Music recommendation method, device and vehicle based on car audio

Abstract

The present invention discloses a method, device, and vehicle for recommending music based on an in-car audio system. The method comprises: obtaining each user's historical behavior data for each piece of music in a music library; calculating each preference score of the current user for each piece of music based on each piece of historical behavior data; determining each audio feature of each piece of music and obtaining configuration features of the in-car audio system; preprocessing the configuration features and each audio feature to obtain a configuration feature vector and each audio feature vector of each piece of music; calculating each first matching score between the current user and each piece of music based on the configuration feature vector, each audio feature vector, and each preference score; and sorting all first matching scores from high to low, recommending music with a first matching score within a preset sorting range to the in-car system. The embodiments of the present invention can be used in conjunction with an in-car audio system to recommend music that meets the requirements of the in-car audio system to the user, thereby improving the user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of in-vehicle music, and in particular to a music recommendation method and device based on in-vehicle audio, and a vehicle. Background Art

[0002] At present, with the rapid development of the Internet of Vehicles, more and more car companies are integrating music players into the central control screen of the car. Music playback has become the most frequently used leisure activity for drivers while driving.

[0003] When the driver starts the music player in the car, the existing technology mostly recommends in-car music based on user preferences, the user's environment, etc., ignoring the requirements of music quality for the in-car audio configuration. If the audio equipment configuration is poor, listening to content with better sound quality at this time will not obtain the optimal listening experience. If the audio equipment configuration is good but the music quality is poor, the advantages of the current vehicle audio configuration cannot be brought into play. Summary of the Invention

[0004] The present invention provides a music recommendation method, device and vehicle based on car audio to solve the problem that the existing technology ignores the requirements of music quality for in-car audio configuration. The present invention can combine with car audio equipment to recommend music that meets the requirements of the car audio equipment to users, thereby improving user experience.

[0005] To achieve the above objectives, an embodiment of the present invention provides a music recommendation method based on a car audio system, comprising:

[0006] Obtain each user's historical behavior data for each piece of music in the music library;

[0007] Calculate each preference score of the current user for each piece of music according to each piece of historical behavior data;

[0008] Obtaining configuration characteristics of the vehicle audio device and determining each audio characteristic of each of the music;

[0009] Preprocessing the configuration feature and each of the audio features to obtain a configuration feature vector and each audio feature vector of each of the music;

[0010] Calculating each first matching score between the current user and each piece of music based on the configuration feature vector, each of the audio feature vectors, and each of the preference scores;

[0011] All the first matching scores are sorted from high to low, and music with first matching scores within the preset sorting range is recommended to the car computer.

[0012] As an improvement to the above solution, the calculation of each preference score of the current user for each music according to each of the historical behavior data includes:

[0013] According to the TF-IDF algorithm, the weight score of each historical behavior data of the current user is calculated;

[0014] Calculating the current user's preference score for the music that has been listened to in the music library based on the weighted score;

[0015] Based on the similarity between users, find K other users from all users who have similar behaviors or preferences to the current user's historical behaviors, where K ≥ 1 and K is an integer;

[0016] The preference scores of K other users are used to calculate the current user's preference score for music that has not been heard in the music library.

[0017] As an improvement to the above solution, the audio features include at least sound quality features, instrument features, waveform features, and style features. The audio features are preprocessed to obtain an audio feature vector, including:

[0018] For each piece of music, generating a sound quality vector according to the sound quality characteristics, generating an instrument vector according to the instrument characteristics, generating a waveform vector according to the waveform characteristics, and generating a style vector according to the style characteristics;

[0019] For each piece of music, a weighted superposition calculation is performed using the sound quality vector, the instrument vector, the waveform vector, and the style vector to obtain each audio feature vector of each piece of music.

[0020] As an improvement to the above solution, the configuration features include at least speaker type features, speaker quantity features, and speaker position features. The configuration features are preprocessed to obtain a configuration feature vector, including:

[0021] generating a speaker type vector according to the speaker type feature, generating a speaker quantity vector according to the speaker quantity feature, and generating a speaker position vector according to the speaker position feature;

[0022] A weighted superposition calculation is performed on the speaker type vector, the speaker quantity vector, and the speaker position vector to obtain a configuration feature vector of the vehicle audio device.

[0023] As an improvement to the above solution, the step of calculating each first matching score between the current user and each piece of music based on the configuration feature vector, each of the audio feature vectors, and each of the preference scores includes:

[0024] Calculating a matching score between the configuration feature vector and each of the audio feature vectors to obtain a second matching score between the vehicle audio device and each of the music;

[0025] Based on each of the second matching scores and the preference score, each first matching score between the current user and each of the music is calculated.

[0026] As an improvement to the above solution, the step of calculating a matching score between the configuration feature vector and each of the audio feature vectors to obtain a second matching score between the vehicle audio device and each of the music includes:

[0027] The second matching score between the vehicle audio device and each piece of music is calculated according to the following formula:

[0028]

[0029] Where x i is the configuration feature vector of the car audio equipment, y i is the audio feature vector of the music, and i represents the i-th element of the corresponding feature vector.

[0030] As an improvement to the above solution, the calculation of each first matching score between the current user and each piece of music based on each second matching score and the preference score includes:

[0031] The first matching score between the current user and each piece of music is calculated according to the following formula:

[0032] F=s a A+s b B

[0033] Where s a is the second matching score between the car audio equipment and the music, A is the preset first weight, s b is the current user's music preference score, and B is the preset second weight.

[0034] To achieve the above-mentioned purpose, an embodiment of the present invention further provides a music recommendation device based on a car audio system, comprising:

[0035] A historical behavior data acquisition module is used to obtain each user's historical behavior data for each music in the music library;

[0036] A preference score calculation module, configured to calculate the current user's preference score for each piece of music based on each piece of historical behavior data;

[0037] a feature acquisition module, configured to acquire configuration features of the vehicle audio equipment and determine each audio feature of each of the music;

[0038] a feature vector calculation module, configured to pre-process the configuration feature and each of the audio features to obtain a configuration feature vector and each audio feature vector of each of the music;

[0039] A first matching score calculation module is configured to calculate a first matching score between the current user and each piece of music based on the configuration feature vector, each of the audio feature vectors, and each of the preference scores;

[0040] The music recommendation module is used to sort all the first matching scores from high to low and recommend music with first matching scores within a preset sorting range to the vehicle computer.

[0041] To achieve the above-mentioned purpose, an embodiment of the present invention further provides a vehicle, comprising a vehicle body and the above-mentioned music recommendation device based on the vehicle audio system.

[0042] Compared to existing technologies, the in-vehicle audio-based music recommendation method, device, and vehicle provided by the present invention calculates a match between the user's music preference score, the configuration characteristics of the in-vehicle audio equipment, and the audio characteristics of the music. This method calculates a first match score between the current user and each piece of music, and recommends music with a first match score within a preset ranking to the in-vehicle computer. This method addresses the fact that different in-vehicle audio equipment configurations produce different listening experiences for music of varying sound quality, recommending music that is suitable for the current user's current vehicle audio equipment configuration and offers superior listening quality. This method maximizes the advantages of the vehicle audio equipment and provides the user with a superior listening experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a flow chart of a music recommendation method based on car audio provided by an embodiment of the present invention;

[0044] Figure 2 This is a structural block diagram of a music recommendation device based on a car audio system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0046] See also Figure 1 , Figure 11 is a flowchart of a method for recommending music based on a car audio system according to an embodiment of the present invention. The method for recommending music based on a car audio system includes:

[0047] S1. Obtain the historical behavior data of each user for each piece of music in the music library;

[0048] S2. Calculate the current user's preference score for each piece of music based on each piece of historical behavior data;

[0049] S3, obtaining configuration features of the vehicle audio equipment, and determining each audio feature of each of the music;

[0050] S4. Preprocess the configuration feature and each of the audio features to obtain a configuration feature vector and each audio feature vector of each of the music;

[0051] S5. Calculate each first matching score between the current user and each piece of music based on the configuration feature vector, each of the audio feature vectors, and each of the preference scores;

[0052] S6. Sort all first matching scores and recommend music with first matching scores within a preset sorting range to the vehicle computer.

[0053] In an optional embodiment, the step S2 of calculating the current user's preference score for each piece of music based on each piece of historical behavior data includes:

[0054] S21. Calculate the weight score of each historical behavior data of the current user according to the TF-IDF algorithm;

[0055] Specifically, the weight score of each historical behavior data of the current user is calculated according to the following formula:

[0056] TF-IDF=TF*IDF

[0057]

[0058]

[0059] S22. Calculating the current user's preference score for music that has been listened to in the music library based on the weighted scores;

[0060] Specifically, the current user's preference score for the music that has been listened to in the music library is calculated according to the following formula:

[0061] score=K a A+K b B+K c C+K d D+Ke E

[0062] Where A, B, C, D, and E are the scores of the user's historical behaviors of switching songs, liking, collecting, looping, and canceling collections, respectively. K a , K b , K c , K d , K e are the weight scores of the corresponding operations respectively;

[0063] S23. Based on the similarity between users, find K other users from all users who have similar historical behaviors or preferences to the current user, where K ≥ 1 and K is an integer.

[0064] It can be understood that the Pearson correlation coefficient is used to represent the correlation between two users, and then K other users with similar behaviors or preferences to the historical behaviors of the current user are found from all users.

[0065] S24. Calculate the current user's preference score for music in the music library that the current user has not heard using the preference scores of K other users.

[0066] It can be understood that each preference score of the current user for each of the music is composed of the current user's preference score for the music that has been listened to in the music library and the current user's preference score for the music that has not been listened to in the music library.

[0067] In an optional embodiment, the audio features include at least sound quality features, instrument features, waveform features, and style features, and the audio features are preprocessed to obtain an audio feature vector, including:

[0068] For each piece of music, generating a sound quality vector according to the sound quality characteristics, generating an instrument vector according to the instrument characteristics, generating a waveform vector according to the waveform characteristics, and generating a style vector according to the style characteristics;

[0069] Specifically, the above-mentioned different dimensional features, namely music quality, musical instruments and their number, music waveform and music style, are embedded to obtain the embedding vector of each dimensional feature.

[0070] For each piece of music, a weighted superposition calculation is performed using the sound quality vector, the instrument vector, the waveform vector, and the style vector to obtain each audio feature vector of each piece of music.

[0071] It is worth noting that the sound quality vector, instrument vector, waveform vector, and style vector are all two-dimensional vectors. Preferably, the corresponding weights are 50%, 10%, 20%, and 20%, respectively. These weights can be adjusted according to the effect.

[0072] For example, in the embodiment of the present invention, the sound quality characteristics are shown in Table 1:

[0073] Table 1 Sound quality characteristics of music

[0074] Sound quality characteristics illustrate 7.1.4 7 represents 7 ground channels, 1 represents 1 bass channel, and 4 represents 4 sky channels. 7.1.2 7 represents 7 ground channels, 1 represents 1 subwoofer channel, and 2 represents 2 overhead channels. 5.1 7 represents 7 ground channels, 1 represents 1 bass channel dual channel There are 2 channels

[0075] For example, in the embodiment of the present invention, the characteristics of musical instruments are shown in Table 2:

[0076] Table 2 Characteristics of musical instruments

[0077]

[0078] For example, in an embodiment of the present invention, the waveform characteristics are shown in Table 3:

[0079] Table 3 Waveform characteristics of music

[0080] Waveform characteristics illustrate frequency Frequency distribution of music amplitude The amplitude and volume changes of music

[0081] For example, in the embodiment of the present invention, the style features are shown in Table 4:

[0082] Table 4 Music style characteristics

[0083]

[0084]

[0085] In an optional embodiment, the configuration features include at least a speaker type feature, a speaker quantity feature, and a speaker position feature. The configuration features are preprocessed to obtain a configuration feature vector, including:

[0086] generating a speaker type vector according to the speaker type feature, generating a speaker quantity vector according to the speaker quantity feature, and generating a speaker position vector according to the speaker position feature;

[0087] Specifically, the above-mentioned different dimensional features, namely, speaker type, number of speakers and speaker position, are embedded to obtain an embedding vector of each dimensional feature.

[0088] A weighted superposition calculation is performed on the speaker type vector, the speaker quantity vector, and the speaker position vector to obtain a configuration feature vector of the vehicle audio device.

[0089] It is worth noting that the speaker type vector, speaker quantity vector, and speaker position vector are all two-dimensional vectors. Preferably, the corresponding weights are 40%, 20%, and 40%, respectively. These weights can be adjusted according to the effect.

[0090] For example, the configuration features in the embodiment of the present invention are shown in Table 5:

[0091] Table 5 Configuration characteristics of vehicle audio equipment

[0092]

[0093]

[0094] In an optional embodiment, the step S5 of calculating each first matching score between the current user and each piece of music based on the configuration feature vector, each of the audio feature vectors, and each of the preference scores includes:

[0095] S51, calculating a matching score between the configuration feature vector and each of the audio feature vectors to obtain a second matching score between the in-vehicle audio device and each of the music;

[0096] Specifically, each second matching score between the vehicle audio device and each piece of music is calculated according to the following formula:

[0097]

[0098] Where x i is the configuration feature vector of the car audio equipment, y i is the audio feature vector of the music, and i represents the i-th element of the corresponding feature vector.

[0099] It can be understood that n is the dimension of the configuration feature vector and the audio feature vector, the dimensions of the two feature vectors are the same, and i represents the i-th dimension, i.e., the element;

[0100] S52: Calculate each first matching score between the current user and each piece of music based on each of the second matching scores and the preference score.

[0101] Specifically, each first matching score between the current user and each piece of music is calculated according to the following formula:

[0102] F=s a A+s b B

[0103] Where s a is the second matching score between the car audio equipment and the music, A is the preset first weight, s b is the current user's music preference score, and B is the preset second weight.

[0104] Preferably, A=0.7, B=0.3.

[0105] An embodiment of the present invention provides a car audio-based music recommendation method that calculates a match between the user's music preference score, the configuration characteristics of the car audio equipment, and the audio characteristics of the music. This method calculates a first match score between the current user and each piece of music, and then recommends music with a first match score within a preset ranking to the car computer. This embodiment of the present invention addresses the fact that different car audio equipment configurations produce different listening experiences for music of different sound quality. By recommending music that is suitable for the current car audio equipment configuration and offers the best listening experience for the current user, the method maximizes the advantages of the car audio equipment and provides the user with the highest-quality listening experience.

[0106] See also Figure 2 , Figure 2 1 is a structural block diagram of a car audio-based music recommendation device 10 provided by an embodiment of the present invention. The car audio-based music recommendation device 10 includes:

[0107] The historical behavior data acquisition module 11 is used to obtain the historical behavior data of each user for each music in the music library;

[0108] a preference score calculation module 12, configured to calculate the current user's preference score for each piece of music based on each piece of historical behavior data;

[0109] A feature acquisition module 13 is used to determine each audio feature of each piece of music and to acquire configuration features of the vehicle audio equipment;

[0110] a feature vector calculation module 14 for preprocessing the configuration feature and each of the audio features to obtain a configuration feature vector and each audio feature vector of each of the music;

[0111] A first matching score calculation module 15 is configured to calculate a first matching score between the current user and each piece of music based on the configuration feature vector, each of the audio feature vectors, and each of the preference scores;

[0112] The music recommendation module 16 is configured to sort all first matching scores and recommend music with first matching scores within a preset sorting range to the vehicle computer.

[0113] It is worth noting that the working process of each module in the car audio-based music recommendation device 10 described in the embodiment of the present invention can refer to the working process of the car audio-based music recommendation method described in the above embodiment, and will not be repeated here.

[0114] An embodiment of the present invention provides a car audio-based music recommendation device 10. This device calculates a match between the user's music preference score, the car audio system's configuration features, and the music's audio characteristics to determine a first match score between the current user and each piece of music. The device then recommends music with a first match score within a preset ranking to the car's computer. This embodiment addresses the varying listening experiences of music with varying sound quality, as existing car audio systems vary in configuration. By recommending music that is suitable for the current user's car audio system and offers superior listening quality, the device maximizes the advantages of the car audio system and provides the user with a superior listening experience.

[0115] An embodiment of the present invention provides a vehicle, including a vehicle body and the vehicle audio-based music recommendation device 10 described in the above embodiment.

[0116] The specific working process of the music recommendation device 10 based on the car audio can refer to the working process of the music recommendation device 10 based on the car audio described in the above embodiment, and will not be repeated here.

[0117] An embodiment of the present invention provides a car audio-based music recommendation system. This system calculates a match between the user's music preference score, the configuration characteristics of the car audio system, and the audio characteristics of the music. This system then determines a first match score between the current user and each piece of music, recommending music with a first match score within a preset order to the car computer. This system addresses the varying listening experiences of music with varying sound quality across existing car audio systems, recommending music that is suitable for the current user's current configuration and offers superior listening quality. This system maximizes the advantages of the car audio system and provides the user with a superior listening experience.

[0118] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A music recommendation method based on car audio, characterized in that: include: Obtain each user's historical behavior data for each piece of music in the music library; Calculate each preference score of the current user for each piece of music according to each piece of historical behavior data; Determining each audio feature of each of the music and obtaining configuration features of the vehicle audio device; Preprocessing the configuration feature and each of the audio features to obtain a configuration feature vector and each audio feature vector of each of the music; Calculating each first matching score between the current user and each piece of music based on the configuration feature vector, each of the audio feature vectors, and each of the preference scores; sorting all the first matching scores from high to low, and recommending music with first matching scores within a preset sorting range to the vehicle computer; The step of calculating the current user's preference score for each piece of music based on each piece of historical behavior data includes: According to the TF-IDF algorithm, the weight score of each historical behavior data of the current user is calculated; Calculating the current user's preference score for the music that has been listened to in the music library based on the weighted score; Based on the similarity between users, find K other users from all users who have similar behaviors or preferences to the current user's historical behaviors, where K ≥ 1 and K is an integer; Use the preference scores of K other users to calculate the current user's preference score for music that has not been heard in the music library; The configuration features include at least speaker type features, speaker quantity features, and speaker position features. The configuration features are preprocessed to obtain a configuration feature vector, including: generating a speaker type vector according to the speaker type feature, generating a speaker quantity vector according to the speaker quantity feature, and generating a speaker position vector according to the speaker position feature; A weighted superposition calculation is performed on the speaker type vector, the speaker quantity vector, and the speaker position vector to obtain a configuration feature vector of the vehicle audio device.

2. The music recommendation method based on the car audio according to claim 1, characterized in that: The audio features include at least sound quality features, instrument features, waveform features, and style features. The audio features are preprocessed to obtain an audio feature vector, including: For each piece of music, generating a sound quality vector according to the sound quality characteristics, generating an instrument vector according to the instrument characteristics, generating a waveform vector according to the waveform characteristics, and generating a style vector according to the style characteristics; For each piece of music, a weighted superposition calculation is performed using the sound quality vector, the instrument vector, the waveform vector, and the style vector to obtain each audio feature vector of each piece of music.

3. The music recommendation method based on the car audio according to claim 1, characterized in that: The step of calculating each first matching score between the current user and each piece of music based on the configuration feature vector, each of the audio feature vectors, and each of the preference scores includes: Calculating a matching score between the configuration feature vector and each of the audio feature vectors to obtain a second matching score between the vehicle audio device and each of the music; Based on each of the second matching scores and the preference score, each first matching score between the current user and each of the music is calculated.

4. The music recommendation method based on the car audio according to claim 3, characterized in that: The calculating the matching score between the configuration feature vector and each of the audio feature vectors to obtain each second matching score between the vehicle audio device and each of the music includes: The second matching score between the vehicle audio device and each piece of music is calculated according to the following formula: Where x i is the configuration feature vector of the car audio equipment, y i is the audio feature vector of the music, and i represents the i-th element of the corresponding feature vector.

5. The music recommendation method based on the car audio according to claim 3, characterized in that: The step of calculating each first matching score between the current user and each piece of music based on each of the second matching scores and the preference score includes: The first matching score between the current user and each piece of music is calculated according to the following formula: F=s a A+s b B Where s a is the second matching score between the car audio equipment and the music, A is the preset first weight, s b is the current user's music preference score, and B is the preset second weight.

6. A music recommendation device based on car audio, characterized in that: include: A historical behavior data acquisition module is used to obtain each user's historical behavior data for each music in the music library; A preference score calculation module, configured to calculate the current user's preference score for each piece of music based on each piece of historical behavior data; a feature acquisition module, configured to acquire configuration features of the vehicle audio equipment and determine each audio feature of each of the music; a feature vector calculation module, for determining each audio feature of each of the music and obtaining configuration features of the vehicle audio equipment; A first matching score calculation module is configured to calculate a first matching score between the current user and each piece of music based on the configuration feature vector, each of the audio feature vectors, and each of the preference scores; a music recommendation module, configured to sort all the first matching scores from high to low, and recommend music with first matching scores within a preset sorting range to the vehicle computer; The step of calculating the current user's preference score for each piece of music based on each piece of historical behavior data includes: According to the TF-IDF algorithm, the weight score of each historical behavior data of the current user is calculated; Calculating the current user's preference score for the music that has been listened to in the music library based on the weighted score; Based on the similarity between users, find K other users from all users who have similar behaviors or preferences to the current user's historical behaviors, where K ≥ 1 and K is an integer; Use the preference scores of K other users to calculate the current user's preference score for music that has not been heard in the music library; The configuration features include at least speaker type features, speaker quantity features, and speaker position features. The configuration features are preprocessed to obtain a configuration feature vector, including: generating a speaker type vector according to the speaker type feature, generating a speaker quantity vector according to the speaker quantity feature, and generating a speaker position vector according to the speaker position feature; A weighted superposition calculation is performed on the speaker type vector, the speaker quantity vector, and the speaker position vector to obtain a configuration feature vector of the vehicle audio device.

7. A vehicle, characterized in that: The invention comprises a vehicle body and the music recommendation device based on the vehicle audio system as claimed in claim 6.

Images