Song singing difficulty determination method, electronic device and storage medium

Abstract

The present application discloses a method for determining the difficulty of singing a song, an electronic device and a computer-readable storage medium, and the method for determining the difficulty of singing a song includes: obtaining training songs and corresponding manual scoring information, extracting audio features of the training songs; obtaining the number of recordings and the number of releases of the training songs, and calculating the ratio of the number of releases to the number of recordings as the release rate information of the training songs; weighting the release rate information and the manual scoring information of the training songs to calculate the singing difficulty label corresponding to the training songs; training a machine learning model using the audio features of the training songs and the corresponding singing difficulty label; determining a target song that needs to be evaluated for singing difficulty, and extracting the audio features of the target song; inputting the audio features of the target song into the trained machine learning model to obtain the singing difficulty score of the target song. It can be seen that the present application improves the accuracy of determining the difficulty of singing a song.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and more specifically, to a method for determining the difficulty of singing a song, an electronic device, and a computer-readable storage medium. Background Art

[0002] With the development of network technology, many applications for karaoke have emerged, through which users can select corresponding songs to sing and record. However, due to the existence of songs of various styles, the singing difficulty of each song is also different. In order to facilitate users to understand the singing difficulty of songs when singing karaoke, so as to select songs suitable for them, and at the same time, in order to reasonably judge the songs sung by users, it is necessary to determine the singing difficulty of songs.

[0003] In the related art, the singing difficulty of songs is manually labeled. However, since manual labeling relies on people's subjective judgment, different subjective judgment standards lead to different singing difficulties given by different people, which ultimately leads to inaccurate singing difficulty of the song.

[0004] Therefore, how to improve the accuracy of determining the difficulty of singing a song is a technical problem that technicians in this field need to solve. Summary of the invention

[0005] The purpose of the present application is to provide a method for determining the difficulty of singing a song, an electronic device and a computer-readable storage medium, so as to improve the accuracy of determining the difficulty of singing a song.

[0006] To achieve the above-mentioned purpose, the first aspect of the present application provides a method for determining the difficulty of singing a song.

[0007] Obtaining training songs and corresponding manual scoring information, and extracting audio features of the training songs;

[0008] Obtaining the number of recordings and the number of releases of the training song, and calculating the ratio of the number of releases to the number of recordings as the release rate information of the training song;

[0009] Weighting the release rate information and manual scoring information of the training songs to calculate the singing difficulty label corresponding to the training songs;

[0010] Training a machine learning model using the audio features of the training songs and corresponding singing difficulty labels;

[0011] Determine a target song for which singing difficulty needs to be evaluated, and extract audio features of the target song;

[0012] The audio features of the target song are input into the trained machine learning model to obtain the singing difficulty score of the target song.

[0013] To achieve the above-mentioned purpose, the second aspect of the present application provides an electronic device, including:

[0014] Memory for storing computer programs;

[0015] A processor is used to implement the steps of the method for determining the difficulty of singing a song as described above when executing the computer program.

[0016] To achieve the above-mentioned purpose, the third aspect of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for determining the difficulty of singing a song as mentioned above are implemented.

[0017] It can be seen from the above scheme that a method for determining the difficulty of singing a song provided in the present application includes: obtaining training songs and corresponding manual scoring information, and extracting audio features of the training songs; obtaining the number of recordings and the number of releases of the training songs, and calculating the ratio of the number of releases to the number of recordings as the release rate information of the training songs; weighting the release rate information and manual scoring information of the training songs to calculate the singing difficulty label corresponding to the training songs; training a machine learning model using the audio features of the training songs and the corresponding singing difficulty labels; determining a target song whose singing difficulty needs to be evaluated, and extracting the audio features of the target song; and inputting the audio features of the target song into the trained machine learning model to obtain a singing difficulty score for the target song.

[0018] The method for determining the difficulty of singing a song provided by the present application calculates labels related to the difficulty of singing, and learns the relationship between different audio features and singing difficulty labels through machine learning, so that the process of predicting the difficulty of singing a song is objective and stable. Furthermore, the present application calculates labels related to the difficulty of singing based on the release rate information and manual scoring information of the song. The release rate information is an objective indicator of the song, and the manual scoring information is a subjective labeling of the difficulty of the song. The objective indicators and subjective labels of the song are combined to make the calculated singing difficulty label more accurate, thereby optimizing the training results of the machine learning model, so that the singing difficulty of the song determined by the trained machine learning model is more accurate. It can be seen that the present application improves the accuracy of determining the difficulty of singing a song. The present application also discloses a song display method, an electronic device, and a computer-readable storage medium, which can also achieve the above-mentioned technical effects.

[0019] It should be understood that the foregoing general description and the following detailed description are exemplary only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. The drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific implementation methods, they are used to explain the present disclosure, but do not constitute a limitation to the present disclosure. In the drawings:

[0021] Figure 1 An architectural diagram of a system for determining the difficulty of singing a song provided in an embodiment of the present application;

[0022] Figure 2 A flowchart of a method for determining the difficulty of singing a song provided in an embodiment of the present application;

[0023] Figure 3a-3n A schematic diagram of the relationship between various audio features of a song, release rate information, and manual scoring information provided in an embodiment of the present application;

[0024] Figure 4 A flowchart of a song display method provided in an embodiment of the present application;

[0025] Figure 5 A structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0027] In order to facilitate understanding of the method for determining the difficulty of singing a song provided by the present application, the system used by the method is introduced below. Figure 1 , which shows an architecture diagram of a system for determining the difficulty of singing a song provided in an embodiment of the present application, such as Figure 1 As shown, it includes a server 10 and a client 20.

[0028] The server 10 is used to determine the singing difficulty of the song. Specifically, it obtains the training song and the corresponding manual scoring information, extracts the audio features of the training song; obtains the number of recordings and the number of releases of the training song, and calculates the ratio of the number of releases to the number of recordings as the release rate information of the training song; weights the release rate information and the manual scoring information of the training song to calculate the singing difficulty label corresponding to the training song; trains the machine learning model using the audio features of the training song and the corresponding singing difficulty label; determines the target song that needs to be evaluated for singing difficulty, and extracts the audio features of the target song; inputs the audio features of the target song into the trained machine learning model to obtain the singing difficulty score of the target song. It is understandable that the server 10 can determine the difficulty level of each song in the song library in the above manner.

[0029] The client 20 may include a fixed terminal such as a PC (Chinese full name: Personal Computer, English full name: Personal Computer) and a mobile terminal such as a mobile phone, which is used to display songs of a difficulty level to be displayed. Specifically, when a song display command of a difficulty level to be displayed is received, the song name of the song of the difficulty level to be displayed is displayed.

[0030] The embodiment of the present application discloses a method for determining the difficulty of singing a song, which improves the accuracy of determining the difficulty of singing a song.

[0031] See also Figure 2 , Figure 2 A flowchart of a method for determining the difficulty of singing a song provided in an embodiment of the present application, such as Figure 2 As shown, including:

[0032] S101: Obtain training songs and corresponding manual scoring information, and extract audio features of the training songs;

[0033] The execution subject of this embodiment is the server in the above-mentioned song singing difficulty determination system. In this step, firstly, the training songs and corresponding manual scoring information for training the machine learning model are obtained. It should be noted that in order to avoid the interference of popular factors on the training results, songs released earlier can be selected as training songs, for example, songs released ten years ago can be selected as training songs.

[0034] Secondly, extract the audio features of the training songs, which may include range, pitch, high-pitched rate, low-pitched rate, long-note rate, breathing difficulty, pronunciation speed, etc. Specifically, the range is the difference between the highest pitch and the lowest pitch of the song; the overall pitch is Max{median pitch value - higher pitch threshold, 0}, wherein the higher pitch threshold for boys is different from that for girls, for example, the higher pitch threshold for boys may be G3, and the higher pitch threshold for girls may be F4; the treble rate is the ratio of the number of notes above the higher pitch threshold to the total number of notes; the bass rate is the ratio of the number of notes below the bass threshold to the total number of notes, wherein the bass threshold for boys is different from that for girls, for example, the bass threshold for boys may be F2, and the bass threshold for girls may be A3; the long note rate is the ratio of the number of notes whose duration exceeds the long note threshold to the total number of notes, and the long note threshold may be 3s; the breathing difficulty is the ratio of the total duration of continuous notes within the breathable threshold that is greater than or equal to the long continuous threshold to the number of breathable times, the breathable threshold may be 0.2s, and the long continuous threshold may be 10s; the pronunciation speed is the ratio of the total number of words in the lyrics to the total duration of the song, in words / second.

[0035] Furthermore, the extracted audio features can be standardized into corresponding threshold ranges. For example, the range can be standardized into the range of [12, 42], that is, when it is greater than 42, the value is 42, and when it is less than 12, the value is 12. The pitch can be standardized into the range of [0, 12]. The high-pitched rate and the low-pitched rate can be standardized into the range of [0, 0.3]. The long-pitched rate can be standardized into the range of [0, 0.03]. The breathing difficulty can be standardized into the range of [0, 0.3]. The breathing difficulty can be standardized into the range of [2, 5].

[0036] S102: Obtaining the number of recordings and the number of releases of the training song, and calculating the ratio of the number of releases to the number of recordings as release rate information of the training song;

[0037] In this step, the release rate information of the training songs is calculated, and the release rate information is specifically the ratio between the number of releases and the number of recordings of the training songs.

[0038] S103: Weighting the release rate information and manual scoring information of the training songs to calculate the singing difficulty label corresponding to the training songs;

[0039] It should be noted that the various audio features of a song are correlated with its release rate information and manual rating information, such as Figure 3a-3nAs shown, further, there is a correlation between the singing difficulty of a song and the audio features of the song. Therefore, the release rate information and the manual scoring information can reflect the singing difficulty of the song to a certain extent. That is, the label related to the singing difficulty can be calculated based on the release rate information and the manual scoring information of the training songs. The release rate information is an objective indicator of the song, and the manual scoring information is a subjective labeling of the difficulty of the song. The objective indicators and subjective labels of the song are combined to make the calculated label related to the singing difficulty of the song more accurate.

[0040] As a feasible implementation, this step may include: assigning a first weighting parameter and a second weighting parameter to the release rate information and the manual scoring information respectively; calculating the product of the release rate information and the first weighting parameter as the first product, and calculating the product of the manual scoring information and the second weighting parameter as the second product; and calculating the sum of the first product and the second product as the singing difficulty label corresponding to the training song. In a specific implementation, the calculation formula for the label corresponding to the training song is: release rate information × i + manual scoring information × j, where i is the first weighting parameter and j is the second weighting parameter.

[0041] S104: training a machine learning model using the audio features of the training song and the corresponding singing difficulty label;

[0042] In this step, the machine learning model is trained using the audio features of the training song and the corresponding singing difficulty labels, and the trained machine learning model is used to determine the singing difficulty of the target song. The machine learning model in this embodiment can be specifically a neural network model, a linear regression model, etc., which is not specifically limited here.

[0043] If the machine learning model is specifically a linear regression model, this step may include: inputting each audio feature of the training song and the corresponding singing difficulty label into the linear regression model to perform linear fitting on each audio feature of the training song and the corresponding singing difficulty label, and obtaining the weight coefficient corresponding to each audio feature of the training song. In a specific implementation, a linear regression model is used to perform linear fitting on each audio feature of the training song and the corresponding singing difficulty label, that is, to determine the weight coefficients a, b, c, d, e, f, g in the formula "singing difficulty label = a×range+b×overall pitch+c×high pitch rate+d×low pitch rate+e×long pitch rate+f×breathing difficulty+g×pronunciation speed". The linear regression model can make each audio feature explanatory through linear fitting, and can explicitly explain the importance of each audio feature through the weight coefficient corresponding to each audio feature. The audio feature with a larger weight coefficient is more important.

[0044] If the machine learning model is specifically a neural network model, this step may include: inputting the audio features of the training song into the neural network model, outputting the predicted singing difficulty score of the training song; calculating the loss value based on the predicted singing difficulty score of the training song and the corresponding singing difficulty label, and adjusting the model parameters of the neural network model according to the loss value, until the loss value is less than the first preset value or the number of adjustments reaches the second preset value, and a trained neural network model is obtained. In a specific implementation, the audio features of the training song are input into the neural network model, the predicted singing difficulty score is output, and the loss function is determined to calculate the loss value based on the predicted singing difficulty score and the singing difficulty label, and it is determined whether the loss value is less than the first preset value or the number of adjustments reaches the second preset value. If so, the training is completed, if not, the model parameters of the neural network model are adjusted, the audio features of the training song are input into the adjusted neural network model again, the loss value is calculated again, and the model parameters are adjusted until the loss value is less than the first preset value or the number of adjustments reaches the second preset value, and a trained neural network model is obtained.

[0045] S105: determining a target song for which singing difficulty needs to be evaluated, and extracting audio features of the target song;

[0046] S106: Input the audio features of the target song into the trained machine learning model to obtain a singing difficulty score of the target song.

[0047] In a specific implementation, for a target song that needs to be evaluated for singing difficulty, first, the target audio features of the target song are extracted, which may include range, pitch, high pitch rate, low pitch rate, long pitch rate, breathing difficulty, pronunciation speed, etc. Secondly, each extracted audio feature is standardized and standardized into a corresponding threshold range. Then, the standardized target audio features are input into the machine learning model trained in the previous embodiment, and the singing difficulty score of the target song is output.

[0048] If the machine learning model is specifically a linear regression model, the weight coefficient corresponding to each audio feature is determined during the training process. Therefore, in this embodiment, the audio features of the target song can be weighted based on the weight coefficients corresponding to each audio feature to obtain the singing difficulty score of the target song. For example, during the training process, the weight coefficients corresponding to the range, pitch, high pitch rate, low pitch rate, long pitch rate, breathing difficulty, and pronunciation speed are determined to be 0.88, 0.48, 0.53, 0.50, 0.31, 0.03, and 0.59, respectively. The singing difficulty score = 0.88 × range + 0.48 × overall pitch + 0.53 × high pitch rate + 0.50 × low pitch rate + 0.31 × long pitch rate + 0.03 × breathing difficulty + 0.59 × pronunciation speed.

[0049] If the machine learning model is specifically a neural network model, the audio features of the target song are input into the trained neural network model to obtain the singing difficulty score of the target song.

[0050] As a preferred implementation, after the target audio is input into the trained machine learning model to obtain the singing difficulty score of the target song, it also includes: determining the target scoring range to which the singing difficulty score of the target song belongs; determining the target difficulty level corresponding to the target scoring range according to the correspondence between the scoring range and the difficulty level; and determining the singing difficulty level of the target song as the target difficulty level. In a specific implementation, the correspondence between the scoring range and the difficulty level can be determined. For example, the scoring range [0, 0.2] corresponds to a simple difficulty level, the scoring range (0.2, 0.4] corresponds to an easy difficulty level, the scoring range (0.4, 0.6] corresponds to a medium difficulty level, the scoring range (0.6, 0.8] corresponds to a difficult difficulty level, and the scoring range (0.8~1.0] corresponds to a difficult difficulty level. Determine the target scoring range to which the singing difficulty score of the target song belongs, and then determine the singing difficulty level of the target song according to the correspondence between the above-mentioned scoring range and the difficulty level, that is, determine the target difficulty level corresponding to the target scoring range as the singing difficulty level of the target song.

[0051] The method for determining the difficulty of singing a song provided in the embodiment of the present application calculates labels related to the difficulty of singing, and learns the relationship between different audio features and singing difficulty labels through machine learning, so that the process of predicting the difficulty of singing a song is objective and stable. Furthermore, the embodiment of the present application calculates labels related to the difficulty of singing based on the release rate information and manual scoring information of the song. The release rate information is an objective indicator of the song, and the manual scoring information is a subjective labeling of the difficulty of the song. The objective indicators and subjective labels of the song are combined to make the calculated singing difficulty label more accurate, thereby optimizing the training results of the machine learning model, so that the singing difficulty of the song determined by the trained machine learning model is more accurate. It can be seen that the embodiment of the present application improves the accuracy of determining the difficulty of singing a song.

[0052] The present application embodiment discloses a song display method, see Figure 4 , Figure 4 A flowchart of a song display method provided in an embodiment of the present application, such as Figure 4 As shown, including:

[0053] S301: The server extracts audio features of each song in the song library;

[0054] S302: The server inputs the audio features of each song into the trained machine learning model to obtain a singing difficulty score of each song;

[0055] S303: The server determines the scoring range to which the singing difficulty score of each of the songs belongs, and determines the difficulty level of each of the songs according to the corresponding relationship between the scoring range and the difficulty level;

[0056] In this embodiment, the server may determine the difficulty level of each song in the song library in the manner described in the previous embodiment.

[0057] S304: The client receives a command to display songs of difficulty levels to be displayed;

[0058] S305: The client sends the song display command to the server;

[0059] S306: The server determines the song name of the song whose difficulty level is the difficulty level to be displayed;

[0060] S307: The server returns the song name of the song with the difficulty level to be displayed to the client;

[0061] S308: The client displays the song title of the song having the difficulty level to be displayed.

[0062] In a specific implementation, the user can select the difficulty level to be displayed on the client, the client receives a song display command for the difficulty level to be displayed, and sends the song display command to the server. The server determines the song name of the song with the difficulty level to be displayed, and returns it to the client, so that the client displays the song name of the song with the difficulty level to be displayed.

[0063] A device for determining the difficulty of singing a song provided in an embodiment of the present application is introduced below. The device for determining the difficulty of singing a song described below is applied to a server and can be referenced to the method for determining the difficulty of singing a song described above.

[0064] Specifically, the device for determining the difficulty of singing a song provided in the embodiment of the present application includes:

[0065] An acquisition module, used to acquire training songs and corresponding manual scoring information, and extract audio features of the training songs;

[0066] A first calculation module is used to obtain the number of recordings and the number of releases of the training song, and calculate the ratio of the number of releases to the number of recordings as the release rate information of the training song;

[0067] A second calculation module is used to weight the release rate information and manual scoring information of the training songs to calculate the singing difficulty label corresponding to the training songs;

[0068] A training module, used to train a machine learning model using the audio features of the training songs and corresponding singing difficulty labels;

[0069] An extraction module, used to determine a target song for which singing difficulty needs to be evaluated, and to extract audio features of the target song;

[0070] An input module is used to input the audio features of the target song into the trained machine learning model to obtain the singing difficulty score of the target song.

[0071] The device for determining the difficulty of singing a song provided in the embodiment of the present application calculates labels related to the difficulty of singing, and learns the relationship between different audio features and singing difficulty labels through machine learning, so that the process of predicting the difficulty of singing a song is objective and stable. Furthermore, the embodiment of the present application calculates labels related to the difficulty of singing based on the release rate information and manual scoring information of the song. The release rate information is an objective indicator of the song, and the manual scoring information is a subjective labeling of the difficulty of the song. The objective indicators and subjective labels of the song are combined to make the calculated singing difficulty label more accurate, thereby optimizing the training results of the machine learning model, so that the singing difficulty of the song determined by the trained machine learning model is more accurate. It can be seen that the embodiment of the present application improves the accuracy of determining the difficulty of singing a song.

[0072] Based on the above embodiments, as a preferred implementation mode, the training module is specifically used to: train the machine learning model using any one or a combination of the range, pitch, high-pitched rate, low-pitched rate, long-pitched rate, breathing difficulty, and pronunciation speed of the training songs, and the singing difficulty labels corresponding to the training songs.

[0073] Based on the above embodiments, as a preferred implementation mode, the training module is specifically used to: input each audio feature of the training song and the corresponding singing difficulty label into a linear regression model to perform linear fitting on each audio feature of the training song and the corresponding singing difficulty label to obtain the weight coefficient corresponding to each audio feature of the training song.

[0074] Based on the above embodiment, as a preferred implementation mode, the input module is specifically used to: perform weighted calculation on the audio features of the target song based on the weight coefficients corresponding to each audio feature to obtain a singing difficulty score of the target song.

[0075] Based on the above embodiments, as a preferred implementation mode, the training module is specifically used to: input the audio features of the training song into the neural network model, and output the predicted singing difficulty score of the training song; calculate the loss value based on the predicted singing difficulty score of the training song and the corresponding singing difficulty label, and adjust the model parameters of the neural network model according to the loss value until the loss value is less than a first preset value or the number of adjustments reaches a second preset value, thereby obtaining a trained neural network model.

[0076] Based on the above embodiments, as a preferred implementation mode, the second calculation module is specifically used to: assign a first weighting parameter and a second weighting parameter to the publication rate information and the manual scoring information respectively; calculate the product of the publication rate information and the first weighting parameter as the first product, and calculate the product of the manual scoring information and the second weighting parameter as the second product; calculate the sum of the first product and the second product as the singing difficulty label corresponding to the training song.

[0077] Based on the above embodiment, as a preferred implementation, it also includes:

[0078] A determination module is used to determine the target score range to which the singing difficulty score of the target song belongs; determine the target difficulty level corresponding to the target score range according to the correspondence between the score range and the difficulty level; and determine the singing difficulty level of the target song as the target difficulty level.

[0079] Based on the above embodiment, as a preferred implementation, it also includes:

[0080] The display module is used to determine the song title of the song with the difficulty level to be displayed when receiving a song display command of the difficulty level to be displayed, so that the client can display the song title of the song with the difficulty level to be displayed.

[0081] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0082] The present application also provides an electronic device, see Figure 5 , a structural diagram of an electronic device provided in an embodiment of the present application, such as Figure 5 As shown, a processor 1 and a memory 2 may be included.

[0083] Among them, the processor 1 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 1 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 1 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 1 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 1 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0084] The memory 2 may include one or more computer-readable storage media, which may be non-transitory. The memory 2 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In this embodiment, the memory 2 is at least used to store the following computer program 21, wherein, after the computer program is loaded and executed by the processor 1, it can implement the relevant steps in the method for determining the difficulty of singing a song performed by the server or the client side disclosed in any of the aforementioned embodiments. In addition, the resources stored in the memory 2 may also include an operating system 22 and data 23, etc., and the storage method may be temporary storage or permanent storage. Among them, the operating system 22 may include Windows, Unix, Linux, etc.

[0085] In some embodiments, the electronic device may further include a display screen 3 , an input / output interface 4 , a communication interface 5 , a sensor 6 , a power source 7 , and a communication bus 8 .

[0086] certainly, Figure 5 The structure of the electronic device shown does not constitute a limitation on the electronic device in the embodiments of the present application. In actual applications, the electronic device may include Figure 5 More or fewer components than shown, or combinations of certain components.

[0087] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided. When the program instructions are executed by a processor, the steps of the method for determining the difficulty of singing a song performed by the server or client of any of the above embodiments are implemented.

[0088] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of this application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of this application.

[0089] It should also be noted that, in this specification, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

Claims

1. A method for determining the difficulty of singing a song, characterized in that: include: Obtaining training songs and corresponding manual scoring information, and extracting audio features of the training songs; Obtaining the number of recordings and the number of releases of the training song, and calculating the ratio of the number of releases to the number of recordings as the release rate information of the training song; Weighting the release rate information and manual scoring information of the training songs to calculate the singing difficulty label corresponding to the training songs; Training a machine learning model using the audio features of the training songs and corresponding singing difficulty labels; Determine a target song for which singing difficulty needs to be evaluated, and extract audio features of the target song; The audio features of the target song are input into the trained machine learning model to obtain the singing difficulty score of the target song.

2. The method for determining the difficulty of singing a song according to claim 1, characterized in that: The method of training the machine learning model using the audio features of the training song and the corresponding singing difficulty label includes: The machine learning model is trained using any one or a combination of the range, pitch, high-pitched rate, low-pitched rate, long-pitched rate, breathing difficulty, and pronunciation speed of the training songs and the singing difficulty labels corresponding to the training songs.

3. The method for determining the difficulty of singing a song according to claim 1, characterized in that: The method of training the machine learning model using the audio features of the training song and the corresponding singing difficulty label includes: The various audio features of the training song and the corresponding singing difficulty labels are input into a linear regression model to perform linear fitting on the various audio features of the training song and the corresponding singing difficulty labels to obtain weight coefficients corresponding to the various audio features of the training song.

4. The method for determining the difficulty of singing a song according to claim 3, characterized in that: Inputting the audio features of the target song into the trained machine learning model to obtain the singing difficulty score of the target song, including: The audio features of the target song are weightedly calculated based on the weight coefficients corresponding to the various audio features to obtain a singing difficulty score for the target song.

5. The method for determining the difficulty of singing a song according to claim 1, characterized in that: The method of training the machine learning model using the audio features of the training song and the corresponding singing difficulty label includes: Inputting the audio features of the training song into a neural network model, and outputting a predicted singing difficulty score of the training song; A loss value is calculated based on the predicted singing difficulty score of the training song and the corresponding singing difficulty label, and the model parameters of the neural network model are adjusted according to the loss value until the loss value is less than a first preset value or the number of adjustments reaches a second preset value, thereby obtaining a trained neural network model.

6. The method for determining the difficulty of singing a song according to claim 1, characterized in that: The publication rate information and manual scoring information of the training songs are weighted to calculate the singing difficulty label corresponding to the training songs, including: Assigning a first weighting parameter and a second weighting parameter to the publication rate information and the manual scoring information respectively; Calculating a product of the publication rate information and the first weighted parameter as a first product, and calculating a product of the manual scoring information and the second weighted parameter as a second product; The sum of the first product and the second product is calculated as the singing difficulty label corresponding to the training song.

7. The method for determining the difficulty of singing a song according to claim 1, characterized in that: After inputting the audio of the target song into the trained machine learning model to obtain the singing difficulty score of the target song, the method further includes: Determine a target score range to which the singing difficulty score of the target song belongs; Determine the target difficulty level corresponding to the target scoring range according to the corresponding relationship between the scoring range and the difficulty level; The singing difficulty level of the target song is determined as the target difficulty level.

8. The method for determining the difficulty of singing a song according to claim 7, characterized in that: Also includes: When a song display command of a difficulty level to be displayed is received, the song title of the song with a difficulty level of the difficulty level to be displayed is determined so that the client displays the song title of the song with a difficulty level of the difficulty level to be displayed.

9. An electronic device, characterized in that: include: Memory for storing computer programs; A processor, used to implement the steps of the method for determining the difficulty of singing a song as described in any one of claims 1 to 8 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for determining the difficulty of singing a song as described in any one of claims 1 to 8.

Images