negative intent; determining the intent score of each behavior in the target operational behavior; wherein the higher the intent score, the higher the user's level of liking for the first music, and the intent score corresponding to a positive intent is greater than the intent score corresponding to a negative intent; and determining the user's level of liking for the first music based on the intent score of each behavior in the target operational behavior.

[0012] In this application, positive intent is used to represent that the user likes the first song, and negative intent is used to represent that the user dislikes the first song. The higher the intent score, the higher the user's liking for the first song. The intent score corresponding to positive intent is greater than the intent score corresponding to negative intent. Therefore, different intent scores can be assigned to different types of operations during the cycle of a song, and the degree of user liking for a song can be determined more accurately by various user operations during the cycle of a song.

[0013] In conjunction with the first aspect, in one possible design approach, the target operation behavior is determined based on the intent type of the operation behavior in the operation behavior data, including: if at the first moment there is a first operation behavior with an intent type of positive intent or negative intent, and there is no behavior to cancel the first operation behavior between the first moment and the moment when the first music ends playing, then the first operation behavior is determined as the target operation behavior.

[0014] In this application, it is understood that within the cycle of a piece of music, a user's action at one moment might cancel an action performed at a previous moment. The actions at these two moments cancel each other out, effectively meaning the user did not perform the action performed at the previous moment. Therefore, it is possible to avoid determining the intent score for these two actions, saving the time and computational effort required for intent score determination, and improving the speed and efficiency of determining the music's intensity level for the user.

[0015] In conjunction with the first aspect, in one possible design approach, positive intentions include explicit positive intentions and predicted positive intentions, and negative intentions include explicit negative intentions and predicted negative intentions. Determining the intention score for each behavior in the target operational behavior includes: if the intention type of the operational behavior to be confirmed in the operational behavior data is explicit positive intention, then the intention score of the operational behavior to be confirmed is determined as a first value; if the intention type of the operational behavior to be confirmed in the operational behavior data is predicted positive intention, then the intention score of the operational behavior to be confirmed is determined as a second value; wherein, the first and second values ​​are greater than zero, and the second value is less than the first value; if the intention type of the operational behavior to be confirmed in the operational behavior data is explicit negative intention, then the intention score of the operational behavior to be confirmed is determined as a third value; if the intention type of the operational behavior to be confirmed in the operational behavior data is predicted negative intention, then the intention score of the operational behavior to be confirmed is determined as a fourth value; wherein, the fourth value is less than zero, and the third value is greater than the third value.

[0016] In this application, positive intentions are further subdivided into explicit positive intentions and predicted positive intentions, and negative intentions are further subdivided into explicit negative intentions and predicted negative intentions. Moreover, the intention scores corresponding to the aforementioned four types of intentions are different, which can improve the accuracy of intention score determination.

[0017] In conjunction with the first aspect, in one possible design approach, if the operational behavior data contains an operational behavior to be confirmed that is a first target operational behavior, then the intent type of the operational behavior to be confirmed is determined as a clear positive intent; wherein, the first target operational behavior includes any one or more of setting the first music to a "like" status, liking the first music, adding the first music to the favorites, and adding the first music to the playback queue; if the operational behavior data contains an operational behavior to be confirmed that is a second target operational behavior, then the intent type of the operational behavior to be confirmed is determined as a predicted positive intent; wherein, the second target operational behavior includes rewinding the playback progress of the first music to the beginning of the first music and the playback duration of the first music being greater than a first duration, and starting the first music on repeat and the repeating time being... If the number of attempts exceeds one or more of the preset number of attempts; if there is an operation behavior to be confirmed in the operation behavior data that is a third target operation behavior, then the intent type of the operation behavior to be confirmed is determined as an explicit negative intent; wherein, the third target operation behavior includes any one or more of setting the first music to a dislike state and setting the first music to a down state; if there is an operation behavior to be confirmed in the operation behavior data that is a fourth target operation behavior, then the intent type of the operation behavior to be confirmed is determined as a predicted negative intent; wherein, the fourth target operation behavior includes any one or more of manually playing the next music and the playback duration of the first music is less than the second duration, and the user manually dragging the progress bar of the first music and the playback duration of the first music is less than the third duration.

[0018] In conjunction with the first aspect, in one possible design approach, after determining the target music that matches the current road scene type and has a liking value greater than or equal to the target threshold from the target mapping relationship, the method further includes: determining music similar to the target music; wherein, music similar to the target music includes music with the same attribute information as the target music, the attribute information including any one or more of the timbre and music style of the target music; recommending the target music to the user, including: recommending the target music and music similar to the target music to the user.

[0019] In this application, the music that users prefer in the current road scenario can not only be music that users have heard before, but also music that is similar to music that users have liked to listen to in the past. This can broaden the range of music that users like to listen to and improve the user experience.

[0020] In conjunction with the first aspect, in one possible design approach, determining the road scene type in which the vehicle plays the first piece of music includes: acquiring the vehicle speed and the duration of each speed during the playback of the first piece of music; and determining the road scene type based on the vehicle speed and the duration.

[0021] In this application, compared with the method of determining the road scene type by analyzing the image, the road scene type can be determined more simply and quickly using only vehicle speed, thus improving the speed of road scene type determination.

[0022] In conjunction with the first aspect, in one possible design approach, determining the current road scene type includes: obtaining the vehicle's current speed; and determining the current road scene type based on the current speed.

[0023] In this application, compared with the method of determining the road scene type by analyzing the image, the road scene type can be determined more simply and quickly using only vehicle speed, thus improving the speed of road scene type determination.

[0024] Secondly, embodiments of this application provide a vehicle, including: a processor; and a memory for storing processor-executable instructions, wherein the processor is used to execute the music recommendation method described in the first aspect.

[0025] Thirdly, embodiments of this application provide a computer-readable storage medium storing a computer program for performing the music recommendation method described in the first aspect.

[0026] Fourthly, embodiments of this application provide a computer program product, which includes a computer program that, when executed by a processor of a computer device, enables the computer device to perform the music recommendation method described in the first aspect.

[0027] The technical effects of any of the design methods in the second to fourth aspects can be found in the technical effects of different design methods in the first aspect, and will not be repeated here. Attached Figure Description

[0028] Figure 1 The diagram shown is an application scenario illustration of the music recommendation method provided in an exemplary embodiment of this application.

[0029] Figure 2 The diagram shown is a flowchart illustrating a music recommendation method provided in an exemplary embodiment of this application.

[0030] Figure 3 The diagram shown is a schematic representation of an exemplary embodiment of this application, illustrating the process of obtaining varying vehicle speeds and the duration of each speed while playing a song.

[0031] Figure 4 The diagram shown is an exemplary embodiment of this application illustrating the user's operational behavior and intent type during the playback of a song.

[0032] Figure 5 The diagram shown is a schematic representation of an exemplary embodiment of this application, illustrating the process of obtaining user actions during the playback of a song.

[0033] Figure 6 The diagram shown is a schematic representation of the model self-learning process provided in an exemplary embodiment of this application.

[0034] Figure 7 The diagram shown is a structural schematic of a music recommendation device provided in an exemplary embodiment of this application. Detailed Implementation

[0035] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0036] Application Overview As mentioned in the background section, common music recommendation methods only recommend music of a certain genre to users, and not all of the many songs in a certain genre are liked by the user, resulting in low accuracy in music recommendation.

[0037] To address the aforementioned technical problems, this application provides a music recommendation method applied to a vehicle. The method includes: the vehicle responding to a user's opening of a music application to determine the current road scene type; wherein the road scene type characterizes the vehicle's driving conditions; based on the current road scene type, determining target music from a target mapping relationship that matches the current road scene type and has a liking value greater than or equal to a target threshold; wherein the target mapping relationship includes media information of a first song played within a historical time period of the vehicle, the user's liking value for the first song, and the correspondence between the road scene type on which the first song was played, and the media information includes at least the name information of the first song; and recommending the target music to the user.

[0038] In related technologies, a user determines whether a song is of interest simply based on whether they have played or saved it. This method is relatively simple, but if the user unsaves the song, it may no longer be of interest, making the determination of a user's interests inaccurate. However, the embodiments of this application determine a user's level of liking for a song based on various user actions during the playback period. This method is more accurate in identifying a user's preferred music. Furthermore, the vehicle can more accurately determine the user's preferred music in the current road scenario based on their past liking of each played song and the road conditions, thus improving the accuracy of music recommendations.

[0039] Exemplary scenario The music recommendation method provided in this application can be applied to various vehicles. Vehicles include various types of vehicles, such as sedans, SUVs, buses, and trucks. This application does not impose any special limitations on the specific form of the vehicle.

[0040] The following is an example application scenario of a music recommendation method. For example, Figure 1 The diagram illustrates an application scenario of the music recommendation method provided in an exemplary embodiment of this application. Figure 1 As shown, the scenario includes a vehicle 100, which includes a central control screen 110, on which a music application can be displayed.

[0041] Vehicle 100 stores media information of the first music played during a historical time period, the user's liking value for the first music, and the corresponding road scene type where the first music was played. For example, Table 1 illustrates an example table showing the correspondence between media information of the first music played during a historical time period, the user's liking value for the first music, and the road scene type where the first music was played.

[0042] Table 1 Understandably, the higher the user's liking score for the music, the greater their level of liking for the music.

[0043] Vehicle 100 can respond to a user's action of opening a music application on the central control screen 110, determine the current road scene type, for example, determine that the current road scene type is urban road and not congested; then, based on the urban road and not congested road scene type, vehicle 100 can determine from Table 1 above the music to be recommended that matches the urban road and not congested road scene type and has a preference value greater than 90: Music 2 and Music 3, and recommend Music 2 and Music 3 to the user. For example, the music recommendation list: Music 2 and Music 3 can be displayed on the display interface 1101 of the central control screen 110 to recommend Music 2 and Music 3 to the user.

[0044] It is understood that the display interface 1101 displays a music recommendation list: Music 2 and Music 3, where Music 2 and Music 3 are the names of the music. However, the aforementioned music recommendation list is only a simple example. In some other embodiments, the music recommendation list may also include any one or more of the following information: singer, actor, lyricist, composer, album name, user's liking value for the music, confidence level of the song recommendation, music application that can play the recommended music, and reasons for recommendation, but is not limited to this.

[0045] It should be understood that the above application scenario examples are only shown to facilitate understanding of the spirit and principles of this application, and the embodiments of this application are not limited thereto. Rather, the embodiments of this application can be applied to any applicable scenario.

[0046] Exemplary methods Figure 2 The diagram shown is a flowchart illustrating a music recommendation method provided in an exemplary embodiment of this application. Figure 2 The method can be performed by a vehicle, such as Figure 1 Vehicle 100 is executed. For example... Figure 2 As shown, this music recommendation method may include the following steps: 210: Determine the current road scene type; where the road scene type is used to characterize the driving conditions of the vehicle.

[0047] It is understood that, in some embodiments, the vehicle may begin to determine the current road scene type in order to determine the music to be recommended, triggered by the user opening a music application.

[0048] Road scene type is used to characterize the road conditions in which vehicles travel. Road conditions can include any one of the following: road type information, road congestion status information, or a combination of both. For example, road type information could be urban road, urban expressway, highway, or racetrack. Congestion status information could be either congested or not congested.

[0049] In some embodiments, the vehicle may acquire its current speed and, based on the current speed, determine the road scene type from the mapping relationship between speed range and road scene type.

[0050] For example, Table 2 shows a mapping relationship between vehicle speed range and road scene type.

[0051] Table 2 Specifically, the vehicle has an internal communication network: Controller Area Network (CAN). This CAN connects various components within the vehicle to transmit data. For example, the vehicle's infotainment system and chassis are connected to this CAN, allowing the infotainment system to obtain the vehicle's current speed.

[0052] While vehicle speed alone can provide a simple way to determine the type of road scene, it is not very accurate. In other embodiments, the vehicle can acquire an image of its surrounding environment and determine the road scene type based on that image. This improves the accuracy of determining the road scene type.

[0053] 220: Based on the current road scene type, determine the target music that matches the current road scene type and has a liking value greater than or equal to the target threshold from the target mapping relationship.

[0054] The target mapping relationship includes the media information of the first music played during the vehicle's historical time period, the user's liking score for the first music, and the correspondence between the road scene type where the first music was played. At the current moment, the vehicle can determine the music to be recommended based on the aforementioned target mapping relationship, given the current road scene type. For example, vehicle 100 can determine, based on the urban road and uncongested road scene type, from Table 1 above, the music to be recommended that matches the urban road and uncongested road scene type and satisfies a liking score greater than 90 as: Music 2 and Music 3.

[0055] A higher "like" score indicates a greater level of liking for the music. For example, if music 1 has a "like" score of 89 and music 2 has a "like" score of 90, then the user likes music 2 more than music 1.

[0056] The liking level is determined based on the user's actions towards the first song from the start to the end of its playback.

[0057] Media information includes at least the name of the first song, and may also include one or more of the following: the name of the singer, actor, lyricist, composer, album name, user's liking score for the music, confidence level of the song recommendation, music application that can play the recommended music, and reasons for the recommendation, but is not limited to these.

[0058] The method for determining the target mapping relationship, which includes the media information of the first music played during the vehicle's historical time period, the user's liking value for the first music, and the corresponding road scene type where the first music was played, will be introduced below.

[0059] 230: Recommend target music to users.

[0060] In some embodiments, target music can be recommended to the user via voice announcement. In other embodiments, the name of the target music can be displayed on the music application's interface to recommend it to the user. Furthermore, in some embodiments, in addition to displaying the name of the target music, any one or more of the following information can be displayed: the artist, actor, lyricist, composer, album name, the user's liking score for the music, the confidence level of the song recommendation, the music application that can play the recommended music, and the reason for the recommendation, but these are not limited to.

[0061] This application provides a music recommendation method. Based on the current road scene type, a vehicle can determine music that matches the current road scene type and has a liking value greater than or equal to a target threshold by analyzing the media information of the first music played in the vehicle's historical time period, the user's liking value for the first music, and the correspondence between the road scene types in which the first music was played. In this way, the vehicle can obtain the music that the user likes in the current road scene based on the liking value of each music played in the past time period and the road scene, thus improving the accuracy of music recommendation.

[0062] The following describes the correspondence between the media information of the first music played during the vehicle's historical time period, the user's preference value for the first music, and the road scene type in which the first music was played, i.e., the method for determining the target mapping relationship. Before determining the current road scene type, the target mapping relationship is determined. The method for determining the target mapping relationship includes the following steps 310 and 320: 310: If a vehicle is detected playing the first music within a historical time period, determine the road scene type when the vehicle is playing the first music.

[0063] In some embodiments, the vehicle speed and the duration of each vehicle speed during the playback of the first music can be obtained; the average vehicle speed can be determined based on the vehicle speed and the duration of each vehicle speed during the playback of the first music; and the road scene type can be determined based on the average vehicle speed.

[0064] The following example illustrates how to obtain changes in vehicle speed and the duration of each speed while a song is playing. For example, Figure 3 The diagram illustrates a process for acquiring changing vehicle speeds and the duration of various speeds during the playback of a song, as provided in an exemplary embodiment of this application. Figure 3 As shown, the vehicle includes a multimedia agent, a unified multimedia service module, and an adapter module. The multimedia agent is a functional module used to determine the level of liking. The unified multimedia service module is a service module that adapts to different media interfaces and provides a unified control interface. The adapter module is used to access the underlying Controller Area Network (CAN) signals, providing the operating system layer with the ability to access CAN data. CAN signals are digital communication signals based on differential voltage (CAN_H / CAN_L), employing a non-destructive arbitration mechanism to achieve multi-master control, with a maximum speed of 1Mbps, supporting priority transmission and error detection, and are widely used in automotive electronics and industrial control fields.

[0065] After the multimedia agent registers with the adapter module to monitor vehicle speed changes, the adapter module, upon detecting the speed change based on the controller area network (SAR) signal, can then relay the changed speed back to the multimedia agent. The SAR signal refers to the data frames transmitted on the SAR bus.

[0066] After the multimedia agent registers a media change monitoring callback with the multimedia unified service module, the module can monitor the music playback status and provide feedback on this information (such as when music has started or stopped playing) to the multimedia agent. A callback refers to the multimedia unified service module proactively pushing the playback status information back to the multimedia agent when its playback status changes, thus eliminating the overhead of continuous polling by the multimedia agent.

[0067] Depend on Figure 3As can be seen, when the multimedia unified service module calls back to start music playback, the multimedia agent queries the current vehicle speed v1. During music playback, a timer is started to record the duration Tn of the vehicle speed Vn. When the multimedia unified service module calls back to end music playback, the multimedia agent calculates the average vehicle speed during the music playback: (V1*T1 + V2*T2 + ... +Vn*Tn) / (T1 +T2 + ... T3 + Tn). Based on this average vehicle speed, the vehicle can determine the road scene type from Table 2 above.

[0068] In some other embodiments, the vehicle can play an image of the surrounding environment accompanied by music, and determine the road scene type based on the image of the surrounding environment. This can improve the accuracy of determining the road scene type.

[0069] There are several scenarios in which a vehicle might play music. For example, it could be music transferred via USB, music transferred via Bluetooth, music played using a local media player, music played via the radio, or music transferred from a mobile phone.

[0070] Vehicles can play music using media applications, which include system applications and third-party applications. System applications refer to the functions integrated into the vehicle's operating system. Third-party applications can be various music apps.

[0071] 320: After the first music ends, determine the user's liking level for the first music; in the target mapping relationship, store the media information of the first music, the user's liking level for the first music, and the correspondence between the road scene type when the first music is played.

[0072] In this embodiment, the songs played in the vehicle can accurately reflect the user's listening habits. After the first song finishes playing, the vehicle can store the media information of the first song, the user's liking level for the first song, and the correspondence between the road scene type when the first song was played and the target mapping relationship in a relatively accurate manner. In this way, the vehicle can accurately store the user's listening habits, making it easier for the vehicle to accurately find the user's favorite music from the target mapping relationship.

[0073] In some embodiments, the user's level of liking for the first piece of music can be determined through the following steps 3201 to 3203: 3201: Obtain user operation behavior data for the first music; wherein, operation behavior data includes any one or more of the following: start behavior, state change behavior, and end behavior.

[0074] Operations can include media playback control actions such as playing the previous track, playing the next track, pausing playback, starting playback, setting playback speed, switching playback modes, setting fast forward, and setting rewind.

[0075] It is understandable that the life cycle of a piece of music mainly includes the following three stages: Part 1: How is the music started? That is, how is music playback initiated? For example, using the steering wheel to control the next song; automatically playing the next song after one song has started, etc.

[0076] Section Two: User Behavior During Music Playback. For example, a user can use voice control to fast forward the first song, or the user can like, favorite, or add the currently playing music to the playlist within the application interface.

[0077] Step 3: How is music playback stopped? For example, clicking "next song" on the application interface, or using voice commands to play other music.

[0078] The user behavior corresponding to stage one is the initiation behavior, the user behavior corresponding to stage two is the state transition behavior, and the user behavior corresponding to stage three is the termination behavior. Among them, the initiation behavior includes the user operation behavior used to trigger the start of playing the first music, the state transition behavior includes the user operation behavior used to change the playback status and / or preference status of the first music during playback, and the termination behavior includes the user operation behavior used to end the playback of the first music.

[0079] 3202: Determine the intent type of the operation in the operation behavior data; wherein, the intent type includes positive intent and / or negative intent, positive intent is used to represent that the user likes the first music, and negative intent is used to represent that the user does not like the first music.

[0080] The following example illustrates the relationship between user actions and intent types during the playback of a song. For example, Figure 4 The diagram shown is an exemplary embodiment of this application illustrating the user's operational behavior and intent type during the playback of a song.

[0081] like Figure 4 As shown, in the music playback initiation methods, the following actions are identified as positive intentions: 1. The previous song triggered the start, and the playback duration is close to the total duration XX.

[0082] 2. The user requests to start playback via voice, and the playback duration is close to the total duration XX.

[0083] 3. The user selects to start playback, and the playback duration is close to the total duration XX.

[0084] "Approximately total duration XX" means that the playback duration is greater than the first duration, which can be set in advance according to actual needs.

[0085] The following actions are identified as having unknown intent: 1. The start of the next track: The reason the current song is being played is because the 'next track' command is triggered, and the system switches to it according to the playback order. The "start" action of this song is not because the user presses the play button, nor is it random interruption, but is triggered by the event that "after the previous song ends, the system automatically switches to the next song in the queue".

[0086] 2. The next song will start automatically when the previous song finishes playing naturally, when you drag the progress bar to the end of the previous song, or when you fast forward to the end of the previous song.

[0087] During music playback, the following actions are identified as positive intentions: 1. Users can like, favorite, or add to the playlist of the currently playing music. (This refers to user actions that change the playback status and / or preference status of the first song during its playback.) 2. Drag the progress bar or fast-forward to the beginning, and the playback time is close to the total duration.

[0088] 3. Enable single-track loop, and the number of loops is greater than XX.

[0089] The following actions are identified as having negative intent: 1. To dislike or downvote the currently playing music.

[0090] 2. Drag the progress bar or fast-forward to the end, and the playback time is less than XX of the total duration.

[0091] The following actions are identified as cancellation intentions: This function allows you to unlike, favorite, or add the currently playing music to the playback queue.

[0092] The following actions are identified as unknown / complex intents: Repeatedly switch modes, drag the progress bar multiple times, and perform fast forward and rewind operations on the currently playing music.

[0093] Among the methods used to end music playback, the following actions are identified as negative intentions: 1. Manually play the next song, and the current song playback progress is less than XX. 2. Drag the progress bar of the current music. When the current music has reached the end, the next song will play automatically, and the current music's playback time is less than XX of the total duration.

[0094] The following actions are identified as having unknown intent: The user can play other music through the interface.

[0095] In some embodiments, if the operational behavior data contains an operational behavior to be confirmed that is a first target operational behavior, then the intent type of the operational behavior to be confirmed is determined to be an explicit positive intent; if the operational behavior data contains an operational behavior to be confirmed that is a second target operational behavior, then the intent type of the operational behavior to be confirmed is determined to be a predicted positive intent; if the operational behavior data contains an operational behavior to be confirmed that is a third target operational behavior, then the intent type of the operational behavior to be confirmed is determined to be an explicit negative intent; if the operational behavior data contains an operational behavior to be confirmed that is a fourth target operational behavior, then the intent type of the operational behavior to be confirmed is determined to be a predicted negative intent.

[0096] Specifically, in some embodiments, the first target operation may include any one or more of the following: setting the first music to a "like" status, liking the first music, adding the first music to a favorites list, or adding the first music to the playback queue.

[0097] Specifically, in some embodiments, the second target operation line may include any one or more of the following: rewinding the playback progress of the first music to the beginning of the first music and the playback duration of the first music is greater than a first duration; and starting single-track looping of the first music and the number of single-track loops is greater than a preset number (e.g., 2 times). It is understood that rewinding the playback progress of the first music to the beginning of the first music and the playback duration of the first music is greater than the first duration means dragging the progress bar or fast-forwarding to the beginning, and the playback duration is close to the total duration.

[0098] Specifically, in some embodiments, the third target operation behavior may include any one or more of setting the first music to a dislike state and setting the first music to a down state.

[0099] Specifically, in some embodiments, the fourth target operation may include any one or more of the following: manually playing the next song and the playback duration of the first song is less than the second duration; and the user manually dragging the progress bar of the first song and the playback duration of the first song is less than the third duration. "The user manually dragging the progress bar of the first song and the playback duration of the first song is less than the third duration" means that the user manually drags the progress bar of the current song until it reaches the end, automatically plays the next song, and the playback duration of the current song is less than the total duration XX.

[0100] 3203: Determine the user's liking level for the first piece of music based on the intent type of the operation behavior in the operation behavior data.

[0101] In this embodiment of the application, the cycle of a piece of music includes a start phase, a playback phase, and an end playback phase. Based on the user's operation behavior in each of the aforementioned phases, the degree of the user's liking for a piece of music can be comprehensively considered, avoiding the judgment of the user's liking for a piece of music based on a single user operation behavior, reducing the misjudgment rate of the user's liking for a song, and improving the accuracy of determining the user's liking for a song.

[0102] In step 3203 above, determining the user's liking level for the first piece of music based on the intent type of the operation behavior in the operation behavior data can be achieved through the following steps 32031 to 32033: 32031: Determine the target operation behavior based on the intent type of the operation behavior in the operation behavior data; wherein, the target operation behavior includes operation behaviors with an intent type of positive intent or negative intent.

[0103] It is understandable that positive intent is used to represent that a user likes the first song, while negative intent is used to represent that a user dislikes the first song. The higher the intent score, the higher the user's liking for the first song. The intent score corresponding to positive intent is greater than that corresponding to negative intent. Therefore, different intent scores can be assigned to different types of operations within a song's cycle, and various user operations within a song's cycle can more accurately determine a user's liking for a song.

[0104] Furthermore, it can be understood that within the cycle of a piece of music, a user's action at one moment might cancel an action performed at a previous moment. The actions at these two moments cancel each other out, effectively meaning the user did not perform the action performed at the previous moment. Therefore, it is unnecessary to determine the intent score for these two actions, saving the step and computational effort of determining intent scores and improving the speed and power consumption of determining the user's music intensity value. Specifically, in some embodiments, if at the first moment there is a first action with an intent type of positive or negative, and there is no action to cancel the first action between the first moment and the end of the first music playback, then the first action is determined as the target action.

[0105] 32032: Determine the intent score for each behavior in the target operation; where, the higher the intent score, the higher the user's liking for the first music, and the intent score corresponding to positive intent is greater than the intent score corresponding to negative intent.

[0106] It is understood that in some embodiments, if a piece of music is played multiple times within a historical time period, the intent score can be determined based on the operation behavior during the most recent playback of that music. However, the determination of the intent score is influenced by the operation behavior during previous playbacks of that music. Specifically, if the intent type of an operation behavior during the most recent playback of that music is a positive intent, and if this operation behavior occurs multiple times within a historical time period, it indicates that the user's accumulated liking for that music is high, and the intent score corresponding to this operation behavior is set to be higher than the intent score of the previous operation behavior. If the intent type of an operation behavior during the most recent playback of that music is a positive intent, and if this operation behavior occurs multiple times within a historical time period, it indicates that the user's accumulated liking for that music is decreasing, and the intent score corresponding to this operation behavior is set to be lower than the intent score of the previous operation behavior.

[0107] It is understood that in some embodiments, if a piece of music is played multiple times within a historical time period, the intent score can be determined based on the operation behavior during the most recent playback of the music, and the determination of the intent score is not affected by the operation behavior during previous playbacks of the music.

[0108] 32033: Determine the user's liking level for the first piece of music based on the intent score of each behavior in the target operation.

[0109] In other embodiments, positive intentions can be further subdivided into explicit positive intentions and predicted positive intentions, and negative intentions can be further subdivided into explicit negative intentions and predicted negative intentions. Furthermore, the intention scores corresponding to the aforementioned four types of intentions are different, which can improve the accuracy of intention score determination.

[0110] Specifically, positive intentions include explicit positive intentions and predicted positive intentions, and negative intentions include explicit negative intentions and predicted negative intentions. The intention score for each behavior in the target operational behavior is determined as follows: If the intention type of the operational behavior to be confirmed in the operational behavior data is explicit positive intention, then the intention score of the operational behavior to be confirmed is determined as a first value; if the intention type of the operational behavior to be confirmed in the operational behavior data is predicted positive intention, then the intention score of the operational behavior to be confirmed is determined as a second value; wherein, the first and second values ​​are greater than zero, and the second value is less than the first value; if the intention type of the operational behavior to be confirmed in the operational behavior data is explicit negative intention, then the intention score of the operational behavior to be confirmed is determined as a third value; if the intention type of the operational behavior to be confirmed in the operational behavior data is predicted negative intention, then the intention score of the operational behavior to be confirmed is determined as a fourth value; wherein, the fourth value is less than zero, and the third value is greater than the third value.

[0111] In the embodiments of this application, the operational behavior can be formed in various ways. Media operation methods in the field of vehicle cockpits mainly include three types: voice operation, steering wheel operation, and operation within the application interface. That is, the operational behavior data includes any one or more of the following: user behavior data of controlling the first music track via voice, user behavior data of controlling the first music track via steering wheel, and user behavior data of controlling the first music track via the interface.

[0112] In some embodiments, the multimedia unified service module can be used to control all the aforementioned media applications, and can also collect user operation behavior, and accurately determine the user's liking for music based on the user's operation behavior. The specific implementation of determining the user's liking for music based on the user's operation behavior is described above and will not be repeated here.

[0113] Figure 5 The diagram illustrates a process for obtaining user actions during the playback of a song, as provided in an exemplary embodiment of this application. Figure 5 As shown in Figure (a), after the multimedia agent registers with the unified multimedia service to listen for user behavior, media information, and playback status callbacks, the user uses voice or square control. The voice / square control module sends a playback control request to the unified multimedia service module. The unified multimedia service module analyzes the command, saves the user behavior, and sends a playback control request to the media application. The playback control request is used to request the corresponding media playback control interface. After receiving the media information or playback status, the unified multimedia service, combined with the saved user behavior, sends a callback notification to the multimedia agent to inform it of song information, playback status, and changes in user behavior.

[0114] It is understandable that "steering wheel control" refers to steering wheel control, and "playback control" refers to playback control. "User behavior" refers to the user's operational actions.

[0115] like Figure 5 As shown in Figure (b), the user directly controls the media application through the media application interface, while the media application notifies the multimedia unified service of the user's behavior. The media application processes the playback control commands, triggering callbacks for media information or playback status changes. After receiving the media information or playback status, the multimedia unified service module, combined with the saved user behavior, then sends callback notifications to the multimedia agent regarding song information, playback status, and user behavior changes.

[0116] At this point, the multimedia intelligent agent can obtain a series of user actions that occur throughout the lifecycle of the music.

[0117] In some embodiments, the target mapping relationship described above can be determined by a model. Specifically, the road scene type and user's operation behavior data when playing the first music are input into a trained model. The trained model can then use any of the above embodiments to determine the user's liking value for the first music, and thus determine the correspondence between the media information of the first music, the user's liking value for the first music, and the road scene type when playing the first music.

[0118] Furthermore, in some embodiments, each time a piece of music is started, the trained model can be triggered to re-determine the correspondence between the media information of the first piece of music being played, the user's liking value for the first piece of music, and the road scene type in which the first piece of music is played, that is, update the correspondence and improve the accuracy of the correspondence.

[0119] In other embodiments, the music to be recommended to the user can be determined by a model. Specifically, triggered by the user opening a music application, the vehicle begins to determine the current road scene type. After determining the current road scene type, it can be input into a pre-trained model to obtain the target music to be recommended to the user, output by the pre-trained model.

[0120] The trained model can determine target music that matches the current road scene type and has a liking value greater than or equal to a target threshold from the target mapping relationship based on the current road scene type. The target mapping relationship includes the media information of the first music played in the vehicle's historical time period, the user's liking value of the first music, and the correspondence between the road scene type in which the first music was played. The model outputs the target music to be recommended to the user.

[0121] The trained model can be a Large Language Model (LLM).

[0122] Furthermore, in some embodiments, the trained model can determine target music that matches the current road scene type and has a liking value greater than or equal to a target threshold from the target mapping relationship based on the current road scene type. The target mapping relationship includes media information of the first music played during the vehicle's historical time period, the user's liking value for the first music, and the correspondence between the road scene types on which the first music was played. The model outputs target music to be recommended to the user. Then, it further determines music similar to the target music. Music similar to the target music includes music with the same attribute information as the target music, including any one or more of the target music's timbre and musical style. Finally, it recommends the target music and music similar to the target music to the user.

[0123] In this embodiment, the music that a user prefers in the current road scenario can be not only music that the user has heard before, but also music similar to music that the user has liked before. This can broaden the range of music that the user likes to listen to and improve the user experience.

[0124] It is understood that, in some embodiments, a trained model can integrate the ability to determine the target music to be recommended to the user based on the current road scene type, and the correspondence between the media information of the first music to be played, the user's liking value for the first music, and the road scene type on which the first music is played, based on the road scene type and user behavior data. In other embodiments, the aforementioned two capabilities can also be implemented using different pre-trained models.

[0125] The trained model can contain prompts to control the target music to be recommended to the user.

[0126] For example, the prompt for a character could be: You are a professional music recommendation engine that recommends songs based on the current road scene and your historical listening preferences in this scene.

[0127] Hints for specific characters can be: Skill 1: Comprehensive analysis of road scenarios, conducting all-round and detailed analysis to determine the driving characteristics under such road scenarios and which genres of songs are suitable to listen to.

[0128] Skill 2: Comprehensive analysis of historical listening preferences, accurately extracting key elements such as the style, artist, and album of songs the user has previously liked; the format of historical listening preferences is: data class LLMMediaInput( val title: String, / / Media title val artist: String?, / / Singer, actor, or author name val album: String?, / / Album name val preferenceLevel:Int=0, / / Preference level, -100 to 100 ) It is understood that the aforementioned preference level is the liking value provided in the embodiments of this application.

[0129] Skill 3: Precise Song Recommendation 1. Prioritize recommending similar songs with high preference for the current road scenario, and do not recommend songs with low preference. If there are multiple matching results, sort them by matching degree, and return a maximum of 10 songs.

[0130] 2. The output songs must actually exist and be searchable.

[0131] 3. Return the song list in the following fixed and standardized format: data class LLMMediaOutput( val title:String, / / Media title val artist: String?, / / Singer, actor, or author name val album: String?, / / Album name val musicSource:String?, / / QQ Music, NetEase Cloud Music, Baidu Music, Xiami Music, Migu Music, Kugou Music, Kuwo Music val confidenceLevel:Int=0, / / Confidence level, 0 to 100 val reason:String? / / Recommendation reason ) Understandably, the aforementioned credibility could refer to the credibility of recommendations for other music similar to the target music.

[0132] The cue words related to the format of the pre-trained model's inference and output of the target music to be recommended can be: 1. The interaction should focus solely on the current road scenario and recommending songs based on historical listening preferences within that scenario. Unrelated topics should not be addressed.

[0133] 2. The output content must strictly follow the given format and must not deviate from the framework requirements in any way.

[0134] 3. The reasons for recommendation should be concise, limited to 50 words, and clearly explain the basis for the recommendation.

[0135] Furthermore, in some embodiments, the aforementioned trained model possesses the ability to learn independently, becoming increasingly accurate in its music recommendations as it learns. For example, Figure 6 The diagram shown is a schematic representation of the model self-learning process provided in an exemplary embodiment of this application. Figure 6 As shown, the multimedia intelligent agent parsing model returns a recommendation list, from which songs with high credibility are selected.

[0136] Then, the multimedia agent searches the unified media service module to see if any recommended songs with high information content exist. Next, the multimedia agent collects the existing recommended songs, generates a playback queue, and requests playback from the media service sequentially. Then, the multimedia agent continuously monitors user listening behavior and, according to the schemes in the above embodiments, updates its inference of user preference for recommended songs. The multimedia agent updates the song preference database and updates user prompts in the next model flow, achieving self-learning; the more it learns, the more accurate it becomes.

[0137] Exemplary device Figure 7 The diagram shown is a structural schematic of a music recommendation device provided in an exemplary embodiment of this application. Figure 7 As shown, the music recommendation device 700 includes: an acquisition module 710, a matching module 720, and a recommendation module 730.

[0138] The acquisition module 710 is used to determine the current road scene type; wherein, the road scene type is used to characterize the vehicle's driving conditions; the matching module 720 is used to determine, based on the current road scene type, target music that matches the current road scene type and has a liking value greater than or equal to a target threshold from the target mapping relationship; wherein, the target mapping relationship includes the media information of the first music played in the vehicle's historical time period, the user's liking value for the first music, and the correspondence between the road scene type in which the first music was played, and the media information includes at least the name information of the first music, and the higher the liking value, the higher the user's liking for the music; the recommendation module 730 is used to recommend target music to the user.

[0139] This application provides a music recommendation device. The music recommendation device can determine the degree of liking for a certain music based on the user's various operation behaviors on the music from the start to the end of the music playback. The determination of the user's favorite music is more accurate. Furthermore, the vehicle can more accurately obtain the music that the user likes in the current road scene based on the degree of liking for each music played in the past time period and the road scene, thereby improving the accuracy of music recommendation.

[0140] This application also provides a computer-readable storage medium storing a computer program for executing the music recommendation method provided in any of the above embodiments.

[0141] This application also provides a computer program product, which includes a computer program. When the computer program is executed by the processor of a computer device, it enables the computer device to execute the music recommendation method or model training method provided in any of the above embodiments.

[0142] All of the above-mentioned optional technical solutions can be combined in any way to form the optional embodiments of this application, and will not be described in detail here.

[0143] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0144] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0145] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0146] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0147] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0148] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program verification codes, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0149] It should be noted that in the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0150] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0151] The above are merely preferred embodiments of this application and are not intended to limit this application. Any modifications or equivalent substitutions made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A music recommendation method, characterized in that, The method is applied to a vehicle, and the method includes: Determine the current road scene type; wherein the road scene type is used to characterize the driving conditions of the vehicle; Based on the current road scene type, target music that matches the current road scene type and has a liking value greater than or equal to a target threshold is determined from the target mapping relationship; wherein, the target mapping relationship includes the media information of the first music played during the vehicle's historical time period, the user's liking value for the first music, and the correspondence between the road scene types in which the first music was played; the liking value is determined based on the user's operational behavior towards the first music from the start to the end of playing the first music; the media information includes at least the name information of the first music, and the higher the liking value, the higher the user's liking for the music; Recommend the target music to the user.

2. The music recommendation method according to claim 1, characterized in that, Before determining the current road scene type, the method further includes: If the vehicle is detected playing the first music within a historical time period, determine the road scene type when the vehicle is playing the first music; After the first music finishes playing, determine the user's level of liking for the first music; The target mapping relationship stores the media information of the first music, the user's liking value for the first music, and the correspondence between the road scene types when the first music is played.

3. The music recommendation method according to claim 2, characterized in that, Determining the user's level of liking for the first music includes: Acquire user operation behavior data for the first music; wherein, the operation behavior data includes any one or more of the following: start behavior, state transition behavior, and end behavior; the start behavior includes user operation behavior for triggering the start of playback of the first music; the state transition behavior includes user operation behavior for changing the playback state and / or preference state of the first music during playback; and the end behavior includes user operation behavior for ending playback of the first music. Determine the intent type of the operation behavior in the operation behavior data; wherein, the intent type includes positive intent and / or negative intent, the positive intent being used to represent that the user likes the first music, and the negative intent being used to represent that the user dislikes the first music; Based on the intent type of the operation behavior in the operation behavior data, the user's liking level for the first music is determined.

4. The music recommendation method according to claim 3, characterized in that, Determining the user's liking level for the first music based on the intent type of the operation behavior in the operation behavior data includes: Based on the intent type of the operational behavior in the operational behavior data, the target operational behavior is determined; wherein, the target operational behavior includes operational behaviors with an intent type of positive intent or negative intent; Determine the intent score for each behavior in the target operation behavior; wherein, the higher the intent score, the higher the user's liking for the first music, and the intent score corresponding to the positive intent is greater than the intent score corresponding to the negative intent; Based on the intent score of each behavior in the target operation, the user's liking level for the first music is determined.

5. The music recommendation method according to claim 4, characterized in that, Determining the target operation behavior based on the intent type of the operation behavior in the operation behavior data includes: If at the first moment there is a first operation with a positive or negative intent, and there is no action to cancel the first operation between the first moment and the moment when the first music ends playing, then the first operation is determined as the target operation.

6. The music recommendation method according to claim 4, characterized in that, The positive intent includes explicit positive intent and predicted positive intent, and the negative intent includes explicit negative intent and predicted negative intent. Determining the intent score for each behavior in the target operational behavior includes: If the intent type of the operation behavior to be confirmed in the operation behavior data is a clear positive intent, then the intent score of the operation behavior to be confirmed is determined as the first value. If the intention type of the operation behavior to be confirmed in the operation behavior data is predicted as positive intention, then the intention score of the operation behavior to be confirmed is determined as the second value; wherein, the first value and the second value are greater than zero, and the second value is less than the first value; If the intention type of the operation behavior data to be confirmed is a clear negative intention, then the intention score of the operation behavior to be confirmed will be determined as the third value. If the intention type of the operation behavior to be confirmed in the operation behavior data is predicted to be negative, the intention score of the operation behavior to be confirmed is determined as the fourth value; wherein the fourth value is less than zero, and the third value is greater than the third value.

7. The music recommendation method according to claim 6, characterized in that, Determining the intent type of the operation behavior in the operation behavior data includes: If the operation behavior data contains an operation behavior to be confirmed that is a first target operation behavior, then the intent type of the operation behavior to be confirmed is determined as a clear positive intent; wherein, the first target operation behavior includes any one or more of setting the first music to a favorite status, liking the first music, collecting the first music, and adding the first music to the playback queue; If there is an operation behavior to be confirmed in the operation behavior data that is a second target operation behavior, then the intent type of the operation behavior to be confirmed is determined as a predicted positive intent; wherein, the second target operation behavior includes any one or more of the following: rewinding the playback progress of the first music to the beginning of the first music and the playback duration of the first music is greater than a first duration, and starting the single loop of the first music and the number of times the single loop is greater than a preset number; If the operation behavior data contains an operation behavior to be confirmed that is a third target operation behavior, then the intent type of the operation behavior to be confirmed is determined to be a clear negative intent; wherein, the third target operation behavior includes any one or more of setting the first music to a dislike state and setting the first music to a down state; If there is an operation behavior to be confirmed in the operation behavior data that is a fourth target operation behavior, then the intent type of the operation behavior to be confirmed is determined as a predicted negative intent; wherein, the fourth target operation behavior includes any one or more of the following: manually playing the next song and the playback duration of the first song is less than the second duration, and the user manually dragging the progress bar of the first song and the playback duration of the first song is less than the third duration.

8. The music recommendation method according to claim 1, characterized in that, After determining the target music from the target mapping relationship that matches the current road scene type and has a liking value greater than or equal to the target threshold, the method further includes: Identify music similar to the target music; wherein, the music similar to the target music includes music with the same attribute information as the target music, the attribute information including any one or more of the timbre and musical style of the target music; Recommending the target music to the user includes: Recommend the target music and music similar to the target music to the user.

9. A vehicle, characterized in that, include: processor; Memory used to store the processor's executable instructions. The processor is used to execute the music recommendation method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program for executing the music recommendation method according to any one of claims 1 to 8.

Images